IRCloggy #git 2011-01-28

Logs Search ←Prev date Next date→ Channels Documentation

Provider of IRC logs since 2005.
WARNING: As Freenode became unjoinable and lost all warnings in topics, we cannot log channels on Freenode anymore.

2011-01-28

keyvan- left00:00
TeckniX left00:00
chrischambers left00:02
rcpeterson left00:03
tylerl joined00:03
tylerl so.. I accidentally committed my changes to "(no branch)" (detached HEAD, I think it's called?). so I switched to master (git checkout master) and tried to merge HEAD, but.. it doesn't have my changes?00:04
gaa!00:04
where'd my stuff go?00:04
Vasistha tylerl: ah, you can still get it back00:05
tylerl Vasistha: please tell!00:05
Vasistha Can you scroll back to where you did the commit to no branch?00:05
tylerl git scroll?00:05
Cro_Crx left00:05
Vasistha tylerl: no, I mean, in your terminal00:05
tylerl oh, ok.00:05
karihre doesn't the reflog list the commit?00:05
Vasistha that might be an easier way00:05
cbreak tylerl: HEAD is where ever you are NOW00:06
so when you did git checkout master, you changed HEAD00:06
what you should have done is git checkout -b newbranchname00:06
what you should now do is git checkout HEAD@{1}00:06
Arrowmaster reflog usage is probably going to be required for this one00:06
tylerl git checkout HEAD@e7ee2b3 -- like that?00:06
Vasistha tylerl: anyway, if you can scroll back and see what the commit number is, you can check it out (or merge it!) from that commit number00:07
Agos left00:07
cbreak tylerl: no00:08
read what I wrote00:08
what you wrote looks totally different!00:08
tylerl cbreak: ok. I though "1" was a placeholder. OK. I think i see now.00:08
keex left00:08
chrischambers joined00:09
eletuchy left00:09
tylerl so it looks like my changes are HEAD@{2}, so I can "git checkout master" and then "git merge HEAD@{2}" to pull my changes into master, right?00:09
Take left00:09
eletuchy joined00:10
panike joined00:10
cbreak no00:10
tylerl: read again what I wrote above00:10
also, it should be in HEAD@{1}00:10
git reflog will show you the heads00:11
the past heads you had00:11
andrewvos left00:11
priidu_ left00:11
priidu left00:11
andrewvos joined00:11
tylerl can I merge a specific commmit referencing its sha1? How do I merge the lost commit into my branch?00:12
bhosie left00:12
excess^ left00:12
panike left00:12
ayust git merge <sha1>00:12
if it's a single commit, it might be simpler to just cherry-pick it - git cherry-pick <sha1>00:13
tylerl /sigh cherry pick. I need to learn more about git.00:14
ibolmo left00:14
noahcampbell left00:14
tylerl ayust: hows that different from merging it?00:14
cbreak tylerl: totally different00:15
for one, it's not a merge00:15
chrischambers left00:15
tylerl cbreak: i guess that goes without saying00:15
otherwise they would have called it "merge" too00:16
ayust cherry-pick only grabs the commits you tell it00:16
rather than grabbing all commits in the chain that aren't already in the current branch00:16
cbreak you get a new commit that contains the state of the repo as if the changes of the commit you cherry pick were applies to the branch you cherry pick into00:16
tylerl ayust: so you mean only the differences in that particular commit from the commit before it?00:17
lastk sou muiteaheau00:17
cbreak while a merge creates a commit with your current head and the commit you want to merge as parent00:17
tylerl sounds like merge is more what I'm after then.00:17
ayust cherry-pick is essentially a fancy way of creating and then applying a patch, from a single commit00:17
cbreak that means that a merge will give you ALL commits of that branch in the second line of history00:17
while cherry picking only gives you a new commit with the changes of one existing commit00:17
what you want is not a merge00:17
it's cherry pick00:18
your problem was that you committed at the wrong point in history00:18
cherry pick can transfer the changes to a different point00:18
merging can not00:18
ayust if you have one history that's "A -> B -> C", and another that's "A -> D -> E", then cherry-pick E will give you "A -> B -> C -> E" while merge E will give you "A -> [B -> C, D -> E] -> F"00:18
Rockj left00:19
cbreak cherry picking E won't give you E00:19
it'll give you something similar00:19
ayust well00:19
tylerl cbreak: ... i WANT a cherry pick? I need to pull about 4 commits, then.00:19
ayust the effects of E00:19
SeanLazer joined00:19
ayust tylerl - so you committed 4 times on your detached head?00:19
kipras left00:19
cbreak tylerl: every single one you committed headlessly00:19
pick them from the oldest to the newest00:19
ayust in newer versions of git you can actually cherry-pick a range00:20
Rockj joined00:20
Heartbroken left00:20
tylerl what I would like is to take all the changes made in the detached head from the time it was detached and pull them into the main branch.00:20
ayust git cherry-pick origdetachedhead..lastcommitondetachedhead00:20
Ryback_ left00:21
SirFunk_ left00:22
SimonP86 left00:22
tylerl ayust: as the origin, do I use the last commit that was part of the branch, or the first that was not?00:22
ISF left00:23
SeanLazer left00:23
ayust so in git, A..B means "everything after but not including A, up to and including B"00:24
so A should be before you made any commits, and B should be the last commit you made00:24
tylerl ayust: ty00:24
MattDiPasquale left00:25
Mannequin1 left00:26
shruggar left00:27
tylerl ugh. fail. conflicts. wtf.00:27
kukks left00:28
tizzo left00:28
tvw left00:29
__name__ left00:32
Mab879[Away] left00:33
panike joined00:33
Simkin left00:33
eean left00:33
eijk_ left00:33
kpreid joined00:34
yaottiyaotti_aw00:34
SimonP86 joined00:39
dec left00:41
dec joined00:41
TheBorg[ReGen] joined00:41
eletuchy left00:41
kukks joined00:42
jhoepken joined00:43
ki__ I want to check which files were updated on a specific date00:43
do I use git log for this?00:43
pdelgallego__ left00:43
scelis left00:43
miloshadzic left00:43
ki__ I am reading git help log and there are sooo many options ... what is most practical to get an idea of which files a developer updated on a specific date?00:43
ayust git whatchanged --since <date> --before <date>00:44
webPragmatist left00:45
ayust dates being in YYYY-MM-DD format00:45
ki__ whoa nice syntax. let me try that00:45
jhoepken left00:46
ki__ git whatchanged --since 2011-01-25 --before 2011-01-26 -- like that?00:46
pdelgallego_ left00:46
ayust both --before and --since are non-inclusive00:46
eean joined00:46
pdelgallego left00:46
SeanLazer joined00:46
ayust oh, no, sorry, im wrong00:46
--since is apparently inclusive00:47
so yes, that should work if you want the 25th's changed files00:47
keyvan- joined00:48
Yuuhi left00:49
ki__ it doesn't like that syntax00:49
ayust what error does it give you?00:49
t0rc joined00:49
spaceonline left00:51
SeanLazer left00:53
ayust ki__: ^00:53
Noldorin_ joined00:53
theoros left00:53
theoros joined00:54
eean left00:54
eean joined00:54
bkosborne left00:55
xiangfu joined00:55
Noldorin left00:56
dfr|mac left00:56
miloshadzic joined00:57
Slix` joined00:58
bcardarella joined00:58
miloshadzic left00:58
Vasistha left00:58
gretch left00:58
c00kiemon5ter joined00:58
miloshadzic joined00:59
_psychic_ joined00:59
sebersole left00:59
TheBorg[ReGen] left01:00
_psychic_ left01:01
dpino left01:02
gozoner left01:02
gozoner joined01:03
zorzarzorzar_away01:04
alnewkirk joined01:04
ki__ ayust: fatal: ambiguous argument '2011-01-25': unknown revision or path not in the working tree.01:05
ayust try --since=2011-01-15 then01:05
(ditto for --before)01:05
c00kiemon5ter left01:05
TheBorg[ReGen] joined01:06
deek0146 left01:07
bitslip_ joined01:07
kenichi left01:07
tjholowaychuk left01:08
bitslip left01:08
consolers joined01:08
mase_wk_ joined01:08
mase_wk left01:09
sako hey guys so a developer accidentally git rm a bunch of stuff :/01:09
my local repo is fine as i havent pulled in changes yet01:09
can i force push my stuff through some how?01:09
ayust git push --force01:09
if you want to overwrite the remote01:09
sako ayust: thanks again :)01:10
ayust (git push --force origin master)01:10
would overwrite 'master' on origin with your local 'master'01:10
necromancer joined01:10
ayust note that if anyone -has- pulled the changes they'll need to reset back to before they pulled01:11
sako ayust: one developer has pulled01:11
SimonP86 left01:11
sako so they would need to git reset --hard?01:11
aalex_laptop joined01:12
artefon left01:13
sitaram any ops here? someone called "consolers" pm-d me asking if he can get voice because he can't access the captcha (I have no clue what all that is about!)01:13
Arrowmaster i think it has something to do with how people keep getting voice in here and i cant figure out why01:15
probably some webchat client01:15
bremner Arrowmaster: see /topic01:15
Arrowmaster ah01:16
ChanServ set mode: +v01:16
consolers thanks!01:16
bremner welcome01:17
ChanServ set mode: +v01:17
_Unode left01:17
Arrowmaster it works!01:17
malfy left01:17
consolers is there a git config variable to overcome this: fatal: Refusing to fetch into current branch refs/heads/vendor of non-bare repository01:17
malfy joined01:18
consolers I set receive.denycurrentbranch set to warn01:18
sako ayust: what would the dev that pushed all the changes have to do?01:19
Eridius consolers: that's used for pushes, not fetches01:19
sako reset before the merge?01:19
consolers yeah, i want the same symmetric behaviour.01:20
now i have to mv .git , checkout a dummy branch, rest and then fetch,01:20
Eridius consolers: there's no config variable for this, but you could fake it by claiming you're in a bare repo... git -c core.bare=true fetch ...01:21
Unode|Work joined01:21
ayust sako - also reset, yes01:21
Anubisss left01:21
aah left01:22
consolers ok i'll try that the next time01:22
Unode|Work after doing a git filter-branch shouldn't a git prune ; git repack -ad ; git gc --aggressive remove all unreferenced objects?01:22
I can't seem to make git lose the 190MB of excess01:23
ayust consolers: if you're just trying to grab a certain remote revision and make it your current one, fetch to a different ref and reset to that ref01:23
Eridius try adding --expire=now to your prune01:23
Unode|Work If I clone the repository however, only 6 MB are cloned01:23
Eridius also anything referenced in the reflogs will still be rooted01:23
so you may need to blow those away too01:23
alvaro_o left01:23
Unode|Work Eridius, git reflog expire changed nothing01:23
SethRobertson Unode|Work: RTFM at the end of man git-filter-branch01:23
jast Unode|Work: the 'git-filter-branch' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-filter-branch.html [automatic message]01:23
SethRobertson Unode|Work: specifically, the bits about changing the expire time01:24
tylerl How do I revert the HEAD of a branch a prior commit (removing a commit)?01:24
SethRobertson tylerl: Have you pushed?01:24
tylerl SethRobertson: no01:24
ayust tylerl: git reset --hard <older sha>01:24
Eridius note, this will blow away uncommitted changes01:25
git reset --keep HEAD^ may be a bit better01:25
tylerl that's the idea01:25
ayust or "git reset --hard HEAD~X" (where X is a number of steps backwards you want to go)01:25
(so "HEAD~2" would be "two steps backwards")01:25
Eridius actually I suppose --merge may be slightly nicer01:25
hyperair left01:25
Unode|Work SethRobertson, if you mean git reflog expire --expire=now --all , I've done it several times with no success, same thing for git gc in combination with git prune01:26
Eridius ehh, I'm still not 100% sure of the difference between --keep and --merge. That manpage really needs some examples to illustrate it01:26
Unode|Work SethRobertson, including --expire-unreachable01:27
SethRobertson Unode|Work: Try doing a clone and see if it goes away in the clone. That helps debug the problem01:27
sako thanks ayust01:27
Unode|Work SethRobertson, it does... that's what puzzles me01:27
Eridius SethRobertson: it sure would be nice if `git fsck` had an option to identify how a particular object is rooted01:27
ayust Eridius: --keep does not touch files at all if they have local changes01:27
--merge tries to merge local changes (relative to the current head) into the new head01:27
Eridius ayust: from reading --merge it kind of sounds like staged changes will get blown away01:28
SethRobertson Unode|Work: When you did a `git prune` and `git gc` did you likewise use the date options?01:28
Unode|Work git gc --prune=now yes01:28
ayust Eridius: --merge essentially applies "git diff HEAD" from before the reset to the new HEAD01:28
(but slightly more intelligently)01:28
SethRobertson Unode|Work: How about `gc repack -Ad `01:29
Eridius ayust: glad to hear it. Try doing a strict reading of the docs, though. It resets the index, and it talks about what happens if there's a difference between the index and the working tree, but nowhere does it say what happens if the index differs from HEAD (and the "resets the index" bit therefore implies those staged changes are lost)01:29
Unode|Work SethRobertson, no change01:29
SethRobertson, the packed file still has 190MB in excess01:29
ayust Eridius: if you git added something, and then reverted the file to the old HEAD (but didn't add the revert), then the index will be blown away01:29
but in most cases, changes that are staged are also in your working tree01:30
consolers left01:30
Eridius ok01:30
ayust in the corner case where a change in the index is not also in your working tree, yes, it would be lost01:30
smolyn left01:30
Eridius what if you git add something, and then have no unstaged changes? In that case the working tree doesn't differ from the index, so it sounds like those changes would be lost01:30
ayust no01:30
Eridius I'm sure that's not the case, but again, that's what the docuemntation seems to be saying01:30
ayust because the index != the current commit01:30
Unode|Work SethRobertson, any additional ideas?01:31
Eridius ayust: read the documentation on --merge and tell me how I'm reading it wrong01:31
mikepack joined01:32
sako left01:32
ayust actually, you might be right; it may blow away index changes in favor of a file that changed01:32
avar You can still recover things that have at one point been in the index, even if you nuke those things01:32
i.e. git add && git reset --hard01:32
Eridius ayust: sounds like something that should be tested01:32
offby1 if anyone has found a way to use "reviewboard"'s "post-review" script with git, I'd be interesting in talking01:32
s/interesting/interested/01:32
khmarbaise left01:33
SethRobertson Unode|Work: Do you have any weird refs? alternative? original?01:33
ayust I mostly just tend to not assume that git reset will keep anything uncommitted around when using anything aside from --mixed or --soft :)01:33
Unode|Work well now I seem to have broken HEAD01:33
fatal: bad object HEAD01:33
after doing git reflog01:33
SethRobertson fsck01:33
Eridius ayust: yep, if the file changed between HEAD and <commit>, the staged changes will be thrown out01:33
Unode|Work notice: HEAD points to an unborn branch (master)01:34
yay for breaking it01:34
ayust did you somehow delete your refs/heads/master ?01:34
offby1: we use post-review01:35
w/git01:35
SethRobertson Unode|Work: git for-each-ref refs says?01:35
Unode|Work I think git reflog rewrote something that it shouldn't01:35
ayust git reflog doesn't rewrite anything01:35
Unode|Work git reflog expire --expire=now --all --rewrite --updateref actually01:35
ayust er01:36
so you cleared out your reflog01:36
SethRobertson Remember that clone we talked about. Do you still have it?01:36
ayust and then set master to the top value of the (now empty) reflog01:36
yep, that could get rid of master :P01:36
Unode|Work :P01:36
SethRobertson, I can make another one from the existing backup01:37
so the backup is actually a copy of the full project01:37
ayust scroll up in your terminal, see what the top item on master used to be, and git branch master <that sha> ?01:37
Unode|Work cloning it only gets rid of the full thing if done via the network01:37
ayust, it reports 00000000000000...0000000001:37
tty2 left01:38
ayust i meant in your terminal history01:38
from before you did the reflog expire bit :P01:38
Unode|Work oh01:38
ayust (alternatively, just do git rev-parse master, inside your backup repo)01:38
tchan left01:38
Unode|Work ayust, that would be to fix the broken master right?01:38
ayust yes01:38
although01:38
if you rewrote a bunch of stuff via filter-branch or whatnot01:39
the master in your backup repo might not be what you want to have as your master tip anymore01:39
eletuchy joined01:39
ayust in which case, terminal history go01:39
khmarbaise joined01:39
Unode|Work well now the master issue is no longer one, the backup is the result of the filter-branch stuff prior to breaking master01:39
ayust ah, k01:39
deek0146 joined01:40
tchan joined01:40
shyam_iyer joined01:40
artefon joined01:40
Unode|Work I'm just trying to understand why doesn't git lose the removed objects01:40
abetaha left01:40
ayust so git prune --expire=now didn't work?01:40
Unode|Work since if I clone the current repo via the net, I get a repo with size 6MB when the current one is 194MB01:41
ayust, nope01:41
ayust try doing the reflog expire but without the --updateref01:41
Unode|Work is it normal that git prune gives no output and finishes in miliseconds?01:41
ayust if it doesnt think there's anything to prune, yes01:41
Unode|Work ayust, git reflog expire --expire=now --all --rewrite ?01:41
Heimidal joined01:41
ayust you don't need the --rewrite01:42
since you're expiring everything01:42
Unode|Work well even with the --rewrite nothing changes01:42
I mean, the size remains01:42
ayust try git gc --prune=now01:43
nfiniteset left01:43
ki__ left01:43
Unode|Work still the same01:43
ayust if that fails, you can try adding --aggressive01:43
Unode|Work scratches head01:43
SethRobertson Unode|Work: `git for-each-ref refs` what does that say01:44
Unode|Work running gc --prune=now --aggressive right now, taking a while01:44
ayust one possibility is that you still have a ref that points at your old object chain01:45
Unode|Work seems like it's still compressing the 190MB objects01:45
ayust and thus it's not getting cleaned up01:45
because it's still referenced01:45
Unode|Work ayust, and would that be picked up by git clone?01:45
it would right?01:45
Cap_J_L_Picard left01:45
ayust if it's a head, yes01:45
SethRobertson not for unusual refs01:45
Like original alternatives etc01:45
Cap_J_L_Picard joined01:45
Unode|Work well this is a single branch repository01:45
SeanLazer joined01:46
SethRobertson that doesbn01:46
that doesn01:46
t matter, and is why I am asking for the for-each-ref output01:46
Unode|Work and I don't think history has ever been touched apart from me today doing the --tree-filter01:46
SethRobertson, sorry, still running the git gc01:46
ayust plx run what SethRobertson is mentioning01:46
Unode|Work one sec01:46
it's taking quite a lot of time01:47
shyam_iyer left01:47
ayust yes, --aggressive does that >.>01:47
Unode|Work :\01:47
ayust since for-each-ref is readonly01:47
you should be able to run it in another terminal01:48
Unode|Work ok01:48
Slix` left01:48
dr_win left01:48
Unode|Work 259ef516f19ae635afefd979e26b739b31d483d8 commit refs/heads/master01:48
3a7440f353f88b5c7458b4b13e17e608cdf94e80 commit refs/remotes/origin/HEAD01:48
3a7440f353f88b5c7458b4b13e17e608cdf94e80 commit refs/remotes/origin/master01:48
jrdavid_ left01:48
ayust yeah01:48
Unode|Work duh01:48
ayust your remotes01:48
are pointing at the old one01:48
Unode|Work yeah I get it xD01:48
Slix` joined01:48
Unode|Work sorry about that01:48
well now things make sense01:48
so in this case, the best way would be to remove remotes since they will need to be rebuilt anyway01:49
ayust sure, that'll work01:49
linusoleander left01:50
linusoleander joined01:50
Godslastering left01:51
Unode|Work ayust, yup that did it01:51
thanks ayust and SethRobertson for the help01:51
gretch joined01:55
gretch left01:55
SeanLazer left01:55
miloshadzic left01:55
deek0146 left01:58
Unode|Work left02:00
zombor left02:04
jonmasters joined02:04
Yuffster joined02:05
shyam_iyer joined02:06
peshalto left02:07
ChillaxChillaway02:08
t0rc left02:09
t0rc joined02:09
project2501a joined02:10
p3rror joined02:11
marienz left02:13
tty1 joined02:14
marienz joined02:16
peshalto joined02:16
Pewpewarrows joined02:17
airborn left02:17
spearce left02:18
smolyn joined02:19
shyam_iyer left02:19
dreiss joined02:21
dfr|mac joined02:21
silanus_ joined02:22
guns joined02:22
peshalto left02:23
timj_ left02:23
silanus left02:24
artefon left02:25
eletuchy left02:25
eletuchy joined02:27
SeanLazer joined02:30
timj joined02:30
Vampire0 left02:31
Evious left02:36
ISF joined02:36
bkosborne joined02:37
mw left02:37
boombatower joined02:37
necromancernecromancer`bzzy02:38
shyam_iyer joined02:39
panike left02:40
mw joined02:41
mw left02:41
mw joined02:41
p3rror left02:43
linusoleander left02:44
shyam_iyer left02:44
shyam_iyer joined02:45
pheaver left02:45
SeanLazer left02:46
hyperair joined02:47
necromancer`bzzynecromancer02:47
rayners joined02:48
gretch joined02:49
Vampire0 joined02:49
Remixman joined02:50
ROBOTARMY left02:50
deek0146 joined02:55
deek0146 Hey02:56
My friend has given me space on his server, how do I upload a git repository to it?02:56
bmahe left02:57
tek0 left02:57
ROBOTARMY joined02:58
guns left02:58
ROBOTARMY left02:58
ROBOTARMY joined02:58
Davey_ joined02:58
Davey_ Hola02:58
I need help doing bad thing to my repo02:59
bremner deek0146: the best way is to have git installed on the server02:59
Noldorin_ left02:59
deek0146 I think it might be02:59
He has a git repository on it already02:59
Which I can push and pull from, but I want to create a new one02:59
Davey_ I want to set commit 0735fa2286177d88074fe5fb83dda243926cacd9 on my release branch, as having been merged from 0dcc93f0836efb5e0153d918c3b56c36ec50ac05 on my master branch... is this possible?03:00
fr0sty-away deek0146: on the server: git init --bare in whatever directory you want (~/repos/project.git)03:00
gozoner left03:01
deek0146 I don't have ssh access03:01
fr0sty-away what access do you have>03:02
deek0146 Its ftp03:02
fr0sty-away ?03:02
deek0146 I'm not sure03:02
I'll tell you the command i did to clone his other repo03:02
madewokherd left03:02
fr0sty-away don't think git works over ftp03:02
yeah, what command did you use?03:02
deek0146 [email@hidden.address]03:02
bremner that looks like a gitolite setup03:03
fr0sty-away deek0146: he probably has gitolite then03:03
madewokherd joined03:03
bremner ssh [email@hidden.address] will tell you03:03
deek0146 Is that the same thing that knows my ssh key? Will I be able to create a new repo03:03
bremner it depends03:04
on the configuration03:04
sitaram is impressed that people now automatically assume gitolite is involved ;-)03:04
deek0146 Emm03:04
that ssh command gave no warning03:05
Authenticity can't be established, are you sure you want to continue blah blah03:05
Permission denied: public key03:05
nb_ joined03:05
deek0146 sitaram: Whys that impressive? Is gitolite new? Or just really cool03:06
sitaram deek0146: you couldn't have got "authenticity cant be established" if you already used that clone command...03:07
project2501a um, if i wanted to post a reward, say USD$5k for someone to fork launchpad and incorporate native git and hg support, instead of bridging which has now, where should I do it?03:07
bremner sitaram: btw, you might find http://people.debian.org/~bremner/gitolite-monkeysphere/ interesting. I'll poke you again when the project is up and running properly.03:07
project2501a and how should i put the money in escrow?03:07
sitaram bremner: I dont suppose you already know about http://sitaramc.github.com/gitolite/doc/monkeysphere.html do you?03:07
linusoleander joined03:08
bremner sitaram: yes I started there, and talked to Clint a fair bit.03:08
dmlloyd_ joined03:08
deek0146 sitaram: I cloned it on a windows install, not this machine03:08
sitaram bremner: oh ok03:08
deek0146: then go there and do the ssh03:08
deek0146 hokay 2 secs03:08
sitaram bremner: you can create by pushing also, you dont have to do it by cloning. (I suppose you wrote it that way so people will have origin and upstream branches set correctly from the start)03:09
deek0146 Err yea it asked for me my password (stupid, of course, his server doesn't know about this computer03:09
dmlloyd left03:09
deek0146 Then it said:03:09
bremner sitaram: I wrote it that way out if ignorance ;)03:09
deek0146 ERROR:gitosis.serve.main:Need SSH_ORIGINAL_COMMAND in environment03:10
Heimidal left03:10
deek0146 So yea, gitosis :) What does that error mean?03:10
bremner ah, gitosis probably means no creation iirc.03:10
sitaram bremner: also, can I ask you to change the URLs? I finally got off my duff and created a whole bunch of much faster loading URLs; they all start with https://sitaramc.github.com/gitolite instead of https://github.com/sitaramc/gitolite/blob/pu03:11
deek0146 Aww :( But he said I could store stuff on the server using ftp, is it not possible to set up git that way?03:11
sitaram bremner: they are statically created; github's dynamic rendering was too slow for a lot of people, plus all the distraction of the top half of the screen being gone03:11
bremner sitaram: no problem. Just search and replace the prefix?03:11
sitaram yup03:11
bcardarella left03:11
sitaram and it can be http:// I guess, not https. Mistake there03:12
stringoO joined03:12
sitaram has to head to work; be back online in an hour or so03:12
Remixman left03:13
bremner sitaram: bye for now.03:13
bmahe joined03:13
fr0sty-away deek0146: git doesn't work over ftp03:14
deek0146 sadness03:14
pheaver joined03:15
fr0sty-away deek0146: have your friend create a gitosis repo for you.03:16
deek0146 Ok cool03:16
Thanks for the help guys :D03:16
dmlloyd_dmlloyd03:17
bremner sitaram: if your still here, /gitolite/doc seems not to work?03:18
sitaram: nvm. must be my error03:19
urbanmonk left03:20
nadavoid joined03:20
linusoleander left03:20
bremner oic, s/mkd/html/03:20
urbanmonk joined03:21
linusoleander joined03:21
xiong left03:21
intripoon_ joined03:22
project2501a um, if i wanted to post a reward, say USD$5k for someone to fork launchpad and incorporate native git and hg support, instead of bridging which has now, where should I do it? and how should i put the money in escrow?03:22
keyvan- left03:22
yaotti_awyaotti03:22
bremner project2501a: why don't you write to the mailing list?03:23
stringoO left03:23
project2501a bremner: good idea.03:24
intripoon left03:25
SeanLazer joined03:28
linusoleander left03:29
linusoleander joined03:29
leo2007 left03:30
linusoleander left03:30
Amorphous left03:30
hipstertrash left03:31
zombor joined03:34
Elfe joined03:36
deek0146 left03:37
justin-george left03:38
sako joined03:38
Elfe_ left03:39
crashanddie left03:39
aalex_laptop left03:43
sitaram bremner: that whole tree is auto-generated; I guess I have to put an index.html there also, will do. Meanwhile go up one level and you'll see an index03:44
urbanmonk left03:44
luis_ joined03:45
urbanmonk joined03:45
Amorphous joined03:45
sitaram bremner: oh yeah I forgot that part -- sorry! (mkd -> html)03:46
mw left03:48
envi joined03:49
coderdad joined03:49
timj_ joined03:52
bkosborne left03:53
coldpizza72i joined03:56
rns left03:56
timj left03:56
coldpizza72i how do i pull once you create a branch03:56
git pull master?03:58
zombor left03:58
flaguy48 left03:58
mdel left04:00
beekhof left04:01
mdel joined04:01
yaottiyaotti_aw04:02
BiggFREE joined04:03
coldpizza72i left04:04
beekhof joined04:06
beekhof left04:06
beekhof joined04:06
dermoth|home joined04:06
guns joined04:07
Remixman joined04:07
project2501a left04:07
noahcampbell joined04:08
QAALTXN joined04:09
OOPMan joined04:10
ausdroid joined04:12
justin-george joined04:15
friskd left04:18
dreiss left04:18
zoo-zed joined04:20
zoo-zed left04:21
Schmallon left04:23
bug__ joined04:25
surfdue joined04:26
surfdue Hello!04:27
StuckMojo Please, commit your changes or stash them before you can merge.04:27
what's "stash"?04:27
surfdue I currently cloned a remote repo which default's branch is master, how do i switch the default branch to dev04:27
StuckMojo or is there a way i can force an overwrite?04:27
bug_ left04:27
linusoleander joined04:28
StuckMojo surfdue: maybe with git-config04:29
surfdue: what you do mean by "default" branch?04:29
to switch to a different branch, it's just: git checkout branch_name04:29
here joined04:29
lastk left04:29
StuckMojo once you switch, that sticks04:29
mdel left04:29
csanz left04:30
harshu joined04:30
StuckMojo on my "stash" question, nevermind. found the doc04:31
justin-george left04:31
surfdue StuckMojo: im trying to do a git pull but it says you asked me to pull without telling me a branch04:32
i want this one to only pull the origin/dev branch04:32
StuckMojo ah04:33
surfdue well this clone atleast04:33
any idea?04:33
StuckMojo well, you can just: git checkout dev; git pull origin dev04:33
surfdue think i got it04:35
:P04:35
silly me04:35
bug__ left04:36
kukks left04:36
Heimidal joined04:39
vandemar joined04:40
nothingHappens left04:40
ceej left04:40
jmorris joined04:40
Nitecon joined04:41
Nitecon Hey folks04:41
How does git handle really long filenames? I've used mercurial to version control my homedir04:42
however in the last few weeks some of the stuff that I have checked in has really long windows filenames04:42
which cause mercurial to die. As such I'm looking for an unbiased answer as to whether the same problem exists in git04:43
StuckMojo Nitecon: i'm very new to git, but i would think it doesn't04:43
Nitecon Yeah thats the problem, I need to make sure before I switch04:44
I don't mind the switch04:44
StuckMojo you could just make a mini git repo and try it04:44
how long a filename are you talking here?04:44
Nitecon but switching a 4 terrabyte homedir over takes a while :P04:44
StuckMojo just init a repo, put one of these long ass files in it, see if it works04:44
Nitecon over 256 chars04:44
including multi deep paths with directories over 256 chars too04:45
hmm yeah may try that instead04:45
I've got about 12 more hours of backup to do before I can start playing haha04:45
vandemar this is really stupid I'm sure but how do I get back to tracking upstream: http://dpaste.com/361305/04:46
wereHamster vandemar: git reset --hard origin/develop-olympus ?04:48
dfr|mac left04:48
shentino left04:48
StuckMojo Nitecon: the filename itself is long, or the pathname?04:49
Nitecon in some instances both04:49
hipstertrash joined04:49
vandemar wereHamster: thanks, wow I'm stupid04:49
StuckMojo i can't find anything in google about it, so i suspect it's basically the same limits as your FS04:49
Nitecon yeah thats what I'm hoping :P04:50
cleaning out my extra nas so I can set up a gitweb site and try out git :)04:50
StuckMojo can you paste a filename here?04:50
i'll try it for you04:51
here heya, having trouble with basic .gitignore exceptions using ! : https://gist.github.com/79984804:51
StuckMojo well, a paste site would be better actually04:51
Nitecon one sec i'll find one in a bit04:51
StuckMojo since it's super long04:51
wereHamster here: git doesn't descend into directories which are ignored, so the !public_html/.htaccess does not even take effect04:52
roop joined04:53
here were, so if i want to ignore most files under public_html, with some exceptions?04:53
StuckMojo Nitecon: ok, on ext4 i can't make a file longer than 256 chars i think04:53
wereHamster here: if you only want to track a few files, just use /* and then explicitely add the files you want to track04:53
StuckMojo i've got one that's 226 not04:53
now04:53
Nitecon ok cool04:53
here were: trying.04:53
Nitecon lol i'm trying to find a file now...04:54
wereHamster here: public_html/* and !public_html/<the files>04:54
Nitecon they always come up with failure but hard to pinpoint them04:54
StuckMojo ok, 25404:54
wereHamster here: also, are you trying to store $HOME in a git repo?04:54
StuckMojo git commit worked fine04:54
here were: hrm.. there are also further rare subdirectories ( !public_html/wp-content/themes/ )04:55
SeanLazer left04:55
StuckMojo Nitecon: and cloning it was fine04:55
ROBOTARMY left04:55
here were: no, but was thinking about locating the base of the repo in $HOME and picking which files to track below that.04:55
vandemar left04:55
hipstertrash_ joined04:56
Nitecon StuckMojo:great thanks for the info04:56
wereHamster here: bad idea, for the reason you've seen. It's better to create the repo where you need it (inside public_html)04:56
hipstertrash left04:56
hipstertrash_hipstertrash04:56
StuckMojo Nitecon: ok works with this too:04:56
./foo/bar/baz/zip/blah/foobie/loadie/dah:04:56
suuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuuperlong.file.txt04:56
Nitecon sweet04:56
ur my hero sire04:56
that should solve my problem then04:57
by bye HG04:57
StuckMojo so it's higher than or as high as ext4 limits ;)04:57
Nitecon yeah04:57
here were: i feel like i'd run into the same problem. i'm only looking for files we change like .htaccess and a handful of deeper directories (themes/plugins)04:57
bronson_ left04:58
sivy left04:58
wereHamster here: the only sensible solution I see is to ignore everything and then explicitely add what you need (git add -f ...04:58
yaotti_awyaotti04:59
here ok i'll look at that more, thx.04:59
Davey_ left04:59
rayners left04:59
wereHamster here: however, I still do not recommend putting your git repo inside $HOME04:59
hagabaka left05:01
nadavoid left05:01
mrball left05:02
hagabaka joined05:02
sivy joined05:03
webPragmatist joined05:03
webPragmatist left05:03
webPragmatist joined05:03
lu_zero left05:04
Eridius left05:04
sivy left05:04
here were: public_html seems equally bad. ;/ ?05:05
shentino joined05:05
shentino left05:05
shentino joined05:05
bronson_ joined05:07
ZzTribble joined05:07
harshu left05:08
harshu joined05:08
napster_123 joined05:09
roop left05:09
Slix` left05:10
kylehayes joined05:12
StuckMojo how do i look at a previous revision of a file?05:12
yaottiyaotti_aw05:12
_jham_ joined05:13
StuckMojo actually, git log -p will do05:13
_jham__ joined05:13
linusoleander left05:13
_jham__ left05:14
lu_zero joined05:14
ausdroid left05:16
vitor-br left05:17
theoros_ joined05:18
_jham_ left05:18
smolyn_ joined05:19
theoros left05:20
smolyn left05:21
doener left05:23
jbd left05:23
ZzTribble left05:23
jbd joined05:25
aah joined05:26
avinashhm joined05:27
Samual joined05:33
Samual_ left05:35
doener joined05:35
duellsy joined05:37
psankar joined05:38
duellsy hey gang... bit of a noob question... how do i use git to keep the framework up to date, while I keep my custom files in the project in a separate git?05:38
avinashhm left05:39
avinashhm joined05:39
jmorris left05:40
StuckMojo duellsy: two separate repos?05:40
that's a good question though05:41
duellsy yeah05:41
boombatower left05:41
StuckMojo i dunno for sure, i'm new too05:41
but i have the same issue05:41
i need boost for my project05:41
so i included it right in my source tree05:41
but it would make more sense for it to be a separate repo, since it almost never changes05:41
duellsy i want to be able to pull the framework when the team makes updates to it, but also be able to have a particular folder (the app folder) be controlled by a separate git only i have access to05:42
StuckMojo i guess the question is: what happens when to working directories overlay eachother05:42
duellsy guess i could ignore the app folder, and make it a git of it's own... but the app folder is already a part of the framework, so not sure that would work05:43
shyam_iyer left05:43
j416 joined05:44
StuckMojo oh05:44
webPragmatist anyone around that uses redmine?05:44
StuckMojo so you want only you to have write access to the app folder, but everyone to have read?05:44
j416: howdy05:44
j416 morning05:44
StuckMojo heh, 11:44pm for me05:44
j416 :)05:46
6.45 here05:46
webPragmatist left05:48
noahcampbell left05:50
here i'm confused moving between two local repos, how do i point git at a different local repo ?05:50
j416 here: are you talking about remotes?05:50
in that case, just add another remote05:50
and remove the first one if you wish.05:50
jrm222 joined05:51
here hmm, i don't think so. they are two different directory trees on the local machine, not even connected to remotes at this point.05:51
j416 ok05:51
so what do you want to do?05:51
HG` joined05:52
here say i'm working with one repo and use 'git add ...', etc. then i want to go and work with the other one..,05:53
j416 yes05:53
here seems git still thinks i'm working with the first, feel like i need to point git at the 2nd .05:53
j416 git is local05:53
meaning05:53
envi left05:53
bmahe left05:53
j416 you have one directory ~/myfirstrepo05:54
and another one ~/mysecondrepo05:54
jrm222 Do I have a voice?05:54
j416 they both contain a ".git" dir05:54
jrm222 yup05:54
j416 they are two separate repositories.05:54
here j416 right.05:54
j416 right.05:54
fadein left05:54
here how do i know which i'm affecting with the 'git add ...' command?05:54
j416 so if you are in ~/myfirstrepo, you will be working against the git repo in that directory.05:55
if you are in ~/mysecondrepo, you will be working against the git repo in that directory.05:55
here what if i'm in neither directory?05:55
j416 then git add will not work.05:55
here hmm. seems to ;p, hehe.05:55
ThunderChicken joined05:55
j416 in that case you have a git repo in that directory.05:55
ThunderChicken #gnu-smalltalk05:55
oops.05:56
StuckMojo j416: you'd be envious of my setup at the moment05:56
j416 StuckMojo: I doubt it.05:56
envi joined05:57
bronson_ left05:57
Remixman left05:57
amondracorp joined05:58
ISF left05:58
StuckMojo j416: http://mojo.forgot.his.name/screenshot.png05:59
wagle left05:59
StuckMojo j416: note that it's an ubuntu desktop, with visual studio in a window from an overlayed win7 vbox instance06:00
j416 ok?06:00
wagle joined06:00
StuckMojo well you said you might have to do some windows dev in the near future06:00
j416 yeah06:00
StuckMojo i'm only suffering juuuust enough windows to get what i want ;)06:01
j416 heh06:01
freimatz joined06:01
StuckMojo its virtualbox "seamless" mode06:01
j416 at least I'm not going to have to use visual studio06:01
sure, I use that too06:01
StuckMojo actually, visual studio is a very very nice IDE06:01
j416 I'll be running proper windows at work though06:01
StuckMojo at least for C++06:02
the debugger is unparalleled06:02
j416 nice.06:02
echrsin joined06:03
StuckMojo atm, i'm actually just editing my files in vim on unbuntu, committing to git, push, then pull on windows and bulid and debug with vc06:04
j416 haha what a hassle06:04
does it work well?06:04
chrisshattuck left06:04
StuckMojo not much hassle really06:04
yeah it does06:04
j416 (by the way, you said visual studio is a good IDE, how come you're doing your editing in vim? :P)06:05
StuckMojo 1) i'm just used to it now06:05
2) this way i don't have to try to git push from windows06:05
j416 are you scared of (2)?06:05
StuckMojo mmmm....a little ;)06:05
j416 why? is it any different from git push on unix?06:05
friskd joined06:06
dreiss joined06:06
avinashhm left06:06
echrsin left06:07
echrsin joined06:08
StuckMojo j416: i suppose not06:08
oh, i'm connecting via a cifs mounted drive and a symlink06:08
rather than ssh06:08
so i don't really trust that06:09
oh hey, with visual sourcesafe, your working tree was readonly06:09
j416 cifs.. never heard of it. ok.06:09
StuckMojo so you had to "checkout" a file to edit it06:09
j416 yeah06:09
I remember that06:09
StuckMojo cifs = samba = windows file sharing06:09
j416 ah ok06:09
StuckMojo cifs = common internet file system or some such06:10
j416 ok06:10
StuckMojo it's the protocol smb uses or something.06:10
anyway...06:10
so i kinda liked that checkout bit, because it was harder to lose changes accidentally06:10
i have p4 configured that way at work06:10
it will do either way though06:10
p4 opened shows all the files in your client you've checked out for edit, so it's hard to lose stuff06:11
kuhrt joined06:11
StuckMojo or change things you didn't mean to06:11
can git work like that?06:11
j416 well git is local06:11
StuckMojo "In computer networking, Server Message Block (SMB), also known as Common Internet File System (CIFS)"06:12
devanjan joined06:12
j416 but you can rm -rf * in your working dir and get it all back06:12
StuckMojo so, just another name for it06:12
j416 and you can always see what changed06:12
so there's really no problem, is there? :)06:12
StuckMojo mmmm06:12
well, if you had more than one working tree, you could inadvertantly work in the wrong one06:13
j416 ty (smb == cifs, didn't know)06:13
StuckMojo i.e. have changes in both06:13
avinashhm joined06:13
j416 StuckMojo: sure, but then you could generate a patch and move changes06:13
StuckMojo yeah in the linux kernel config, they call it the cifs module06:13
j416 k06:13
StuckMojo dunno why. i think that started when MS finally published the protocol06:13
luis_ left06:13
StuckMojo j416: yeah but you'd have to notice that you did it06:14
offby1 left06:14
j416 StuckMojo: that's true for sourcesafe too06:14
StuckMojo if you didn't realize that two terminals were in two diff working trees06:14
j416 you can check out a file from the wrong repo06:14
work with it06:14
check it in06:14
StuckMojo whereas if you had to check out, one would give you an error as being readonly when you tried to change it06:14
j416 then when you realise it was the wrong one, you'll have a hassle putting it back in order06:14
StuckMojo hehehe06:15
j416 in git, it's a piece of cake06:15
StuckMojo ok this is not productive :)06:15
arguring the merits is silly :P06:15
j416 StuckMojo: if you're working in the wrong repo, you should not be working at all06:15
you should be taking a break. :)06:15
StuckMojo the answer was: no06:15
and that's fine06:15
yeah i'm getting used to it06:15
j416 no?06:15
StuckMojo i've gotten into the habit of 1) running git status a lot06:16
and 2) taking care not to have multiple clones lying around06:16
j416 good habit.06:16
not sure why you'd want more than one clone06:17
StuckMojo i should probably just have one repo per gox06:17
box06:17
j416 I only have one local copy06:17
ThomasLocke joined06:17
StuckMojo well, i think i started with 2 for safety06:17
just in case i fubared the one i work in06:17
j416 backups are for safety06:17
hehe ok06:17
StuckMojo since i didn't know shit, i was gunshy06:17
much less so after last night, thanks to you06:17
j416 :)06:18
cool06:18
jankoprowski joined06:18
j416 you should spend an afternoon reading up on git and its internals06:18
StuckMojo again: much appriciated06:18
j416 it's a good investment06:18
you're very welcome06:18
StuckMojo yeah i've read a bit in the progit book06:18
up to now the problem has been not having call to use it06:19
unfortuately after adding this small functionality change to my billing software, i probably won't touch it again for quite some time06:19
and i'm stuck with p4 at $job06:19
j416 StuckMojo: I learnt most of what I know from here: http://book.git-scm.com/06:19
went through it step by step06:20
StuckMojo have you checked out progit, and if so, is this better?06:20
j416 ok.. regular use helps you remember. :)06:20
StuckMojo i suspect this is more complete06:20
j416 I've never read progit06:20
StuckMojo progit seems like basically a primer06:20
j416 maybe, dunno.06:20
StuckMojo see TOC: http://progit.org/book/06:20
mdel joined06:20
j416 if something is a primer, it should be Jun Hamano's book :)06:21
StuckMojo goes back to hacking06:22
joel_falcou left06:22
amerine joined06:23
kirthi joined06:24
justin-george joined06:25
OOPMan left06:27
guns left06:30
smolyn_ left06:30
duellsy left06:30
justin-george left06:32
Utkarsh joined06:36
rpbertp13 joined06:36
Pupeno left06:37
avinashhm left06:38
Utkarsh_ left06:39
avinashhm joined06:39
opalka joined06:40
orafu left06:40
orafu joined06:40
engla left06:41
giallu left06:42
Yuffster left06:42
ProLoser|Mobile joined06:43
ProLoser|Mobile how do i diff a specific file in 2 branches06:43
sudonotp1ease left06:43
rpbertp13 left06:44
sudonotplease joined06:44
coderdad left06:45
HG` left06:46
DiGitalX joined06:46
JohnFlux ProLoser|Mobile: git diff branchA branchB -- filename06:48
ProLoser|Mobile: I think06:48
jankoprowski Hi06:48
necromancer left06:49
jankoprowski I'm just wondering - there is so cool tools like gitalist and gitolite in example06:49
ProLoser|Mobile what's the -- for?06:49
Heimidal left06:49
JohnFlux ProLoser|Mobile: just to be clearly seperate branch name and filename06:49
ProLoser|Mobile: git usually guesses correctly though06:49
jankoprowski and You can't make use of it when You want choose Gitorius06:50
ProLoser|Mobile not working06:50
jankoprowski This is frustrating :|06:50
ProLoser|Mobile git diff master -- app/webroot/js/flight_options.js06:50
JohnFlux jankoprowski: what do they do?06:50
ProLoser|Mobile: and you're in a branch at the moment?06:50
ProLoser|Mobile ya i'm in 1 branch and want to compare a file to another branch06:51
HG` joined06:51
JohnFlux ProLoser|Mobile: I mean, you're in a branch that isn't master06:51
ProLoser|Mobile it just goes back to the console input06:51
i don't see anything, not even an error06:51
jankoprowski JohnFlux: The problem is Gitorius have own very stiff way they keep repository structure06:51
ProLoser|Mobile yes06:51
jankoprowski JohnFlux: Own hooks and other stuff...06:51
JohnFlux ProLoser|Mobile: you could try just "git diff master" and see if there are any changes at all06:51
ProLoser|Mobile: maybe there aren't differences :-)06:52
ProLoser|Mobile eh06:52
too many files06:52
and, well this is stupid, but my mac is old and overheating06:52
JohnFlux ProLoser|Mobile: you could do "git log flight_options.js"06:52
ProLoser|Mobile it doesn't scroll through pages of text well :/, i get visual taering06:52
jankoprowski JohnFlux: This is not the best approach because You must choose vs "Nice GUI Gitorius" and "Cool functionality in gitolite". You can't have both.06:52
JohnFlux: Onestly I discover that when I start writing gitogen - key manger/generator with LDAP/NIS integration06:53
sitaram jankoprowski: "GUI" != "cool" so problem solved ;-)06:53
JohnFlux ProLoser|Mobile: I mean, do git log flight_options.js and compare to git log master flight_options.js06:53
shyam_iyer joined06:53
madewokherd left06:53
jankoprowski sitaram: depends on your users06:54
JohnFlux ProLoser|Mobile: so if one has more commits than the other :-)06:54
jankoprowski sitaram: not from You06:54
JohnFlux jankoprowski: what is gitalist and gitolite? I'm not getting that part.06:54
jankoprowski JohnFlux: This two cool git-related stuff very useful when You are administrator.06:55
JohnFlux: First one allow You manage many repositories and access to it based on username/branchname etc...06:55
amondracorp left06:56
jankoprowski JohnFlux: Second is just nice (really nice) repository browser via webpage06:56
j0ran joined06:59
edulix joined07:03
sudonotplease left07:03
sudonotplease joined07:04
VxJasonxV left07:07
VxJasonxV joined07:07
echrsin left07:09
dreiss left07:12
Schmallon joined07:13
joel_falcou joined07:14
cbreak_work joined07:14
shyam_iyer left07:14
ExtraSpice joined07:15
odol joined07:15
kuhrt left07:16
t_b joined07:17
t_b left07:17
t0rc left07:17
Titosemi joined07:17
echrsin joined07:18
hobodave_ joined07:20
alip left07:23
hobodave left07:23
echrsin left07:24
Bass10 left07:24
alip joined07:25
fredp2 joined07:25
ramonmaruko joined07:26
roop joined07:27
asmega joined07:28
fredp2 left07:28
saidi joined07:28
k04n joined07:29
LouisJB left07:29
edulix left07:30
radarek joined07:32
hobodave joined07:32
echrsin joined07:33
jefferai left07:33
hobodave_ left07:33
TrineLocke joined07:34
Teo` left07:34
davetoo left07:36
priidu_ joined07:37
priidu joined07:37
thiago_home joined07:38
avinashhm left07:38
avinashhm joined07:39
giallu joined07:43
Deesl joined07:43
jankoprowski Is it possible to use Apache SSPI with TortoiseGit?07:44
tchan left07:44
ProLoser|Mobile what's sspi?07:44
tchan joined07:45
jankoprowski It is one of Single Sign On protocols from Microsoft07:45
henrikbjorn joined07:45
ProLoser|Mobile i don't get how the 2 are in any way related07:45
iband joined07:46
sako left07:47
jankoprowski You can configure Apache to authrize user using Microsoft Domain Server07:47
werdan7 joined07:47
jankoprowski If You use SPNEGO in IE or Kerberos in Firefox or Chrome You are authorized by default without Basic HTTP Auth prompt07:47
asmega left07:47
jankoprowski TortoiseSVN make use of it and use Your windows credential to autorize You in SVN server07:48
I'm wondering Git can do that.07:48
pratz joined07:48
Teo` joined07:49
JohnFlux jankoprowski: why would anyone tie themselves in so strongly to microsoft?07:49
jankoprowski: when perfectly good alternatives exist?07:49
saschpe joined07:51
jankoprowski JohnFlux: Noone tie to anyone. This is just support for some protocol.07:52
yaotti_awyaotti07:52
bigpeted joined07:52
jankoprowski JohnFlux: This is feature which make Git more suitable in corporate environment07:52
Git was enought long time ago for OpenSource projects07:52
Projects like gitolite bring git to corporates07:53
cbreak_work git works fine here07:53
kirthi left07:53
sdwrage_ joined07:54
sdwrage left07:54
jankoprowski cbreak_work: git works fine - but git access-control/authentication abilities barely exists07:54
cbreak_work you should look at gitolite07:54
it works really well07:54
jankoprowski I use gitolite07:54
JohnFlux cbreak_work: he's been talking about gitolite07:54
cbreak_work and it's very secure07:54
Unknown[OFF]Unknown[NF]07:55
jankoprowski I'm talkin about NTLM/SPNEGO/SSPI support which is present in Subversion world07:55
cbreak_work what's that good for?07:55
more security holes?07:55
jankoprowski cbreak_work: no - goodnes of Single Sign On.07:55
You are login to Windows domain07:56
cbreak_work so, more security holes07:56
jankoprowski and forgot about everything07:56
cbreak_work once I log into my computer, I don't have to enter passwords to use git either...07:56
jankoprowski cbreak_work: You don't like Windows/Microsoft stuff - be my guest. But this technology exists and this is nothing exotic to support them.07:56
cbreak_work without all that additional code that introduces potential vulnerabilities07:56
so it's simply not needed07:57
technology should not be used for it's own sake07:57
it should be used for a purpose07:57
jankoprowski cbreak_work: Purpose is very simple - comfort of users.07:57
sitaram jankoprowski: use gitolite in smart http mode, and use those fancy schmancy schemes to authenticate to apache. Done07:57
cbreak_work how's it more comfortable?07:57
JohnFlux jankoprowski: what's the advantage over the way gitorious does it - simply upload your ssh key to them?07:57
jankoprowski sitaram: Problem is in Git clients. They don't get it.07:58
cbreak_work you have to configure those weird ALL CAPS things07:58
Badiss joined07:58
cbreak_work I don't have to07:58
I just have to create a private key and tell the server my public key07:58
done.07:58
JohnFlux jankoprowski: is there something that git itself could do to help support SSPI ?07:59
cbreak_work and it doesn't even need to piggy back on hyper text transport07:59
teweWork joined07:59
jankoprowski cbreak_work: good luck with > 300 users which send You > 300 keys. By the way we don't use OpenSSH implementation which more complicate everything.08:00
cbreak_work: and one more thing08:00
cbreak_work if you're in a controlled environment, you can create the keys for them08:00
jefferai joined08:01
lostern joined08:01
hobodave left08:01
jankoprowski cbreak_work: yeap and manage them for many SVN/GIT repositories - this smell like "new dedicated employee is needed here".08:01
hobodave joined08:02
laj joined08:02
cbreak_work it doesn't add any complication to the normal user setup08:02
jankoprowski cbreak_work: If You will see corporate world where every permission is Active Directory group ... You will understand that SSH keys without integration with this whole LDAP stuff is not so god.08:03
cbreak_work at least not if you have centralized home directory storage08:03
jankoprowski I'm already writing "gitogen" project08:03
cbreak_work I am in a company, we don't use active directory08:03
Badiss left08:03
cbreak_work and we don't use this ldap stuff08:03
jankoprowski cbreak_work: many companies use LDAP stuff and ActiveDirectory and git is not enough good for them.08:04
tchan left08:04
mdel left08:04
cbreak_work active directory itself is rather useless08:05
jankoprowski cbreak_work: msysgit have --ntlm switch :D08:05
cbreak_work all it can deal with is windows08:05
jankoprowski: what does it do?08:05
tchan joined08:05
shentino left08:06
jankoprowski I must check exactly how this work today but somehow simplify logging from windows domain: http://goo.gl/2eCvN08:06
By the way08:06
If somone create client for Windows08:06
cbreak_work git client?08:07
jankoprowski This client should be windows-authentication-method compliance08:07
*compliante08:07
cbreak_work there are two I know of08:07
msysgit and cygwingit08:07
jankoprowski Yeap08:07
TortoiseGit frontend works with msysgit... i guess08:08
I must check this08:08
opalka left08:08
cbreak_work it does08:08
but badly08:08
thiago_home "windows authentication" is NTLM08:08
cbreak_work I think who ever made tortoise git wanted to strap svn working model onto git08:08
thiago_home points to tortoisesvn08:09
cbreak_work tortoise git can't even commit properly :/08:09
you always have to manually select what you want to commit08:09
odol left08:10
jankoprowski cbreak_work: What is proper git client for Windows in Your opinion?08:10
boombatower joined08:10
TrineLocke left08:10
pratz left08:10
harshu left08:10
opalka joined08:10
cbreak_work jankoprowski: msysgit is the most proper08:11
thiago_home msysgit08:11
cbreak_work but unfortunately, windows itself is rather terrible08:11
I mean the shell windows provides08:11
and it seems msysgit uses that for git-bash08:11
so it's not very comfortable08:11
ludde joined08:12
psoo joined08:12
chrischambers joined08:13
jankoprowski Yeap buy I think about some GUI stuff08:13
andy__ joined08:14
jankoprowski The wors thing about TortoiseGit which looks promissing they don't know how implement "saving passwords"08:14
jimmy1980 left08:14
jankoprowski http://code.google.com/p/tortoisegit/issues/detail?id=52208:14
This is frustrating08:14
kirthi joined08:15
pratz joined08:15
odol joined08:15
cbreak_work saving passwords?08:15
benishor_ left08:15
cbreak_work that's more like microsoft's job08:15
jimmy1980 joined08:15
Arrowmaster cbreak_work: yeah imho they should include Mintty in msysgit but they seem opposed to including anything extra in it, they wont even put nano in it08:15
cbreak_work the tortoise git authors have no means to save passwords08:15
jankoprowski Comparing TortoiseSVN when You must pass password for HTTP repository then can save it and use again08:15
harshu_ joined08:15
cbreak_work besides, git doesn't use passwords08:16
jhoepken joined08:16
cbreak_work most transports are either unauthenticated like dumb http or git08:16
or they use secure authentication like ssh or openssl08:16
psoo left08:17
psoo joined08:17
Arrowmaster for https there was a problem at one time that it didnt cache the authentication and asked multiple times for a single operation08:17
but that should be fixed08:17
cbreak_work asked for?08:18
pheaver left08:18
flaguy48 joined08:18
cbreak_work https isn't supposed to ask for passwords, just the client certificate if you want client authentication08:18
Arrowmaster i honestly cant remember if it had something to do with ssl certificates or http auth passwords08:18
cbreak_work hmm.08:18
Error404NotFound left08:18
cbreak_work people still use the http password stuff?08:19
I guess they don't really care about security:)08:19
Arrowmaster but that bug report on tortoisegit looks strange, i thought tortoisegit used msysgit underneath? why would _netrc work for msysgit but not tortoisegit08:19
OOPMan joined08:20
thiago_home NTLM http auth is a little more secure08:20
andy__ left08:21
psoo left08:21
workmad3 joined08:21
psoo joined08:22
d0k joined08:23
Error404NotFound joined08:25
gretch left08:25
gretch_ joined08:25
agateau joined08:25
gretch_ left08:25
psoo left08:25
gretch joined08:26
serard joined08:27
napster_123 left08:27
skoop joined08:28
thiago_home left08:29
[Po]lentino joined08:31
mishok13 joined08:31
tty1 left08:32
sdwrage_ left08:33
jankoprowski thiago_home: I think exactly like You08:33
bluenovember joined08:33
jankoprowski +Arrowmaster: _netrc work also with TortoiseGit08:34
Arrowmaster: I have just checked08:34
avinashhm left08:38
Schmallon left08:38
avinashhm joined08:39
ProLoser|Mobile left08:39
sejo left08:41
sejo joined08:41
jankoprowski cbreak_work: So Git won't ask me about password when I use SSL?08:41
Arrowmaster theres two different things, theres http auth and there are client side ssl certificates08:43
thermal joined08:44
Arrowmaster for example pushing to github over https only uses http auth, there are no certificates08:44
bigpeted left08:44
cbreak_work jankoprowski: no08:44
Arrowmaster but you can also use http auth with normal http instead of https08:44
cbreak_work git doesn't use passwords08:44
tty1 joined08:44
Arrowmaster its just a VERY bad idea08:44
cbreak_work unless you use the http based authentication I guess08:44
but who would do that...08:45
resmo_ joined08:45
cbreak_work with ssl you'd probably use certificates08:45
Arrowmaster technically git still doesnt use the password, i think it uses curl which would do the password part from _netrc08:45
ormaaj left08:45
cbreak_work because they are many magnitudes more secure than any password I could come up with08:45
and the passwords I come up with are probably many magnitudes more secure than passwords normal people come up with08:46
xiangfu left08:46
cbreak_work in general, passwords are weak.08:46
even if they are good, compared to public key crypto they are weak08:46
mikepack left08:46
ormaaj joined08:47
jankoprowski cbreak_work: How this SSL stuff works? It just verify user based on SSL certificate key?08:48
cbreak_work ssl?08:48
you generate a public and a private key08:48
tvw joined08:48
jankoprowski same as SSH - ok08:48
cbreak_work the public key gets signed by a certificate authority08:48
it's different because you have a CA08:48
jankoprowski We have CA in our company08:49
cbreak_work the server doesn't need to know every public key08:49
jankoprowski so this is not a problem08:49
cbreak_work unlike with SSH08:49
jankoprowski ok08:49
how server will now who is who?08:49
cbreak_work I am no ssl expert, but when setting up an ssl based vpn on my router, I just had to create my own ca and then tell the router to trust all certs signed by it08:49
same way you know that you are talking to your bank when you do online banking08:50
jimmy1980 left08:50
cbreak_work ssl certificates have a common name field08:50
jankoprowski right08:50
cbreak_work and tones of other field08:50
jankoprowski About banks: O08:50
cbreak_work the ssl standard is terrible :/08:50
(meaning the certification format)08:50
jankoprowski lol08:50
However08:51
cbreak_work there were lots of problems with implementing it08:51
jankoprowski It looks like git doesn't like http + password stuff08:51
crashanddie joined08:52
nailbar joined08:52
jankoprowski So the only way is write project which link ssh keys with LDAP/NIS stuff and manage them automaticaly08:52
to use gitolite without dedicated administrator in corporation08:52
So I just must end my project08:52
:|08:52
This is the only way08:52
airborn joined08:52
benishor_ joined08:52
theoros_ left08:53
Arrowmaster what version are you using?08:53
i thought all the http issues got fixed08:53
jankoprowski Arrowmaster: I didn't know anything about http issues08:53
xiong joined08:53
jankoprowski I just see that using Git comfortable way is not support Microsoft stuff08:54
And will not do this in future08:54
cbreak_work most comfortable way to use git is with ssh transport08:54
jankoprowski with gitolite08:54
cbreak_work yeah, or without08:54
jankoprowski but I see that gitolite is only half of solution08:54
cbreak_work you can use real user authentication too08:55
jankoprowski because You still must manage this stuff08:55
roop left08:55
jankoprowski some extra tool which run webpage where You come-in08:55
cbreak_work web page?08:55
jankoprowski yeap08:55
cbreak_work what'd you need that for?08:55
jimmy1980 joined08:56
emptyflask left08:56
jankoprowski Because Apache or IIS can check who You actually are (without asking You about credentials) and set REMOTE_USER variable for some PHP/Python/Ruby/PERL webpage in browser08:56
Then You can ask about LDAP/AD groups08:56
And based on some configuration files08:57
generated SSH pair of keys08:57
cbreak_work how would they know?08:57
nailbar left08:57
jankoprowski VAS/NTLM/SPNEGO/LDAP/Kerberos08:57
There are many ways08:57
SSPI08:57
cbreak_work hmm...08:58
that'd be a way08:58
jankoprowski however08:58
they generate ssh keys08:58
cbreak_work but wouldn't it be easier to just create an ssh key when users are created?08:58
and deposit that into the user's home folder?08:58
jankoprowski what with existing users?08:58
cbreak_work just run some shell script some time08:58
well, only works if you can access the user's homedirectories08:59
so if it's server backed or similar08:59
j416 you could always print the key on paper and have the user enter it.08:59
:P08:59
cbreak_work that sounds painful08:59
j416 it does08:59
especially when you find out you made a mistake in your input09:00
hobodave left09:00
cbreak_work must suck even more than installing windows09:01
do you still have to call microsoft to ask for permission to use the thing you bought? :)09:01
laj left09:01
Badiss joined09:02
amerine left09:03
Mage left09:03
Error404NotFound left09:03
MikeyB joined09:05
j416 :D09:05
napster_123 joined09:09
Error404NotFound joined09:09
tylerl left09:10
psankar left09:11
Agos joined09:12
shentino joined09:12
shentino left09:12
shentino joined09:12
ryec joined09:14
napster_123 left09:15
BiggFREE left09:16
PigeonFriend joined09:16
robinbowes left09:19
gretch left09:21
gretch joined09:21
echrsin left09:22
timj_ left09:25
lostern left09:25
timj joined09:25
saidi left09:25
MikhX left09:26
MikhX joined09:26
pdelgallego_ joined09:26
GaVrA left09:26
napster_123 joined09:27
xyfix joined09:27
echrsin joined09:27
echrsin_ joined09:27
pfrenssen joined09:28
darkredandyellow joined09:28
Neo left09:29
Unknown[NF]Unknown[OFF]09:29
dest joined09:30
dest Hi. Is there a way to know how many times a git repo has been cloned ?09:30
roop joined09:31
echrsin left09:32
j416 dest: no.09:32
gretch left09:33
j416 the repository being cloned is not modified in any way.09:33
k04n left09:33
Neo joined09:33
cbz It's like asking 'is there any way to know how many times a file has been copied"09:34
gretch joined09:34
teweWork i want to push tags and if i try git push --tags looks like all the commits are pushed, or something strange happens because my pre-receive hook gets all the commits and doesn't allow the push (it checks the commit msg and old commits have invalid messages)09:36
and i only have one new tag09:36
dest j416: thanks09:37
j416 yw09:37
xyfix Hi, I'm new to git ( coming from cvs ), I have just committed some files from one local copy and wanted to update my second local copy with the code that I just committed. I've read a few stuff on the net ( fetch-merge or pull ) and done all of those things but I still can update my second local working dir. WindowsXP-TortoiseGit09:37
tango_ xyfix: my guess is that you are trying to push from one local copy to the other09:38
which you shouldn't, faq non-bare09:38
jast Pushing to non-bare repositories is discouraged: please see https://git.wiki.kernel.org/index.php/GitFaq#non-bare [automatic message]09:38
avinashhm left09:38
avinashhm joined09:39
xyfix tango_: I committed my local copy to a repository and the tried to update my second local working dir from this repository. I've done a fresh new checkout and I can see my code that I committed09:39
tango_: I'm not pushing from one local copy to another local copy09:41
napster_123 left09:41
tango_ first of all, small syntax change: you don't commit a local copy to a repository. you clone, push and fetch between repositories. commits are always local.09:41
next: ok, going through a remote is a good thing. so what is it exactly that is failing for you?09:41
tizzo joined09:42
pdelgallego__ joined09:42
teweWork xyfix: i don't understand u can't update your second repo but if u clone a new one there are your changes?09:43
xyfix when I run an update( fetch-merge) on my second local working dir I don't see the code that I just cloned to the repository09:43
yep09:43
pdelgallego joined09:43
gretch left09:43
xyfix @ teweWork09:43
when I do a fetch on my second local copy it ends up in a success09:44
tatsuya joined09:45
xyfix but no code that I just cloned from my first local copy09:45
rseifert joined09:45
j416 xyfix: files are not automaticlly checked out when you fetch09:46
fetch just updates your (local copies of) remote branches09:46
maybe that is the issue.09:46
xyfix but how can I update my local copy from the HEAD( master )09:47
robinbowes joined09:47
j416 xyfix: you have to merge or rebase your local branches09:47
kalusn joined09:47
j416 'git pull' does 'git fetch' and 'git merge' in one go.09:48
koltroll joined09:48
teweWork i don't know what feedback tortoisegit gives u but if it only sais 'Success' then it's not enough information :)09:48
j416 I don't know much about windows and git, sorry.09:49
xyfix this is the output of my git pull http://pastebin.com/cAYitCAQ09:49
ntoll joined09:49
j416 xyfix: this is because you have changes in that file09:50
do something with those changes first.09:50
xyfix yes but I expect a conflict resolve option09:50
teweWork back to my question why does the pre-receive hook gets all the commits if i do git push --tags ?09:50
j416 xyfix: git merges committed things, you have to commit it first.09:50
xyfix: any changed not staged or committed, are not in git.09:51
s/changed/changes/09:51
xyfix hmmm? but the second local copy is not finished yet ( code ways ) so I don't want to commit just update09:52
j416 xyfix: then, commit and rebase.09:52
git add [your stuff] && git commit -m 'Add stuff' && git pull --rebase09:53
I think should do it.09:53
(understand that before you run it)09:53
:)09:53
^Simon^ joined09:53
^Simon^ Every time I merge the live branch into my working branch I get a conflict on a file that has to be different between live and dev. It's only a couple of lines that are different, the rest of the file is identical.09:54
tvw left09:54
j416 does anyone know what the deal is with git adding a nested repo as a submodule, but without a config file?09:55
I'd be interested in any info at all on this.09:55
xyfix j416: I already added the new files in my second local copy, but I didn't know that I should also commit them because that will put them in the repository ( I think )09:55
priidu left09:55
j416 xyfix: "git add" merely adds the changes to the index. You have to run "git commit" to actually put them in the branch.09:55
so to speak.09:56
patmaddox left09:56
torbenh3 joined09:56
j416 xyfix: you should invest an afternoon reading this: http://book.git-scm.com/09:56
you'll be glad you did09:57
especially since you seem a bit affected by your previous experience :)09:57
xyfix j416: you're absolutely right but why not start this morning ( right now)09:57
:-)09:57
j416 xyfix: why not :)09:57
^Simon^ myfix/j416 - getting good with git is quite a good book too09:57
j416 ^Simon^: but the git community book is free :P09:58
(why is there an illustration of a computer mouse on the cover of that book -- I can't recall ever using the mouse when working with git)09:59
haha.09:59
xyfix thanks you guys, it seems like I have homework to do :-)09:59
kirthi left09:59
j416 xyfix: have fun :)09:59
kirthi joined10:00
j416 xyfix: there is a lot of good material out there, I recommended that only because I learnt most of what I know from it.10:00
at least I think I did.10:00
gretch joined10:01
andrewvo1 joined10:01
foocraft joined10:01
^Simon^ j416 - agreed yes, the git community book is free, and it's worth reading..10:02
LouisJB joined10:02
andrewvo1 left10:03
j416 ^Simon^: not to say the book you suggest isn't good, but I seem to have no way of finding out except buying it :P10:03
^Simon^ I picked it up when it was free.10:03
j416 aha, sample content availabel10:03
andrewvos left10:03
^Simon^ it was "book of the week" or something.10:03
I figured out how to use git add -i from it last night.10:04
andrewvos joined10:04
workmad3 left10:04
j416 git add -i10:07
never got around that one10:07
tried it a couple times10:07
^Simon^ I wanted to commit a couple of parts of a file, but not all of it.10:07
j416 I use "git add -p" from time to time though10:07
^Simon^ git add -i10:07
j416 I thought -i was too messy10:07
-p is more straightforward10:07
^Simon^ I think it's the same.. I chose 'p' from the menu.10:07
j416 yep10:07
-i is just some kind of wrapper thing for a bunch of commands10:08
^Simon^ getting conflicts on the file now though :(10:08
beekhof left10:08
Unknown[OFF] left10:09
^Simon^ I merged the live branch back into my working branch and now I have: (a) changes to be commited which are one of the files that didn't conflict.(b) Unmerged paths (the conflicting file), (c) changed but not updated - something ready to git add/git commit.10:09
I resolved the conflict in the file, but don't want to commit it.. just leave it as a modified file.10:10
seems to happen regularly that it goes weird10:10
j416 ^Simon^: well, if you don't commit, there will be no merge10:11
that means, you don't want to merge?10:11
perhaps rebase is what you want10:11
^Simon^ there were no new commits on my branch - they'd all been pushed and then merged into staging and pushed back to the repo.10:11
j416 (not sure why you are merging the live branch back into your working branch, though)10:11
^Simon^ we use the integration manager approach (or my take on it) - devs have personal repos that get pulled into staging ready to go live and then pushed back to a central repo which has updates pulled back to each personal10:12
tizzo left10:13
tty1 left10:13
^Simon^ maybe I just don't do it right.10:13
I have a branch "main" which tracks the central repository and is pretty much a copy of the live code.10:14
I also have master which is my working version10:14
I have to update the main branch using git checkout main; git pull10:15
and then git checkout master10:15
git merge main10:15
echrsin_ left10:15
^Simon^ should I just skip the main branch and pull the remote main branch into my working branch?10:16
eijk joined10:17
^Simon^ I couldn't find very much useful information on exactly how to setup the integration manager method of working so cobbled it together with the few git commands I knew. :)10:17
kirthi left10:18
kirthi joined10:18
j416 ^Simon^: man gitworkflows10:19
jast the 'gitworkflows' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/gitworkflows.html [automatic message]10:19
matjas joined10:19
j416 also google git workflows10:19
jeeger joined10:19
j416 there are many ways...10:19
which is right for you, I don't know.10:19
Flcn left10:19
jeeger Greetings! Is it possible to add branch directories to an svn cloned repository after the initial clone? I can't really find anything in the docs (git-config and git-svn).10:20
drizzd_ joined10:20
^Simon^ do you mean creating a new branch on the svn repository?10:21
mithro joined10:21
^Simon^ or just creating one in git based on the svn clone?10:21
pratz left10:22
jeeger ^Simon^: I didn't pass any --branch arguments to the initial git svn clone, and want to add them later so I can push branches to svn10:22
^Simon^ I'm afraid I don't know - when I cloned our svn repo, it was for the initial import into git and we just ditched svn afterwards.10:23
we only once used a branch in svn10:23
yotaff joined10:23
jeeger *shudder*10:23
drizzd left10:23
^Simon^ most of the dev stuff sucked because we'd keep the files modified on the personal dev systems until the feature was ready to go live - got really messy, really quickly.10:24
not to mention no way to recover a working but not ready version from a few days before if a change broke something.10:24
since switching to git it's been a bit of a learning curve but on the whole it's so much tidier.10:25
pratz joined10:25
bentob0x joined10:26
thermal left10:27
dr_win joined10:27
transplant joined10:28
avar what's the "git show --stat sha" that also shows you the diff? Just adds a stat to the log message10:30
mithro left10:30
foocraft left10:30
avar ah, just add -p10:31
napster_123 joined10:32
petrux joined10:33
bauruine left10:35
artefon joined10:36
winterweaver joined10:38
Weust`afkWeust`10:38
avinashhm left10:38
Neo left10:38
avinashhm joined10:39
roger_padactor left10:39
transplant left10:39
tvw joined10:40
tvw left10:40
tvw joined10:41
bigpeted joined10:41
Neo joined10:43
_pingu joined10:43
Unknown[OFF] joined10:43
snearch joined10:44
Cro_Crx joined10:47
fjb4 joined10:47
devanjan left10:47
mduft joined10:48
mduft hey :)10:48
mithro joined10:49
mduft i have a question about moving from svn to git (acutally first of all testing git, and _possibly_ moving to it, so a git-svn side-by-side solution would be cool for the start):10:49
is there a way to get multiple svn repositories into one git master branch?10:49
markelikalderon left10:49
jeeger oof10:50
Don't think so10:50
mduft i have https://server/{dir1,dir2}/{trunk,branches,tags}/dirNproj110:50
foocraft joined10:50
mduft and want a master git repo with dir1proj1 and dir2proj2 both in it...10:50
Titosemi1 joined10:51
mduft hmmm that would be bad...10:51
i have approx 70 svn repositories, and don't want to end up with 70 git repositories...10:51
Titosemi left10:51
^Simon^ why not?10:52
do you have the svn repositories as sub-directories of each other or something?10:52
mduft because i want to move stuff from one to the other... isn't this easier if all things end up in one repo?10:52
^Simon^ not necessarily.10:52
selckin many small repos is usually better10:52
mduft ah, and i need to get EGit working on the repos...10:52
hmmm10:52
^Simon^ you can add remotes to the various repositories and pull stuff from a different repo.10:53
mduft it's a bit eclipse-driven java project with > 300 plugins...10:53
hmm10:53
ok, i see10:53
^Simon^ might be a pain to setup if you're talking all 70 need to pull from the other 69.. but could just write a script to run in each of the repos to add the remotes.10:53
mduft right, that would not be so much of a problem10:54
i just fear the developers used to svn will complain about the complexity... even though in svn it was 70 repos, they were all on the same root, so the eclipse repo browser hat one server, and multiple directories they easily navigated10:55
now they'd have to add 70 repos separately and browse in the right one to get their plugins...10:55
any other ideas to get this right (tm)?10:55
fr0sty-away left10:56
nicoulaj joined10:56
[Po]lentino__ joined10:57
fr0sty-away joined10:57
[Po]lentino left10:58
rgr joined10:58
^Simon^ Not really. We moved from a single svn repo to a single central git repo + personal repos for each dev.10:58
and we don't use any kind of repo browser10:58
unless cd/ls counts :)10:59
mduft heh10:59
yeah, thats what i use too...10:59
still i need to construct something that our windows-guys can use too ;)10:59
Noldorin_ joined11:00
ndim mduft: If there was a git repo for a prototype plugin, and all the other plugins were basically forks of that, that would make merging things between plugins easier. However, I have no idea how to achieve that with your current svn or future git-svn setup.11:01
mduft hmmm... ok, thanks anyway. need ${food} to think about it...11:02
_Vi joined11:02
Cromulent joined11:04
j416 ^Simon^: do you guys do windows dev?11:05
deek0146 joined11:05
j416 ^Simon^: I'd love to hear your reasons for switching from svn.11:06
^Simon^ it's all perl scripts/mod_perl stuff.11:06
j416 ok11:06
[Po]lentino__ left11:07
^Simon^ main reason is that svn repo access was setup over ssh and we have a some not-very-technical people doing the website changes - html/css/images but it's all in with the perl code in the repo and didn't want them being able to commit stuff randomly.11:07
verdon joined11:08
^Simon^ it's too easy to type "svn commit" and have it commit EVERYTHING that's changed instead of specific files.11:08
markelikalderon joined11:08
pi_benjamin joined11:09
[Po]lentino joined11:09
^Simon^ going the integration manager approach with git means the more technical have to put stuff live, but the less technical can still use commit to save working (but not quite ready) versions of things as often as they like without impacting anything.11:09
andrewvos left11:09
deek0146 left11:10
^Simon^ that and being able to create a branch for each new feature.. very handy..11:10
svn branching was a pain.11:10
hyperair left11:10
^Simon^ of course, getting the least technical web person using git is proving a bit of a struggle.. "it's too hard.. I'll never learn all these commands.. can you just put these two files live for me"11:11
tibbe joined11:14
j416 ^Simon^: thanks11:14
^Simon^: yeah I can imagine..11:14
otoh11:14
you employed the person11:14
get the person to learn11:14
or, employ a person who can actually do the work :)11:14
tibbe What's a good flow for maintaing bug fix branches for stable releases? I tried to look into Linus' Linux repo for inspiration but there's only a master branch there.11:14
^Simon^ laughs, "Actually, he employed me.. it's the boss!"11:14
j416 hahahaha11:14
fire the boss11:15
krfs joined11:16
^Simon^ To be fair, he's got the hang of building html from templates using the command line11:16
krfs left11:16
^Simon^ and it's only really git add/git commit/git push that he needs to learn11:16
fenrig joined11:17
fenrig Hi11:17
voker57 joined11:18
j416 sec phone11:18
fenrig i get a "fatal: 'gitosis-admin.git' does not appear to be a git repository" after following the guides on the net for setting up a git :o I tried it 2 times already and I always seem to get the same error (did this on 2 different vm's with a different linux flavor)11:19
voker57 How to apply patches made by git-format-patch so git would create appropriate commits? git apply only applies the patch11:20
frerich ^Simon^: Youre situation sounds familiar. What we did was to write a little script which would run 'git commit -a && git push'. So our non-technical stuff could edit some web pages (or, documentation files in our case) and then run 'git publish' and it's live.11:20
ntoll left11:20
frerich ^Simon^: It's not nice and it doesn't leverage any of the good git provides but it got the job done for him.11:20
voker57: Try 'git am *.patch'11:21
fenrig after doing this: "sudo -H -u git gitosis-init < /path/to/public_key.pub"11:21
voker57 frerich: Patch format detection failed.11:21
^Simon^ frerich - he has a habit of dumping all kinds of work-in-progress images all over the filesystem.11:21
I wouldn't want him adding and committing those files11:21
fenrig - try gitolite instead :)11:22
voker57 ah, never mind, patches were mangled a bit11:22
frerich ^Simon^: Hm I see. So maybe let the 'publish' script accept file names and/or directories.11:22
So you run 'git add <arguments>; git commit; git push'11:22
fenrig ^Simon^: do u have a good working guide or howto?11:23
^Simon^ nods to frerich, "you can create extra git commands by putting a script somewhere to make 'git publish' valid?"11:23
^Simon^ fenrig - one sec.11:23
frerich ^Simon^: Yep, actually there's a neat trick: I think you can have a script called 'git-publish' somewhere in your path and then run 'git publish' (the 'git' command will attempt to run 'git-publish').11:24
gretch left11:24
frerich ^Simon^: At least that works on Linux and Mac if I remember correctly. Not sure whether it works with batch files on Windows.11:24
^Simon^ fenrig - https://github.com/sitaramc/gitolite11:25
fjb4 left11:25
^Simon^ the docs are pretty straightforward - if you've partially setup gitosis it's quite similar.11:25
frerich - not a problem, all dev is on linux and mac anyway.11:26
frerich ^Simon^: Lucky you! :-)11:26
zealiod joined11:26
tibbe left11:28
fenrig ^Simon^: thx :D11:28
StrawberryKitten joined11:28
ChanServ set mode: +v11:29
StrawberryKitten left11:30
fenrig ^Simon^: is there a trick to undo all gitosis witchcraft?11:30
^Simon^ I think I just removed the gitosis-admin repository and deleted a couple of files.11:31
there's a doc for migration from gitosis.11:31
http://sitaramc.github.com/gitolite/doc/migrate.html11:31
I used the first 4 steps from that, and then went through the install.11:32
we didn't have a fully working gitosis install at that point.11:32
OOPMan left11:33
^Simon^ wasn't worth using gl-conf-convert11:33
markelikalderon left11:34
simplechat joined11:35
simplechat left11:35
simplechat joined11:35
eighty4 joined11:35
zorzar_away left11:38
Unknown[OFF] left11:38
avinashhm left11:38
hyperair joined11:38
avinashhm joined11:39
radarek left11:43
pi_benjamin left11:43
pfrenssen_ joined11:44
v0n left11:44
angelsl left11:46
radarek joined11:46
teweWork why does the pre-receive hook get all the commits if i do git push --tags ?11:47
pfrenssen left11:47
cbz how do i see the git dag from the command line?11:48
FauxFaux log --graph11:49
esc cbz: git log --oneline --graph --decorate --all11:49
FauxFaux I use log --graph --pretty=format:'%Cred%h%Creset -%C(yellow)%d%Creset %s %Cgreen(%cr) %C(bold blue)<%an>%Creset' --abbrev-commit --date=relative11:50
zorzar_away joined11:51
rodrigo_golive joined11:52
rodrigo_golive morning all11:52
cbz can i put that in as an alias somewhere?11:54
dl9pf_ joined11:54
dl9pf left11:54
rodrigo_golive so, I already installed gitosis successful on my personal server, but now I'm having trouble installing on a virtualized linux system; it appears to not load the .bashrc or .bash_profile PATH var11:54
Cromulent left11:54
rodrigo_golive what I can do now??11:54
^Simon^ try gitolite instead? :)11:55
rodrigo_golive ^Simon^: why gitolite?11:55
^Simon^ Are the virtual server and personal server running the same OS?11:55
divVerent first of all, because gitolite is still maintained11:56
kirthi left11:56
rodrigo_golive ^Simon^: yes, slack13.111:56
divVerent but anyway11:56
how are these files needed by gitosis?11:56
gitolite does not use them11:56
rodrigo_golive divVerent: they contains the path to a bin dir, that have the gitosis-serve command11:57
divVerent ah, okay11:57
rodrigo_golive and others :p11:57
pfrenssen_ left11:57
divVerent can you give the ssh users a forced command that sets the environment, then starts the gitosis server process?11:58
of course, if you do that, the setup becomes more and more gitolite-like :P11:58
rodrigo_golive divVerent: well, I tried to force with the environmet file on .ssh dir, but not works too11:59
hyperboreean joined11:59
divVerent that file must be allowed in the ssh server's configuration11:59
rodrigo_golive divVerent: already done it too, and restarted the sshd, but no luck12:00
bluenovember FauxFaux, that's very nice, ty. *steals*12:00
divVerent see PermitUserEnvironment in man sshd_config12:00
the option looks very dangerous to enable, though12:00
bauruine joined12:01
fjb4 joined12:02
^Simon^ could you put something in /etc/profile ?12:03
or the equivalent12:03
not ideal, but perhaps use if to test the username and only do it if the username is 'git'?12:03
StuckMojo how do i pull in a change to a single file from one branch to another?12:04
divVerent ^Simon^: /etc/profile is also not run12:04
for a non-login shell12:04
^Simon^ ah.12:04
StuckMojo im in a branch, change a few files. one of those file's changes i'd like to merge into master12:04
rodrigo_golive ^Simon^: yes, I'm using a "git" user12:06
StuckMojo hmmm...git checkout branch; git checkout <commit> filename12:06
then commit?12:06
borwar joined12:06
rodrigo_golive the problem thats is that I'm using exactly the same conf that I have used on my personal server... it's driving me crazy :p12:06
srisw left12:06
rodrigo_golive maybe I'll take a look on gitolite :p12:07
StuckMojo ah yes, that did it12:07
zealiod left12:07
^Simon^ rodrigo_golive - http://sitaramc.github.com/gitolite/doc/migrate.html12:08
migration from gitosis.12:08
rodrigo_golive ^Simon^: :) thanks12:08
^Simon^ ponders actually getting some work done this morning..12:08
^Simon^ cbz/FauxFaux - was there an answer to this?<cbz> can i put that in as an alias somewhere?12:10
spaceonline joined12:11
dthigpen joined12:12
pdelgallego__ left12:12
pdelgallego left12:13
pdelgallego_ left12:13
stamina_ joined12:14
^Simon^ left12:15
Unknown[OFF] joined12:16
borwar left12:19
borwar joined12:22
borwar left12:22
rgr left12:22
borwar joined12:23
Agos I have a local developement repo, and would like to “put” it on the test server... like an invese git clone. what is the way to this?12:24
jettero pushing to a checked out repo is naughty12:25
pushing to a remote is called 'git push'12:25
pushing to an empty directory might be ssh host (cd somewhere && git clone blarg) or maybe look at rsync12:26
what is put to you?12:26
cbreak_work the inverse to git clone is rm -rf12:26
bartj joined12:26
jettero haha12:26
Agos ahaha12:26
I mean12:26
cbreak_work remote clone?12:26
Agos I would like to clone my local repo to the remote host12:27
is copying the files the same?12:27
cbreak_work usually you wold make bare clone from the other host, or just rsync over a bare clone12:27
jettero rsync would work fine... if you're not going to edit there, I would rsync -vaP --exclude .git . remotehost:/path/12:27
cbreak_work if you want to push to it, make sure it is bare12:27
teweWork i can't find how can i set merge --no-ff the default12:27
nvm found it12:28
fjb4 left12:28
fjb4 joined12:28
MikhX left12:28
cbreak_work teweWork: what was it?12:28
Agos let me rephrase.12:28
jettero please do12:29
cbreak_work teweWork: only solution I can think of is making an alias12:29
didn't find a config field for it12:29
teweWork cbreak_work: git config branch.master.mergeoptions "--no-ff"12:29
cbreak_work ! clever12:29
jettero are you saying not clever or wow clever?12:29
goshawk joined12:29
cbreak_work ! sounds like the attention sound12:30
Ryback_ joined12:30
cbreak_work when enemies notice solid snake12:30
jettero :)12:30
teweWork :)12:30
i wonder if i can set this globally for every master branch but guess not12:31
yaottiyaotti_aw12:31
cbreak_work every master branch?12:31
Agos current situation is: normal repo to which I'm committing to is on my laptop. Now the test and production hosts are ready. I'd like to set up a repo on the development server so that I can push to it as a remote12:31
cbreak_work you only have one of those per repo12:31
maybe you can set it with git config --global12:31
teweWork yeah i meant that12:31
cbreak_work (so it's in ~/.gitconfig)12:31
pdelgallego joined12:32
pdelgallego_ joined12:32
cbreak_work Agos: just make a bare clone and push to that12:32
pdelgallego__ joined12:32
OOPMan joined12:32
cbreak_work and when you work on the test/production hosts, do a git pull from the bare12:32
bauruine left12:32
Agos cbreak_work then I think I have a very wrong idea of what a bare clone is.12:33
cbreak_work there are some people who abuse git for deployment12:33
Agos: open a .git directory12:33
a bare clone is approximately that12:33
teweWork it works globally12:33
cbreak_work a repository you can push to12:33
teweWork: neat.12:33
bkosborne joined12:34
cbreak_work teweWork: although I would hate to have --no-ff in there :) I'd probably add -ff12:34
Agos but the point is, I can't go on the server and do git clone --bare /mymachineathome/12:34
I'd love to do the inverse12:34
from my machine12:34
cbreak_work Agos: you can do git clone --bare on your machine12:34
and then rsync the clone over12:34
teweWork it's company policy right now (i don't think it's a good idea either but time will tell :))12:34
cbreak_work or ftp12:34
Agos got it12:34
thanks cbreak_work12:34
bauruine joined12:35
cbreak_work Agos: you have ssh access to the machine? without that you won't be able to push12:35
at least not sanely... :)12:35
Agos cbreak_work yes, I do :)12:35
nicoulaj left12:35
snearch left12:35
teweWork and u can override it with --ff :)12:36
Agos cbreak_work last question: why bare?12:36
nicoulaj joined12:36
cbreak_work Agos: because you can only push to bare repositories (without getting into trouble)12:37
shruggar joined12:37
teweWork back to my previous question, why does the pre-receive hook get all the commits if i do git push --tags ?12:37
cbreak_work all commits?12:37
it should get all commits pointed to by tags12:38
avinashhm left12:38
avinashhm joined12:39
linusoleander joined12:39
linusoleander left12:39
Agos cbreak_work that's interesting. so I'll put my bare clone somewhere (I do have a nice repos directory), then have a full-blown "normal" repo in, say, my /www/ dir which will pull from the bare clone?12:39
Dave^| joined12:39
teweWork cbreak_work: it gets every single commit12:42
bkosborne left12:42
hegge left12:43
[Po]lentino left12:44
shruggar when I push into gitolite-admin, the message "already on master" is printed. Can this be avoided?12:45
nullvariable left12:46
lostern joined12:46
shruggar it's due to "git checkout -f master", in the post-update hook. But I don't know the intended purpose of that command12:46
OOPMan left12:47
sitaram shruggar: other than redirecting output to devnull, no...12:47
and I'd rather not do that12:47
shruggar well I could --quiet, but same effect, I assume12:48
lostern left12:48
nullvariable joined12:48
shruggar what is it trying to do? I'm wondering if it's something I could detect whether it's necessary12:48
hegge joined12:48
sitaram shruggar: does --quiet suppress only "normal" messages while allowing error messages through? (if any)12:48
shruggar: if so, that would be fine12:49
shruggar I believe so12:49
thermal joined12:49
sitaram shruggar: ok let me try and confirm it by some experimentation and then add it12:49
shruggar but I don't know what notes about differing files being overwritten due to -f count as12:49
sitaram shruggar: why is it bothering you anyway? just mentally ignore it :)12:49
shruggar sitaram, because it's otherwise completely silent, so it means I get a cron notification instead of none12:50
sitaram you're pushing to the admin repo via *cron*?12:51
shruggar: if you can confirm that --quiet thing under some common checkout errors, could you send me an email? I may not be on for a while now... going drinking ;-)12:51
--quiet thing ==> does not quiet real errors12:52
shruggar yeah, that key updating thing I was trying a couple of days ago. Just updating the key database without storing it in the admin had various pointy corners which I can't recall at the moment but could probably think of if pressed12:52
ah, right. I was becaause if I then made a regular change in admin, all the generated keys would be cleared out12:54
I'll send myself an e-mail telling me to play with it when I get home. I'm also going to lunch now12:54
jeeger shruggar: sounds like a plan.12:55
hyperair left12:56
Sigma left12:58
fenrig Hahaha i've got a noob question12:59
Textmode left12:59
fenrig The editor i get when doing "gl-setup" how do i close it without saving anything (I'm used to nano)12:59
roger_padactor joined13:00
DiGitalX left13:01
henrikbjorn left13:02
fenrig It would be really handy If someone can help me cause I really don't know the editor13:02
MAK-Jr joined13:02
j416 fenrig: what is the editor?13:04
(you can set the default editor in git config)13:04
loa joined13:04
loa hello how i can remove old commits?13:04
fenrig j416: that's what I don't know, it will be the default :o13:04
teukka fenrig: if it is vi or vim, type ":q<enter>"13:05
j416 fenrig: what does it look like?13:05
loa for example i want save only 5 last commits.13:05
j416 loa: do you want to remove them entirely, or do you want to log that you removed them?13:05
fenrig black with white text :p and when I press f1 the screen flashes white for a sec13:05
loa j416, just remove.13:05
j416 loa: removing them entirely means you're altering your history, be careful if you have pushed your changes and others have fetched.13:05
TheBorg[ReGen] left13:06
j416 loa: man git-reset13:06
jast loa: the 'git-reset' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-reset.html [automatic message]13:06
j416 is your friend.13:06
fenrig j416: it's not distinguisable I really don't see any help13:06
j416 fenrig: are there wigglies on the left hand side? (~)?13:06
loa j416, i wont to remove "old" commits, not set to old commit state.13:06
j416 loa: not sure what you mean by that13:07
you want to remove commits that are early in history?13:07
project2501a joined13:07
thijso joined13:07
fenrig j416: No, it looks a lot like nano but it's not nano :s and i don't think it's vi either13:07
j416 loa: in that case, man git-rebase13:07
jast loa: the 'git-rebase' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-rebase.html [automatic message]13:07
loa j416, yes.13:07
j416 "git rebase -i" may do what you ask.13:07
Cromulent joined13:08
j416 for more advanced features, check out man git-filter-branch13:08
jast the 'git-filter-branch' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-filter-branch.html [automatic message]13:08
j416 fenrig: weird... o_O13:08
emacs perhaps?13:08
fenrig i tried pressing escape and afterwards X13:08
and twice escape and afterwards X13:09
j416 fenrig: if it's emacs, I believe it's ctrl+x ctrl+c13:09
to exit13:09
if it's vim, it's :q13:09
thermal left13:09
j416 (then enter)13:09
fenrig j416: thx it's vim i think but uhm when i do :q it says "No write since last change (:quit! overrides13:10
j416 fenrig: ok!13:10
fenrig: if you don't want to save any of the changes you made, do :q!13:10
<enter>13:10
(q is shorthand for quit)13:11
fenrig j416: :quit! saved me :o but when i do that the file doesn't get saved13:12
j416 fenrig: of course not13:12
you said you didn't want to save the changes.13:12
fenrig :D got thx :o13:12
j416 if you wanted to save it you should have done :wq13:13
fenrig but uhm I'm getting a "fatal: empty ident [email@hidden.address] not allowed"13:13
j416 supposedly there was a reason it brought up the editor...13:13
MAK-Jr left13:13
MAK-Jr joined13:13
j416 perhaps setting your identity first would help13:13
MAK-Jr left13:13
j416 fenrig: (man git-config)13:14
jast fenrig: the 'git-config' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-config.html [automatic message]13:14
miloshadzic joined13:14
j416 (user.name, user.email)13:14
mw joined13:15
mw left13:15
mw joined13:15
j416 fenrig: you might also want to set core.editor to something other than vim, if you're not friends with that yet.13:15
hunterloftis joined13:15
fenrig j416: I'm following the installation guide, why don't they mention this stuff13:16
hunterloftis left13:17
j416 fenrig: because they assume you know what you're doing? :P13:18
I dunno.13:18
fenrig j416: I don't have any exp with setting up git :o and the guide says it's for people without exp :d13:19
j416 dunno13:19
roop left13:19
j416 I never used what you mention13:19
fenrig j416: could u give me a better guide then?13:20
Pewpewarrows left13:20
j416 to installing git?13:20
what OS?13:21
and what is that you've been trying to use? where is it from?13:21
fenrig j416: for installing gitolite :)13:22
j416 oh ok13:22
never done that13:22
sorry13:22
fenrig j416: linux (archlinux)13:22
j416 can't help you thre13:22
there*13:22
esc hey, what units will git cat-file output if i ask for size (-s) ?13:22
fenrig it's for setting up gitserver :o13:22
j416 fenrig: perhaps sitaram can help13:22
his away message says "gone drinking" so maybe he's out for a while :P13:23
harshu_ left13:23
TheBorg[ReGen] joined13:23
j416 esc: man git-cat-file explains that13:24
jast esc: the 'git-cat-file' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-cat-file.html [automatic message]13:24
j416 esc: from that man page: "If -s is specified, the size of the <object> in bytes."13:24
schweizer hi13:24
esc j416: thanks13:24
j416 yw13:24
banisterfiend joined13:24
schweizer how can i make git difff" to word-wrap the output?13:24
-f13:25
banisterfiend how do i squash everything into a single commit when i merge a branch?13:25
j416 schweizer: perhaps set your terminal to wrap13:25
schweizer: a diff shouldn't be wrapped, or it isn't a proper diff...13:25
dunno if there is an option to wrap it.13:25
banisterfiend: git rebase -i13:25
banisterfiend: squash first, then merge13:26
banisterfiend without rebasing13:26
j416 then you don't13:26
banisterfiend i think there's like a --squash option or some shit i can pass to merge?13:26
malesca banisterfiend: git merge --squash mybranch13:26
j416 ah13:26
juanjoc joined13:26
banisterfiend malesca: awesome thanks13:26
malesca will put the changes in your working dir, so you can make a single commit from that point13:26
fenrig aaarggh setting up git is driving me crazy :o13:26
j416 that will make a commit that doesn't look like a merge, right13:26
jaql joined13:26
j416 fenrig: I thought you were setting up gitolite13:27
fenrig j416: yeah a git server, tried it already with gitosis and gitolite :o13:27
j416 gitolite is the way to go of those13:27
I believe gitosis isn't very active these days13:27
j416 shrugs13:28
j416 fenrig: why not use github or similar if you find it hard to set up your own server?13:28
shiba_yu36 joined13:28
fenrig j416: cause I want my own gitserver and because it's for my internship to find out how to do it :S13:29
j416 hahaha13:29
ok13:29
then start reading :)13:29
fenrig :D i'm doing it for 3 days long now :o i just hate when something doesn't work13:29
j416 read some more then :)13:29
for comparison, it took me a full two weeks to set up my first mail server.13:30
miloshadzic left13:30
fenrig j416: XD i know but still it drives me crazy13:30
pratz left13:30
j416 fenrig: a hint, make notes of every step you do. You'll thank yourself next time.13:31
rcpeterson joined13:31
srisw joined13:32
kulhas joined13:32
cbreak_work bash will do that for you :_13:32
OOPMan joined13:34
engla joined13:34
engla left13:35
engla joined13:35
LiamH joined13:35
kulhas how can I compile git with static link ?13:36
roger_padactor so I untracked some things doing git rm -r --cached and later I did a clean but now a bunch of files from my working tree keep dissapearing when I do commits. how do I get this back to normal and track everything?13:36
jefferai left13:38
Teo` left13:38
avinashhm left13:38
skelterjohn joined13:38
alnewkirk left13:39
avinashhm joined13:39
normanrichards left13:41
coderdad joined13:42
winterweaver left13:42
tizbac joined13:42
MrJeep left13:42
gretch joined13:43
jhoepken left13:43
nicoulaj left13:43
miloshadzic joined13:45
nicoulaj joined13:45
linusoleander joined13:47
borwar left13:48
linusoleander left13:49
voker57 left13:50
selckin what you're saying doesn't make much sense13:50
_ollie joined13:51
_ollie left13:51
kalusn left13:52
spaceonline left13:53
jankoprowski left13:55
jxpx777 joined13:55
ferd joined13:56
salty-horse joined13:56
ChanServ set mode: +v13:57
verdon left13:57
ferd hello all - I had a hopefully quick question if anyone could help out13:57
ptl left13:57
ferd i have a non-working directory that represents what should be a git branch13:58
i want to make a working directory, but preserve the changes in the code13:58
any tips as to how to do this?13:58
verdon joined14:00
guardian joined14:00
shruggar ferd: do you have a git checkout already?14:01
project2501a left14:01
ferd i have the master branch pushed to a remote repository14:01
normanrichards joined14:02
ferd so i got that far... and i can clone it, and manually move the changed files to the clone.. but iw as wondering if there was an easier way to do that14:02
emptyflask joined14:02
shruggar there are more direct methods, but they all involve you having a local clone14:03
ferd ok i can grab a local clone14:03
roger_padactor how do I re track everything I untracked?14:05
mdel joined14:05
Pewpewarrows joined14:06
Teo` joined14:06
Teo` left14:06
Teo` joined14:06
jefferai joined14:06
urbanmonk left14:06
j416 roger_padactor: remove from ignore file, add again.14:07
shruggar ferd: GIT_DIR=$new_clone/.git GIT_WORK_TREE=$the_dir git add $the_dir; will add all the files from $the_dir into the index. You can then examine the changes with git diff --cached, and commit if it's appropriate14:07
j416 roger_padactor: you can add without removing from ignore file, but it's sane to remove from the ignore file, of course, if you have such a file.14:07
SeanLazer joined14:07
Cromulent left14:08
ferd ah fantastic - thanks for your help14:08
RaceCondition_ joined14:08
RaceCondition_RaceCondition14:09
RaceCondition hey, I just lost a commot by doing a rebase -i and messing something up, but I cannot seem to find the right commit ID from git reflog... any hints?14:09
fenrig I've got the same issue with gitolite as with gitosis14:09
fatal: 'gitolite-admin' does not appear to be a git repository14:09
after completing the gl-setup14:10
teuf RaceCondition: depends on why you cannot find it exactly :) You can find HEAD sha1 before the rebase, and look at git log $SHA114:10
ben_alman left14:10
koltroll left14:10
RaceCondition teuf: find HEAD sha1? how?14:10
roger_padactor so after somethings been tracked if you add it to the ignore it still gets tracked thats why you have to remove it then add it to the ignore?14:10
jefferai left14:11
RaceCondition teuf: actually, I think I got it :)14:11
thanks!14:11
\mjd joined14:11
posciak joined14:11
Deesl left14:11
odol left14:12
cptG joined14:12
rbq joined14:12
fenrig How can this be :S I tried gitolite and gitosis and none of them seem to create a existing repository on either debian or archlinux :(14:13
shruggar what do you mean by "create a existing repository" ?14:14
fenrig, creation is generally something which happens to move something from a state of non-existence to a state of existence :)14:15
kulhas How can I compile git with static link?14:15
nevyn^ joined14:15
RaceCondition teuf: I wish there were "git show-recently-lost-commits" or smth.. would make life so much easier, especially for newcomers14:15
jefferai joined14:15
fenrig shruggar: I'm sorry i meant to create a repo :o I followed the sitaramc install gitolite guide but it doesn't work14:15
everytime i try to check in i get that fatal :(14:16
jxpx777 left14:16
bkosborne joined14:16
jxpx777 joined14:16
shruggar fenrig, hm. Are you using the latest version of git?14:16
meebey joined14:16
nicoulaj left14:16
teuf RaceCondition: maybe git lost-found does that14:17
fenrig shruggar: I'm using archlinux, so yes14:17
RaceCondition teuf: ah14:17
skelterjohn is there an easy way to sync a git project with an hg project?14:17
shyam_iyer joined14:17
RaceCondition teuf: git fsck --lost-found14:17
nicoulaj joined14:18
teuf ah fsck, I knew there was this too but I couldn't remember its name14:18
shruggar skelterjohn, yes, though I don't know where :)14:18
skelterjohn, it's not part of the main distribution yet14:18
Tasser joined14:18
RaceCondition teuf: but that just gives a huge list of commit ID's.. no commit messages or dates or nothing :)14:18
Tasser is there a command to start a git server in a repo so people can pull?14:19
skelterjohn or vice versa, is fine - i have a project hosted on googlecode/hg and github, but i don't like committing/pushing twice14:19
teuf RaceCondition: for sha1 in $(git ...); do git show $sha1; done or so14:19
guardian left14:20
nevyn^ left14:20
RaceCondition teuf: ah, cool... I thought there had to be smth like git show14:20
Titosemi1 left14:21
shruggar fenrig, just to confirm, what was the URL of the install guide you used?14:22
nothingHappens joined14:22
fenrig shruggar: http://sitaramc.github.com/gitolite/doc/1-INSTALL.html14:23
shruggar that's the one all right14:23
which method?14:23
fenrig deb/rpm method (but actually I'm using archlinux)14:24
panike joined14:24
RaceCondition weird, git show --online sha1 seems to ignore --online14:25
oneline*14:25
shruggar fenrig, are you running the commands exactly as shown, or are you substituting some of your own values for things like username?14:26
bkosborne left14:26
fenrig shruggar: unfortunatly i do :o14:27
only i created the .pub file with ssh-keygen -t rsa14:27
because it's not in the guide and i have it from the gitosis guides :)14:28
shruggar that should be fine14:28
fenrig shruggar: I really don't understand :(14:28
shruggar does ~gitolite/repositories exist on the remote server?14:29
srisw_ joined14:29
shruggar (making wild guesses here, since I didn't use the package method, I really have no idea where it tries to put it)14:29
fenrig yes it does :)14:29
shruggar is gitolite-admin.git in there?14:30
srisw left14:30
srisw_srisw14:30
fenrig yes14:30
shruggar and you are doing git clone gitolite@...., not git clone yourname@.... right?14:31
fenrig shruggar: Instead i named gitolite git :) because well gitolite is too long XD14:31
so yes14:31
Anubisss joined14:32
fenrig git clone git@localhost:gitolite-admin.git (from another user in another folder)14:32
jason237 left14:32
shruggar what about: git clone git@localhost:gitolite-admin ?14:32
fenrig shruggar: also not working14:32
Nazcafan joined14:33
jason237 joined14:33
Nazcafan hello14:33
I made a commit in a detached state, how do I create a branch that includes this commit?14:33
tango_ git checkout -b branchname14:34
Gunni left14:35
jeeger Hm, where is the git alternative to svn $Revision$ tags?14:35
ChillawayChillax14:36
shruggar jeeger: the keyword you're looking for is "smudge"14:36
iirc14:36
jeeger australian punk band?14:36
shruggar man git-attributes14:37
man gitattributes14:37
jast the 'gitattributes' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/gitattributes.html [automatic message]14:37
shruggar there it is14:37
look for smudge in there14:37
Lord_Khelben joined14:38
Anti-X joined14:39
montylounge joined14:40
sivy_ joined14:40
fenrig shruggar: I don't understand why it's not working :s I'm googling it but none of the solutions work14:40
angelsl joined14:41
sako joined14:41
jeeger shruggar: ah, thanks.14:41
opalka left14:41
ch077179 joined14:41
project2501a joined14:41
goshawk left14:41
tizzo joined14:41
shruggar fenrig: what happens when you try to just ssh git@localhost ?14:42
torvalamo left14:42
fenrig I log in as git using ssh, works perfect :)14:42
shruggar fenrig: there's your problem, I bet :)14:43
fenrig shruggar: why?14:43
Nazcafan tango_, thank you14:43
shruggar fenrig: what's in ~git/.ssh/authorized_keys ?14:43
fjb4_ joined14:44
airborn left14:44
fenrig #gitolite start14:44
...14:45
#gitolite end14:45
project2501a left14:45
shruggar just one key, then? And it has a command=" " part at the beginning?14:45
fenrig wow :o14:45
shruggar: yes but14:45
shruggar: it ends in "stage@archvm"14:46
instead of14:46
gip@archvm14:46
so i tried doing14:46
git clone stage@localhost:gitosis-admin14:46
\mjd left14:46
srisw left14:46
fenrig and it warns me about cloning a empty repo :)14:46
leo2007 joined14:46
fjb4 left14:47
sonnym left14:47
fenrig ls14:47
shruggar there you go, then. Looks like the install guide needs to be updated14:47
fenrig that's sooo strange14:47
well i think its because of ssh keygen14:47
maybe i did it in stage instead of git :o but nobody said it was important :s14:48
shruggar I don't think it is14:48
what did you name it?14:48
fenrig shruggar: named what?14:48
shruggar the temporary key14:48
marc_kdab left14:48
fenrig id_rsa.pub or something like that14:49
kukks joined14:49
Nitecon left14:49
shruggar hm, okay, no idea :/14:50
I really don't know how it would even detect that stage@ was involved14:50
Error404NotFound left14:50
shruggar so my only guess is "something weird about the .deb"14:50
project2501a joined14:51
simplechat left14:51
fenrig shruggar: I used a packagebuild :s14:51
Gunni joined14:52
fenrig shruggar: how do i change stage to git :o properly cause stage is not intended for this14:52
avena joined14:52
siganderson joined14:52
dfr|mac joined14:52
zombor joined14:52
Number5_ joined14:52
siganderson how do I when I want to select a particular commit in a project tree?14:52
atrofast joined14:53
teuf siganderson: what do you mean exactly by "select" ?14:53
fjb4_ left14:53
shruggar fenrig: clone as stage, then rename the keyfile (in the keyfiles, or whatever it's called, dir) from stage.pub to git.pub, then commit and push14:54
fenrig shruggar: but it's a empty repo :s14:54
atrofast Heya, I want to manage several drupal sites inside a drupal core repository and the problem I have is I don't want to HAVE to download all sites when I clone the core repository (since these sites add up to GBs of data)... Does anyone have a best approach to this?14:54
shruggar fenrig, ah, right...14:54
sako left14:55
siganderson teuf, think when you do git-bisect, every step it selects a particular commit, I want to do that having the commit's number14:55
Number5_ Hello guys, I want to compare my local master with remote master and after googling I did this: git diff master origin/master . There should be no difference in my situation, so is it normal the output will be an empty line? I see no message like "no difference detected" or something. Is this normal?14:55
teuf siganderson: you want that the content of your working directory matches what was in commit $SHA1 ?14:55
siganderson: if so, git checkout $SHA1 will do that14:55
shruggar fenrig, okay, you'll probably need to re-install, 'cause it looks like it didn't finish installing. First, copy the existing ~git/.ssh/authorized_key 's key to under the #end line14:55
siganderson teuf, exactly :)14:55
teuf, thank you14:55
teuf siganderson: when you do that, you are in what is called "detached head" state, meaning it's not recommended that you do commits or any changes14:56
shruggar fenrig, then re-install. I expect that may get you started14:56
fenrig: come to think of it, I'm completely wrong and I have no clue how to help you :)14:56
siganderson teuf, yes, I'm doing that only for testing purposes14:56
angelsl left14:57
mattp1 joined14:57
teuf siganderson: ok, have a nice testing session then14:57
:)14:57
mattp_ left14:57
fenrig shruggar: I redid the whole thing :s14:57
shruggar fenrig, fair enough14:58
fenrig shruggar: and the situation is still the same14:58
ferd how do I diff unstaged files...14:58
mattp1 left14:58
ferd files that git status says are modified14:58
shruggar fenrig, all I can recommend at this point is to try another installation method14:58
bhosie joined14:58
mattp1 joined14:58
shruggar ferd, "git diff" ?14:58
ferd it only works when i stage a file14:58
shruggar ferd: I think you might be using the word "stage" wrong, in that case14:59
ferd so, git status says a file is modified but i don't want to stage it14:59
probably14:59
haha14:59
khmarbaise left14:59
rcpeterson left14:59
rcpeterson joined14:59
siganderson teuf, thank you again14:59
mattp1 left14:59
ferd "# Changed but not updated:14:59
# (use "git add <file>..." to update what will be committed)"14:59
siganderson left14:59
ferd files under that , i cannot git diff14:59
mattp_ joined15:00
xiangfu joined15:00
project2501a left15:00
shruggar "git diff" gives differences between the index (staging area) and the work-tree. "git diff --cached" shows differences between the index and HEAD15:00
Utkarsh_ joined15:00
fenrig shruggar: I'm going to reinstall my vm and try again from scratch :)15:00
yaotti_awyaotti15:00
shruggar fenrig, ouch... good luck?15:00
Number5_ left15:01
Error404NotFound joined15:01
fenrig shruggar: It's a good thing I'm doing it on a VirtualMachine first otherwise i'd be real pissed XD15:01
shruggar I just read the docs until I thought I understood things, then did it all on the live server... then sitaram held my hand for a while because I misread various things :/15:02
Utkarsh left15:02
shruggar lesson is: install when sitaram isn't out drinking ;)15:02
cbreak helped too, I'll mention. credit where it's due.15:03
TheMoonMaster joined15:03
airborn joined15:03
guardian joined15:04
bauruine left15:04
verdon left15:05
shyam_iyer left15:05
fenrig shruggar: hahahaha :D i'm so happy I have a VM for testing all of this cause my server is almost perfectly tuned to suit my needs :d15:05
shruggar: though someone should make a better guide XD15:06
feindbild joined15:06
shruggar I'll guess it's a problem with the package15:06
feindbild hi ^^15:06
RaceCondition left15:06
khmarbaise joined15:07
fenrig shruggar: u certeanly don't know pkgbuild from archlinux. It does the manual installation in a fakeroot for me :) and then I just install the package (so actually it extracts itself to /"15:07
OOPMan left15:08
verdon joined15:08
fenrig shruggar: you should check it out, it's a real fine and cool system helps you compile everything in a package :d15:08
feindbild is it possible to rewrite the history to be linear? some parameter to rebase so it doesn't stop all the time because of merge-conflicts?15:09
psundaram joined15:09
shruggar feindbild, when rebase complains of merge conflicts, it means "merge" in the low-level sense15:10
psundaram left15:10
fowlduck joined15:10
shiba_yu36 left15:10
nevyn^ joined15:11
feindbild shruggar: yes, but shouldn't it be able to solve it by itself?15:11
xiong left15:11
shruggar feindbild, conflicts are by definition the parts of a merge which git was unable to solve by itself :)15:11
feindbild shruggar: like just doing a diff to the previous commit and applying it15:12
shruggar feindbild, that's pretty much what it does15:12
webPragmatist joined15:13
webPragmatist left15:13
webPragmatist joined15:13
benishor_ left15:13
spaceonline joined15:14
shruggar feindbild, sometimes it can be helpful to not rebase everything at the same time. For example, if you're rebasing to a point 20 commits ahead of where you were, start by rebasing to a point just 10 commits, or 5, or 1, and repeating until finished. There are sometimes fewer conflicts, and they generally are easier to resolve or at least justify15:14
feindbild shruggar: I'm trying to rebase more like 4.5k commits >.<15:15
freimatz left15:15
shruggar feindbild, sorry, I meant "onto a point..." not, "to a point"15:16
Drakevr left15:16
Drakevr joined15:16
patrikf feindbild: why?15:17
shruggar feindbild, may I ask what your situation is? What has caused this?15:17
EricInBNE left15:17
yaottiyaotti_aw15:19
__name__ joined15:19
__name__ left15:19
__name__ joined15:19
yaotti_awyaotti15:20
schweizer is it possible to restore deleted files in git?15:21
jeeger schweizer: just go back to a previous version?15:21
schweizer hm how do i even delete files in git? there is no git delete15:22
yaottiyaotti_aw15:22
yaotti_awyaotti15:22
schweizer ah git rm15:22
shruggar schweizer, of course. That's the primary thing version control systems do :)15:22
schweizer well but its not always very easy15:22
sako joined15:23
_ollie1 joined15:23
ben_alman joined15:23
yaottiyaotti_aw15:24
yaotti_awyaotti15:24
sonnym joined15:24
_ollie1 left15:25
shruggar schweizer, well how you do it depends on exactly what you want to do: make it so the delete never happened? make the file NOW be what it was THEN? re-create the file without tracking the file? Roll everything back to what it was when the file last exists? Revert the change which made the file go away, while keeping the log of it having gone away?15:26
Snake_Byte joined15:27
guardian left15:27
shruggar that's at least four different things one might mean by "restore deleted files"15:27
guardian joined15:27
shruggar hm, no, it is five. I was thinking two of those were the same, but it only is in a trivial case15:28
panike left15:28
j416 hmm15:29
did he say "restore deleted files"?15:29
oh15:29
yes he did.15:29
j416 hides15:29
ExtraSpice left15:30
ceej joined15:31
yaottiyaotti_aw15:31
avena left15:31
sivy_ left15:32
teweWork anyone? http://stackoverflow.com/q/4828629/29978815:32
sivy joined15:33
doener teweWork: it should get each updated ref, not each commit15:35
teweWork: maybe you're just using some thing like "rev-list <new-state>" which walks the whole history because you don't limit the set?15:36
fenrig shruggar: same problem allover again :(15:36
shruggar fenrig, congratulations? :)15:36
rbq left15:37
fenrig shruggar: why?15:37
shruggar: and now "git clone stage@..." doesn't work anymore (i consider that a good thing)15:38
shruggar fenrig, check ~git/.ssh/authorized_keys15:38
fenrig shruggar: now i get "fatal: 'gitolite-admin' does not appear to be a git repository"15:38
shruggar: it has a command and stuff so it works :o15:39
shruggar: I mean the authorized_keys is fine15:40
offby1 joined15:40
vmiklos left15:40
smolyn joined15:41
ptl joined15:42
psankar joined15:42
spearce joined15:44
jeeger left15:44
goshawk joined15:45
alvaro_o joined15:45
saschpe left15:46
jaql left15:46
avinashhm left15:46
fenrig it really makes no sense :(15:47
feindbild shruggar: sorry, was called off to a meeting ^^ boss wants SVN mirror, I wasn't here for 3 months so now history is a mess of merge-commits ^^15:48
_pingu left15:49
shruggar feindbild, stall him until after this years GSoC? :)15:49
Yuffster joined15:49
Yuffster left15:49
feindbild shruggar: why? xD15:49
Yuffster joined15:50
ROBOTARMY joined15:51
shruggar feindbild, because I remain hopeful that git-remote-svn will be finished someday :)15:51
feindbild shruggar: :o)15:51
shyam_iyer joined15:51
spike joined15:51
tcurdt joined15:51
sivysivy_biab15:52
spike hi, I've done a git fetch and it pulled down some stuff, but git log doesn't show any of the new commits. I thought the point of fetch was that you could preview the changes before merging them. not sure what I'm doing wrong15:52
sako left15:53
jxpx777 left15:53
teuf spike: they are in origin/branchname, have you looked there ?15:53
spike: git log origin/master for example15:53
spaceonline left15:53
shruggar spike: you probably need to specify what you want to view a log of. "git log" defaults to the current HEAD, whereas "fetch" defaults to refs/remotes/<someremote>/<somebranch>15:53
spike well, I as doing --all, I thought that looked there too15:53
jxpx777 joined15:53
spike s/as/was/15:53
err_ok joined15:54
shruggar hm, okay, I think that /is/ in fact the case15:54
Error404NotFound left15:54
shruggar I think --all sorts by date, though15:54
scelis joined15:54
gretch left15:55
guardian left15:55
teweWork doener: i think u r right, my fault15:55
spike I've only got 5 commits, was just experimenting. the thing is if I throw away the clone and fetch again it says it's got 3 new objects15:55
so it's clearly fetching something15:55
ChUrP joined15:55
spike and show up in log15:56
steph021 joined15:56
spike and if I do a git pull the changesets come down as expected15:56
chrisshattuck joined15:56
cbz git fetch does a git merge doesn't it?15:57
selckin other way around15:58
spike no, that's what pull does, altho that's what I thought15:58
cbz okay15:58
spike s/altho/at least/15:58
selckin git pull = fetch+merge (or rebase)15:58
beatak joined15:58
_psychic_ joined15:59
fenrig could it be that i dont have a15:59
~/.ssh/config?15:59
shruggar fenrig, I doubt it16:01
fenrig shruggar: Well i don't have a /home/git/.ssh/config16:01
Error404NotFound joined16:01
Error404NotFound left16:01
Error404NotFound joined16:01
guardian joined16:01
bhosie left16:02
mishok13 left16:02
aah left16:02
nona joined16:02
nothingHappens left16:03
necromancer joined16:03
juanjoc left16:03
necromancer left16:03
rayners joined16:03
pgib joined16:03
nicoulaj_ joined16:04
Mannequin joined16:04
krfs joined16:04
krfs left16:05
fenrig shruggar: I don't have it :s is this normal16:05
xiong joined16:06
tizzo has anyone here used git as a backup tool?16:06
alberto56 joined16:06
shruggar tizzo, yes, but they all stopped shortly afterwards because git is not a backup tool16:06
SethRobertson I have heard people doing such, but it isn't precisely a good idea.16:07
tizzo shruggar & SethRobertson : why not?16:07
SethRobertson Isn't there a new linux distribution which puts all files under version control?16:07
aalex_laptop joined16:07
tizzo SethRobertson: idk, is there?16:07
spaceonline joined16:07
shruggar tizzo, generally one doesn't want the entire history of all backups ever in every place they need to back up16:07
nicoulaj left16:07
tizzo shruggar: fair point, so we'd need to add backup rotation at some point16:08
my use case is a web app that basically always adds files16:08
they virtually never get deleted16:08
sitaram fenrig: almost everything you have would appear to be "using the wrong key or allowing the wrong key to be used"16:08
SethRobertson Foresight Linux16:08
agile joined16:08
sitaram fenrig: see http://sitaramc.github.com/gitolite/doc/ssh-troubleshooting.html16:09
fenrig sitaram: Can you please explain it to me XD :)16:09
sitaram: I'm looking at it right now :o16:09
tizzo so with git there would be very little unnecessary objects16:09
shruggar there we go, someone competent is here. I will now leave :)16:09
sitaram fenrig: it also really, *really* helps if you read http://sitaramc.github.com/gitolite/doc/gitolite-and-ssh.html16:09
fowlduck left16:09
marc_kdab joined16:10
tizzo SethRobertson & shruggar : so the only catch with git-as-backup is backup rotation?16:10
guardian left16:10
SethRobertson Git also isn't happy with large binary files16:10
tizzo oh I didn't know that, unhappy how?16:10
SethRobertson You need at least as much ram as the size of your largest file, possibly a multiple of that size (it used to be like 2.5x, but that may have changed recently)16:11
shruggar tizzo: No, in the sense that I wouldn't call "backup rotation" a sane solution when it means "throw out all your old backups"16:11
warbee joined16:11
Schmallon joined16:11
tizzo shruggar: huh?16:11
SethRobertson shruggar: In theory you could squash commits together to reduce the size and eliminate old backups16:12
Or outright delete old commits. Still, it requires a lot of infrastructure around git, and is thus no longer pure git16:12
tizzo SethRobertson: wouldn't even need to do that, rebase can just drop commits, right?16:12
shruggar tizzo: git is not a backup tool. Don't use it as a backup tool. Don't even think about using it as a backup tool. Use a /backup tool/ as a backup tool16:12
ROBOTARMY left16:13
bhosie joined16:13
SethRobertson Dropping commits doesn't necessarily give you a consistent filesystem. Squashing does. If you add a file and later delete it and squash those commits, the file goes away.16:13
_ollie joined16:14
tizzo shruggar: I was using a /backup tool/ as a backup tool and I was getting frustrated with it, threw together a 2 bash scripts in about 10 minutes that give me automated git commits and fetches and it's working great16:14
but I figured there has to be a reason people aren't doing this16:14
SethRobertson: that makes sense16:14
bremner tizzo: large binary files16:14
tizzo bremner: is there any doco on the limitations with this?16:15
SethRobertson No, but see disucssions on the mailing list16:15
tizzo most of the files in my use case are actually pretty small but most are binary (web uploaded images)16:15
cbreak_work left16:15
tizzo SethRobertson: ok, thanks16:15
datagrok joined16:17
teweWork left16:17
nadavoid joined16:18
spike left16:18
datagrok When using git log --graph, is there a way to force a particular branch to the left-hand side?16:18
jxpx777 left16:18
ddkilzer joined16:19
fenrig sitaram: I read the 2 documents, and well it didn't help me fix it :(16:19
teweWork joined16:19
Nazcafan left16:19
datagrok Oops, I meant: when using git log --graph --all, is there a way to force a particular branch to the left-hand side?16:20
sitaram fenrig: alright lets get onto /query to avoid boring everyone with YASP (yet another ssh problem)16:20
alberto56 left16:20
shruggar fenrig: as you're using a VM, trying another install method seems like a very sane method of eliminating (or controlling) variables16:20
fenrig sitaram: http://sites.google.com/site/senawario/home/gitolite-tutorial ; i compared what i did with this tut and there is almost no diff, except that I used a pkgbuild from archlinux :o16:20
join /query16:21
napster_123 left16:21
fenrig eeeuhm what is /query? (long time not used irc anymore) :p16:21
keyvan- joined16:22
sitaram fenrig: archlinux... I have no idea what they would do. Please start with one of my versions so I don't have to guess if/what changes they have16:22
Yuffster left16:23
dfr|mac left16:23
kuhrt joined16:24
webPragmatist left16:24
jdav_gone left16:25
webPragmatist joined16:26
webPragmatist left16:26
webPragmatist joined16:26
guardian joined16:27
alberto56 joined16:27
nona left16:27
sivy_biabsivy16:27
datagrok left16:28
keyvan- left16:29
posciak left16:29
_ollie left16:29
dest left16:30
skoop left16:30
jdav_gone joined16:30
Heimidal joined16:32
romv i'm doing some repo maintenance and I'm getting rid of some remote branches that we don't use anymore using git push origin :branch_to_delete ... is a git fetch sufficient for the other developer to sync up? if they have any local branches tracking that deleted remote branch will it recreate it if they accidentally push on that branch?16:33
jxpx777 joined16:33
j416 left16:34
tizzo left16:35
kulhas left16:35
alberto56 left16:36
nicoulaj_ left16:36
Weust`Weust`afk16:37
nicoulaj_ joined16:37
dreiss joined16:37
guardian left16:37
warbee left16:37
Cro_Crx left16:40
tjholowaychuk joined16:40
rorist joined16:42
alberto56 joined16:42
bcardarella joined16:43
keyvan- joined16:44
homa_rano is there a difference between 'git reset HEAD' and 'git reset HEAD .' ?16:44
alberto56 left16:45
rorist left16:45
teweWork left16:46
marc_kdab left16:46
envi left16:47
hipstertrash_ joined16:47
hipstertrash left16:47
hipstertrash_hipstertrash16:47
jcdny joined16:48
feindbild shruggar: /quit16:49
oops16:49
yrlnry homa_rano: 'git reset HEAD .' resets only the paths in the current directory.16:49
feindbild ^^16:49
feindbild left16:49
yrlnry homa_rano: 'git reset HEAD' resets everything16:49
teweWork joined16:49
homa_rano yrlnry: ah, so . was a path after all; thanks16:49
yrlnry YW16:50
kpreid left16:50
sako joined16:51
nevyn^ left16:51
Schmallon left16:51
stliu joined16:52
stliu left16:52
stliu joined16:52
Zhonghua left16:52
guardian joined16:53
yhager left16:54
steph021 left16:55
atrofast is git submodules the only way to include subdirectories that are not always downloaded when you clone a git repository16:55
cubuanic joined16:56
MattDiPasquale joined16:56
Dieterbe joined16:56
nicoulaj_ left16:56
Dieterbe Hi, i know how to squash commits with git rebase -i, but how do i squash a commit with the very first commit?16:57
when i do git rebase -i $first_commit, it starts with the commit after $first_commit16:57
and if i move the squash line to the top, i get "Cannot 'squash' without a previous commit"16:58
joel_falcou left16:58
aah joined16:58
joel_falcou joined16:58
Dieterbe and if i do : git rebase -i $first_commit^ i get:16:58
fatal: Needed a single revision16:58
Invalid base16:58
keyvan- left16:58
offby1 Dieterbe: I think you've got the right idea; you just need to "spell" that first_commit^ properly17:00
I don't know why what you did didn't work17:00
just run 'git log' or something, get the actual SHA1 for the parent of first_commit, and pass that to "git rebase -i"17:00
oriba joined17:00
_psychic__ joined17:00
Dieterbe the parent of the first commit?17:01
afaict, it has no parent17:01
xiangfu left17:01
bronson_ joined17:01
teweWork left17:02
Angel-SL joined17:02
pgib left17:03
bcardarella left17:03
nevyn^ joined17:03
_psychic_ left17:04
_psychic___psychic_17:04
jrmuizel joined17:04
offby1 oh.17:04
smolyn left17:04
offby1 well, then you can't squash it into its parent, then can you?17:04
that's what squashing is -- combining a commit with its parent.17:04
snearch joined17:04
Dieterbe i want to squash into the first commit, not the parent of the first commit17:04
offby1 oh17:04
ah17:04
xiong left17:05
offby1 I dimly recall that this is a sort-of misfeature, or bug, in git; and that there's some trick to get around it.17:05
dotCOMmie left17:05
offby1 Unfortunately, my brain is too small to contain the details of the trick17:05
Angel-SL left17:05
steph021 joined17:05
TheFox21 joined17:06
steph021 left17:06
Dieterbe aha http://stackoverflow.com/questions/598672/git-how-to-squash-the-first-two-commits17:06
Angel-SL joined17:06
Angel-SL left17:06
angelsl joined17:07
serard left17:08
offby1 yeah!17:08
alvarezp joined17:08
bronson_ left17:08
offby1 heh ... "OCD heaven". Who gave that guy the right to spy on me? :-)17:08
Dieterbe anyway, this will be for Monday, when i have a clear head, weekend is starting!17:08
MattDiPasquale left17:09
dreiss left17:09
ramonmaruko left17:10
patmaddox joined17:11
tstclair left17:11
ISF joined17:12
montylounge left17:12
Utkarsh_ left17:13
airborn left17:14
hyperair joined17:14
Utkarsh_ joined17:15
hipstertrash left17:15
amerine joined17:15
steph021 joined17:16
siyb how would i create aliases for branches (symbolic-ref?) so that i have an alias like experimental, which always points at the current experimental branch?17:16
radarek left17:16
ISF_ joined17:17
phishy_ joined17:18
csanz joined17:18
Dieterbe is a shell variable ok?17:19
noahcampbell joined17:19
MattDiPasquale joined17:19
patrikf siyb: uh, you can do that, but you can also use a branching model where your experimental branch is always named experimental ;-)17:19
Dieterbe or that, of course :P17:19
keyvan- joined17:19
ISF left17:20
ptl left17:20
patrikf or your local branch is named experimental and points to some other remote branch17:20
s/points to/tracks/17:20
joel_falcou left17:22
keyvan- left17:22
siyb well, i was planning on creating an experimental branch at origin, which always points to the branch, which is currently used to develop in17:22
bhosie left17:22
SimonP86 joined17:23
giallu left17:23
siyb so that origin/experimental tracks origin/0.1 for instance17:23
krfs_ joined17:23
krfs_ left17:23
bhosie joined17:23
siyb when 0.1 becomes sort of stable, experimental should track 0.2 and so on17:23
tro left17:24
angelsl left17:24
offby1 I'd just advance the branch by hand17:24
patrikf why don't you develop experimental, and branch off 0.1 when it's stable?17:24
fenrig left17:24
patrikf develop on*17:24
krosenvold left17:24
patrikf or rather, when you start development on 0.217:24
angelsl joined17:24
patrikf or branch to "stable" and only tag it as 0.117:24
Dieterbe left17:26
Agos left17:26
siyb well, i though git is all about personalizing your workflow (please don't missunderstand me, i am not trying to be cocky!). so there isn't a way to do what i am thinking about, right?17:26
Badiss left17:27
nothingmuch left17:27
shruggar siyb, git is flexible, but the one which is specifically about personalizing your workflow is bazaar. Git is only about allowing you to have /sane/ workflows ;)17:27
patrikf actually, i think symbolic-ref will work out the way you described17:28
ferd left17:28
patrikf our approach here is usually to solve problems instead of merely answering questions17:28
nevyn^ left17:28
ptrb left17:29
coderdad left17:29
siyb shruggar: so my workflow isn't sane? :p17:29
marc_kdab joined17:29
ptrb joined17:29
shruggar siyb, possibly. I wasn't looking when you described it :)17:30
nona joined17:30
hipstertrash joined17:30
marc_kdab left17:31
shruggar siyb, but I agree with others: if you always want to develop in a branch called "experimental", call it that :)17:31
siyb well the idea is to have an experimental branch on which users can stay and a versionized branch. experimental always reflect the newest version, while the versionized branch becomes stable eventually17:31
kpreid joined17:31
GaVrA joined17:32
tro joined17:32
j0rd joined17:32
leavittx joined17:32
siyb ok thanks for your help, going to figure that out now17:33
tizzo joined17:33
ben225 joined17:33
nothingmuch joined17:34
salty-horse left17:35
enherit joined17:35
patrikf siyb: you are aware that you can create, delete, rename, modify branches at any time, right?17:35
shruggar siyb, your workflow sounds common, but your proposed implementation is not how to do things17:36
patrikf siyb: and also that git does not care what branch a commit is created on?17:36
pheaver joined17:37
Cro_Crx joined17:37
bronson_ joined17:37
bronson_ left17:37
project2501a joined17:38
Schmallon joined17:39
bigpeted left17:39
Tasser left17:39
steph021 left17:39
bronson_ joined17:40
jotik left17:41
shruggar left17:42
steph021 joined17:42
angelsl left17:44
sivy left17:44
hobodave joined17:44
ilteris_ left17:45
siyb shachaf: what do you propose then?17:45
ilteris_ joined17:45
siyb sorry ... wrong nick17:45
bronson_ left17:46
siyb patrikf: i do, but i am not sure how that could be helping me now17:46
i know git manages objects which may have relations between them17:46
PigeonFriend left17:47
marc_kdab joined17:47
alvarezp left17:47
nona left17:48
linusoleander joined17:48
siyb patrikf: may i hear your proposal. i am trying to implement a system where there is an experimental branch pointing at a versionized branch, both branches coexist17:48
kpreid left17:49
siyb i mean i could create a hook of some kind which syncs two branches on push or something, but i am not sure17:49
patrikf siyb: i just don't get why you want that versionized branch17:50
you can create it later on, when "experimental" moves on to be something else17:50
justin-george joined17:50
smolyn joined17:50
siyb if you are running debiann, you got the lenny -> stable mapping. if your sourcefile contains lenny instead of stable, you will remain on lenny, even though stable might have advanced17:51
nona joined17:51
lresende joined17:52
siyb if you are on stable, you will always be on the stable release, no matter which codename this release has17:52
aalex_laptop left17:52
sm joined17:52
dfr|mac joined17:53
patrikf yes. so?17:53
sm g'day all. For a normal project tag, eg of a release, should I use git tag -a ? or just git tag ("lightweight tag") ?17:53
siyb patrikf: well that is the system i am trying to implement ...17:54
that mean that the versionized branch and the experimental branch need to coexist and be linked in some manner, right?17:55
patrikf siyb: you are doing development for debian?17:55
skelterjohn left17:56
rorist joined17:56
siyb patrikf: no. i just want to apply a similar system17:57
TheMoonMaster left17:57
patrikf that's what i thought... but a git repository is not a debian package archive...17:57
leavittx left17:57
patrikf i still don't know17:58
sigh17:58
coderdad joined17:58
patrikf i still don't know *why* you want that experimental refs, but if you insist, by all means go ahead and implement it with symbolic-ref17:59
leavittx joined17:59
LouisJB left17:59
oriba left18:00
marc_kdab left18:01
siyb patrikf: ok, fair enough. thanks for your help/time :)18:01
yhager joined18:01
saschpe joined18:02
psankar left18:02
miloshadzic left18:02
_psychic_ left18:03
_psychic_ joined18:03
Weust`afkWeust`18:04
Weust`Weust`afk18:05
nothingHappens joined18:05
nothingHappens left18:06
shruggar joined18:06
nothingHappens joined18:09
jarrod_ joined18:09
nothingHappens left18:11
sm after doing tag -a, how do/will I see the tag in git log ?18:11
Lord_Khelben "git tag" will show it18:12
nothingHappens joined18:12
rayners left18:13
henrikbjorn joined18:13
WebDragon joined18:14
sm thanks. Is there an easy way to list tags in time sequence with commits at the command line ?18:14
jmil left18:15
cbreak sm: git log --decorate18:15
diphthong joined18:15
banisterfiend left18:15
bluenovember left18:15
KipMacy left18:15
project2501a left18:15
sm thanks18:16
MikeyB left18:16
MikhX joined18:16
sm next q: I'm moving a github wiki repo to a full github repo. I cloned the wiki, did git remote rm origin and git remote add origin ...stuff recommended by github for the new repo...18:17
dreiss joined18:17
sm I was able to push to the new repo by name, but what do I need to do to make simple git push/git pull work ?18:18
sivy joined18:18
leavittx left18:18
sm the error implies I need to configure/adjust the default refspec or something18:18
elmob joined18:18
sm look like I can configure the push.default preference. I want to be sure to configure this repo in a clean "default" way, so I don't cause confusion down the road18:20
Lord_Khelben if you do once "git push origin master", after that git push doesn't work ?18:20
sm Lord_Khelben: right. https://gist.github.com/80068218:20
kenichi joined18:21
yotaff left18:21
Lord_Khelben i don't get this message, but you can choose a behavior you want18:23
for example git config push.default matching18:23
wait to hear from some more knowledgeable though18:23
sm thanks, I've done that and it has fixed push18:24
git pull gives different advice18:24
ie, configure four variables in a configuration file - https://gist.github.com/80068618:24
tizzo left18:25
cbreak sm: do a git push -u with the correct settings18:25
darkredandyellow left18:26
cbreak that will set the upstream for every local ref you push (well, branches)18:26
tizzo joined18:26
sm my git 1.6.3.3 doesn't support that18:26
xiong joined18:27
sm hopefully it's equivalent to the git config I just did18:27
marc_kdab joined18:27
j416 joined18:27
sm cbreak: would that take care of pull as well ?18:27
volano joined18:28
leavittx joined18:28
cbreak sm: try :D18:28
volano left18:28
cbreak get a new git18:28
1.7 is ancient18:28
vortec joined18:28
vortec hi there18:28
ymasory left18:28
SIGBUS joined18:28
jxpx777 left18:29
vortec does "git show" have a functionality to show all commits in all branches?18:29
Lord_Khelben sm if you do "git config branch.master.remote origin" and "git config branch.master.merge refs/heads/master" does pull work then ?18:30
sm Lord_Khelben: yes, that takes care of it18:30
phew. thank the lord18:30
ksmith joined18:31
ksmith hey guys18:31
cbreak vortec: no, git show only shows one thing18:31
vortec: you want git log18:31
just use git log --all18:31
throw in more options if you want more stuff18:31
FylGood joined18:31
ksmith is there some way to look at all the configuration options I have set?18:31
sm next q. After pushing, how come I don't see the tag that git tag lists locally, at https://github.com/simonmichael/ledger-website ?18:31
ldh joined18:32
cbreak ksmith: git config -l18:32
sm: pushing doesn't push tags by default18:32
use git push --tags18:32
ksmith cbreak: thanks a lot :)18:32
just getting started with git ...18:32
sm I see, thanks18:32
LouisJB joined18:33
vortec cbreak: thank you very much :)18:34
tcurdt left18:34
sm it's strange to me that tags aren't shown prominently along with changes, eg at https://github.com/simonmichael/ledger-website/commits . WRT to the github ui, I guess I need to think of them as extra branches. But it's odd not to see when they happened18:35
sraue left18:35
cbreak sm: maybe it only shows annotated tags18:36
sm I think that's what I made.. git tag -a -m 'some message' 'some-tag-name'18:37
jonata__ joined18:37
jonata__ left18:38
cbreak hmm.18:38
tatsuya left18:38
benishor_ joined18:38
malfy left18:38
thiago_home joined18:38
marc_kdab left18:39
jga joined18:39
kipras joined18:40
marc_kdab joined18:40
resmo_ left18:40
nothingHappens left18:41
malfy joined18:41
nothingHappens joined18:42
romv i'm doing some repo maintenance and I'm getting rid of some remote branches that we don't use anymore using git push origin :branch_to_delete ... is a git fetch sufficient for the other developer to sync up? if they have any local branches tracking that deleted remote branch will it recreate it if they accidentally push on that branch?18:42
sraue joined18:42
cbreak romv: git remote prune18:43
romv aah18:43
cbreak that will not delete local branches18:43
but it will delete local remote tracking branches18:43
romv which you can't push right?18:43
which means you can't push anymore*18:44
booginga joined18:44
cbreak sure you can push18:44
ksmith what do you guys recommend to get started with git?18:44
cbreak but you have to do it intentionally18:44
default pushing only pushes matching, and if you delete the remote, there won't be anything that matches18:45
Nugget ksmith: an open terminal window and a few hours to experiment18:45
romv cbreak: right18:45
cbreak: perfect ... thanks18:45
cbreak ksmith: http://eagain.net/articles/git-for-computer-scientists/18:45
FylGood left18:45
ksmith Nugget: hehehe, I'll need some info on what things are, first. I don't really get the whole concept of git right now.18:45
dr_win left18:46
tvw left18:46
cptG left18:46
Nugget ksmith: I've been using it for years and I don't get the whole concept either. You're in good company. :)18:46
aah left18:47
MattDiPasquale left18:47
miloshadzic joined18:48
dr_win joined18:49
thiago_home left18:50
radarek joined18:50
cbreak you two, read that page and you will get the concept18:50
ksmith: man gitglossary also helps18:51
jast ksmith: the 'gitglossary' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/gitglossary.html [automatic message]18:51
giallu joined18:51
thiago_home joined18:51
ksmith nugget: well, to be honest, I'm just starting software development. I'd like to use some kind of svm for my private projects and hope git is easy enough to learn.18:52
Nugget Does anyone have any idea why adding a pre-receive hook to a repo causes it to hang on pushes about half the time? It just sticks on git-receive-pack and never finishes18:53
but not every time, just some of the time18:53
theoros joined18:53
theoros left18:53
theoros joined18:53
bauruine joined18:54
ksmith cbreak: thanks for the links :)18:54
Nugget it doesn't appear to ever actually run the hook, but the hook being there induces the hang18:54
sdwrage joined18:54
Nugget git version 1.7.3.518:54
timepilot joined18:55
sm ksmith: git is not the easiest, but it's certainly one of the useful ones to know18:56
ki__ joined18:56
petrux left18:56
j416 sm: now that is entirely subjective18:56
cbreak I'd say git is easier than svn18:56
j416 personally I think git is the easiest18:56
Lord_Khelben nowadays it isn't so difficult to use18:56
cbreak no need to set up a server18:56
Lord_Khelben very easy i would say18:56
cbreak you can rewrite history if you want18:56
you can easily communicate with others18:57
you can commit as much as you want, branch, merge, tag18:57
ksmith sm: I think they're moving to git where I work as well, so it'll come in handy when they do.18:57
j416 ksmith: lucky bastard.18:57
:P18:57
ldh be glad you're not still stuck in CVS land18:58
j416 SVN land isn't very comfortable either18:58
ksmith ldh: well, I think it's kinda funny. they were using cvs. then one guy came along: we can keep all the data inside our git repository without any trouble.18:59
TheMoonMaster joined18:59
j416 what, they checked in the git repo into cvs?19:00
lol19:00
ldh ksmith: i have to use it because we've got a lot of build/deployment process built around CVS. but i still use git in stealth mode as my personal vcs19:00
disappearedng left19:01
shyam_iyer left19:01
Nugget When I migrated us from cvs to git it resulted in a profound increase in developer activity and productivity19:01
ksmith j416: I think it's the other way around.19:01
nona left19:04
rorist left19:04
spaceonline left19:05
divVerent Nugget: when I migrated from cvs to dumping random stuff at random places in the file system, you know, directories named foo_old, foo_really_old, foo_older, foo_oldest, foo_older_than_oldest, I also got a big increase19:05
just kidding, just saying, this comparison doesn't mean much for a SCM where it is a nontrivial task to just get a local copy of the files in the repo19:05
export CVSROOT=..., or cryptic long -d:pserver:foo@bar.com:/path my ass (and "what was the module called again, sourceforge does not list them") :P19:06
csanz left19:06
bremner cvs log. that is all.19:06
bookwar1 joined19:06
divVerent bremner: which part19:07
the part "global revisions are overrated"19:07
or the part "everyone knows what 1.3.2.2.1.109.2.1 stands for"?19:07
bremner it's a holistic pain thing19:07
cbreak commit dec32a4644330965d2236ba199dc24b943f644a119:07
divVerent seriously, cvs managed to have even more cryptic revision specifiers than git :P19:07
with git one at least KNOWS they mean nothing :P19:07
cbreak it's a hint.19:08
divVerent cbreak: is that commit published anywhere?19:08
Google doesn't have it in the index :P19:08
foocraft left19:08
cbreak it's private to my company19:08
well, the one I work for19:08
divVerent but seriously, it would be nice if there were some global index of git revisions19:08
ChanServ set mode: +v19:08
divVerent so you have a revision hash, and get clone URL and branch of it :P19:09
saysjonathan joined19:09
tango_ ldh: btw if the cvs stuff only needs a server, git can fake one19:09
divVerent I had hoped Google does it, but my quite public project is not indexed19:09
ldh tango_: good to know. unfortunately in my case it involves lots of developers (who may be hesitant to switch) and it's out of my hand19:09
s19:09
bookwar1 hello, nice silent company you have here )19:10
_iron joined19:10
tango_ I assume firing them is not an option, uh? ;-)19:10
pdelgallego left19:11
pdelgallego_ left19:11
pdelgallego__ left19:11
foocraft joined19:11
deadlyicon joined19:12
raz joined19:13
markelikalderon joined19:13
bookwar1 help me please with the small question: how to get the current revision of a folder with subfolders from a huge remote git repository?19:13
leavittx left19:13
PigeonFriend joined19:13
cbreak folders don't have revisions19:13
crashanddie left19:13
cbreak you want to get the currently checked out HEAD hash?19:14
git rev-parse HEAD will give that19:14
j416 cbreak: I'm going to guess he wants the latest version of the project without its history19:14
Lord_Khelben bookwar1: you want to clone a repo without all the commits ?19:14
cbreak git rev-parse master?19:14
Lord_Khelben then use --depth in clone19:14
bmahe joined19:14
bookwar1 j416: yes, and not the whole project but some part of it19:15
j416 bookwar1: then as Lord_Khelben says19:15
pdelgallego_ joined19:15
pdelgallego joined19:15
j416 git clone with --depth19:15
pdelgallego__ joined19:15
j416 bookwar1: man git-clone19:15
jast bookwar1: the 'git-clone' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-clone.html [automatic message]19:15
cbreak git can not clone only parts of a project19:15
you can either clone all of it, or nothing. all --depth allows you to do is limit history19:16
but such a repository is gimped19:16
kakos joined19:16
cbreak and won't work in many situations19:16
FauxFaux Also it's incredibly bad at actually saving space in practice.19:16
j416 bookwar1: if the project has a website with gitweb or so, you should be able to get an archive with just some specific subdirectory.19:16
cbreak I don't think it'll save much space indeed19:16
keyvan- joined19:16
cbreak but why would you only want part of a project?19:16
it'll be of no use due to dependencies19:17
j416 perhaps he's just interested in a particular file or collection of files19:17
joel_falcou joined19:17
j416 shrugs19:17
miloshadzic left19:17
FauxFaux There's no reason you can't request only HEAD's tree and relevant children over dumb transports. ¬_¬19:17
kakos I have a tracked file in my git repo that is stable and doesn't have any significant code changes, except for one line which is used point the program to a server. Devs change it to their localhost and consistently push up that file with the change because they are lazy and use "git commit -a". Is there a way to tell git to ignore an already tracked file?19:17
dekroning joined19:17
cbreak FauxFaux: there is19:18
FauxFaux kakos: man git-update-index --assume-unchanged19:18
jast kakos: the 'git-update-index' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-update-index.html [automatic message]19:18
dekroning how do I checkout a remote branch? git checkout origin/foobar right ?19:18
cbreak dumb transports are dumb19:18
FauxFaux dekroning: On modern enough versions, yeah.19:18
cbreak they can't unpack pack files and only send you the things you want19:18
bronson_ joined19:18
kakos FauxFaux: Thanks19:18
j416 kakos: why don't they have their own branches?19:18
dekroning FauxFaux: what does that mean exactly? :)19:18
cbreak dekroning: don't do that19:18
dekroning: use git checkout -t origin/foobar19:18
dekroning cbreak: aha why not ?19:18
cbreak because it will detach your head otherwise19:18
dekroning cbreak: -t is track right ?19:18
j416 kakos: easy enough to have a local branch with local settings and rebase that19:18
cbreak yes19:18
bookwar1 j416: your telepathy works great )19:18
FauxFaux Oh, is it actually "git checkout foobar" that works in modern versions?19:18
kakos j416: Because we're highly undisciplined and it's like herding squirrels to get them to do things half-right, let alone full-right19:18
cbreak it implies -b foobar19:19
j416 bookwar1: cool19:19
cbreak FauxFaux: yes19:19
FauxFaux: but I don't think it's good to rely on it in here19:19
since many people have old gits19:19
FauxFaux Debian'll release soon, then the problem will go away.19:19
cbreak no, they'll probably only update to 1.619:19
FauxFaux 1.7.2.3 is code-frozen.19:20
MikhX left19:20
LouisJB_ joined19:20
cbreak do they still have mplayer rc2? after four years? :)19:20
or did they finally give up on stupid releases and ust get svn head19:20
LouisJB left19:22
LouisJB_LouisJB19:22
MattDiPasquale joined19:23
stew it looks like it will be "1.0~rc3++final". excellent version string19:23
opalka joined19:23
foocraft left19:23
coletg joined19:24
melvinram joined19:24
bookwar1 left19:24
melvinram is there a way to undo a commit without loosing the changes?19:24
I committed too much19:24
I want to undo the commit and redo them as smaller commits19:25
HG` left19:25
FauxFaux melvinram: FAQ undo last commit.19:25
cbreak melvinram: git reset --mixed HEAD~119:26
you can leave away --mixed since it's the default19:26
but do NOT write in --hard or --keep19:26
LouisJB left19:27
melvinram thank you!19:27
kakos FauxFaux: Thank you once again. You have saved what little hair I have left19:27
GaVrA left19:27
peshalto joined19:28
MikhX joined19:28
pantsman- joined19:28
sm left19:28
Evious joined19:29
GaVrA joined19:30
MikhX_ joined19:30
PigeonFriend left19:32
montylounge joined19:34
miloshadzic joined19:34
Kronuz joined19:34
Kronuz hello19:34
rodrigo_golive left19:34
tstclair joined19:35
coderdad left19:35
Kronuz hey, if I created a repository using git init… then I just put the .git folder in my server to serve my repo...19:35
MikhX left19:35
MikhX_MikhX19:35
Kronuz I get this error when I try to clone from it: error: The requested URL returned error: 500 while accessing https://git.myserver.com/git/project/info/refs19:35
how do I create that refs?19:36
ksmith left19:36
Lord_Khelben after git init, did you add any content and make commits ?19:36
foocraft joined19:36
Kronuz Lord_Khelben: yes, I did19:37
then I copied my .git to my server19:37
I thought that was the easiest to publish my repo19:37
(until I got that "missing refs" error)19:37
Lord_Khelben normally you create a bare repo on the server and from your normal repo you push to it19:37
AAA_awright Kronuz: What clone method are you using?19:38
Kronuz just git clone19:38
from a web server19:38
AAA_awright In the past you've had to run git update-index or something19:38
If it's a stupid clone, like HTTP19:38
Kronuz it's the "smart" one or something19:38
thru https19:39
AAA_awright That just means it's encrypted it has nothing do do with smart19:39
Lord_Khelben you mean git update-server-info19:39
AAA_awright If you want smart over HTTP you have to setup some CGI program19:39
Kronuz: You should just git push to the repository if you can19:39
ki__ left19:40
Kronuz AAA_awright: the other repos are working fine19:40
AAA_awright Yeah, git update-server-info19:40
Lord_Khelben also the config from the .git folder you copied will have bare=false19:40
ki__ joined19:40
jpalmer joined19:40
Lord_Khelben when it should have bare=true19:40
change that too19:40
AAA_awright I'm not sure what that does but yeah set that too19:40
How it handles HEAD?19:40
jpalmer quick question, if I want to migrate a repo from one server to another, can I just do a git clone, change the origin to the other (new) server, and git push? will it retain all the previous commits and tags/branches?19:41
nstorer joined19:41
Kronuz git update-server-info returns command doesn't not exist19:41
I changed the bare to true19:41
update-index did work, but it didn't add the refs to info/19:42
nstorer left19:42
marc_kdab left19:42
pantsman- left19:42
Lord_Khelben jpalmer: i think the cloned will have everything besides the hooks you may have, but i may be wrong19:43
PovAddict joined19:43
PovAddict can git rev-list show all commits *except* those from a certain author?19:43
--not --author=foo doesn't seem to work19:43
cbreak cloning will not clone hooks, configurations, ...19:43
it will only clone refs and their history19:43
Lord_Khelben i remembered it correctly then :)19:44
cbreak there are different clone methods that differ in which refs are cloned and how19:44
Kronuz in my other repos ther's a info/refs file19:44
which is not in this new repo19:44
I tried git init --bare19:45
but it didn't either create a info/refs file19:45
cbreak don't know what that file is good for19:45
just ignore it19:45
PovAddict it's for HTTP cloning19:45
it's created by git update-server-info19:45
Kronuz update-server-info is giving me an error, it doesn't exist (the command)19:46
:(19:46
Lord_Khelben it should be there19:46
Kronuz :P19:46
I was trying update-server-index19:47
sorry19:47
bronson_ left19:47
redondos left19:48
tvw joined19:49
aalex_laptop joined19:50
bronson_ joined19:51
Eridius joined19:51
PigeonFriend joined19:51
panike joined19:52
tizzo left19:54
tizzo joined19:54
necrodearia joined19:56
necrodearia make a new branch based on the master branch, fix bug, merge back into master AND into staging branch - see http://nvie.com/posts/a-successful-git-branching-model/19:57
tgunr left19:57
mkramer1 left19:57
necrodearia How do I make a new branch based on the master branch? is it simply `git checkoug -b <branchname>?19:57
iband left19:57
iband joined19:57
necrodearia s/checkoug/checkout/19:57
malfy left19:58
aah joined19:58
Lord_Khelben necrodearia: if your current branch is master yes, otherwise git checkout -b branchname master19:59
necrodearia okay, thanks.19:59
PovAddict left20:00
TheFox21 left20:01
malfy joined20:01
TheMoonMaster left20:02
booginga left20:03
dfr|mac left20:03
marc_kdab joined20:05
_Vi left20:06
asmega joined20:06
shennyg joined20:06
marc_kdab left20:06
marc_kdab joined20:06
_Vi joined20:07
krosenvold joined20:08
trobrock joined20:08
arkonova joined20:09
hipstertrash left20:14
engla left20:14
amondracorp joined20:15
malfy left20:15
wtfness joined20:17
foocraft left20:17
hipstertrash joined20:17
verdon left20:18
malfy joined20:18
marc_kdab left20:20
malfy left20:22
joeytwiddle left20:22
marc_kdab joined20:23
echrsin joined20:24
panike left20:25
marc_kdab left20:25
marc_kdab joined20:25
panike joined20:25
s0ber_ joined20:26
necrodearia After fixing bugs (as mentioned above) and I want to merge the new brach back into master, how do I do that? `git merge dev` outputs "Already up-to-date."20:26
s/master/dev/20:27
imMute necrodearia: switch to trunk, *then run git merge dev20:27
madewokherd joined20:27
bigpeted joined20:28
s0ber left20:28
s0ber_s0ber20:28
fr0sty-away left20:28
necrodearia imMute, hmm, that also shows "Already up-to-date."20:29
imMute :/ I could have sworn that was how you do it..20:29
meebey left20:29
ldh should be. maybe the changes on the branch weren't committed?20:29
bmahe left20:29
imMute or you didn't switch to the master branch.20:30
necrodearia imMute, While in dev branch I `git checkout -b dev.nobitcoin`ed and then made changed, commited and pushed and now I want to merge them wit the dev branch20:31
imMute necrodearia: so `git checkout dev` then `git merge dev.nobitcoin`20:32
necrodearia error: Your local changes to 'httpdocs/config.php' would be overwritten by merge. Aborting. ------ Please, commit your changes or stash them before you can merge.20:32
imMute you have uncommitted changes20:33
deal with them (in either way the error told you to)20:33
necrodearia The same changes that I pushed to dev.nobitcoin20:33
necrodearia nods20:33
ldh you didn't push them because they were unstaged when you committed, i'm guessing20:34
imMute which means they weren't actually comitted20:34
joeytwiddle joined20:34
radarek left20:34
joeytwiddle left20:34
Merbo joined20:35
larsmoil left20:35
Merbo hello20:35
larsmoil joined20:35
larsmoil left20:35
aalex_laptop left20:36
Merbo left20:36
larsmoil joined20:36
joeytwiddle joined20:36
hipstertrash left20:37
aalex_laptop joined20:37
hipstertrash joined20:37
PigeonFriend left20:38
Kronuz thnaks Lord_Khelben :) …that update-server-info made it work!!20:39
larsmoil left20:39
Lord_Khelben Kronuz: i am glad it worked. actually it was AAA_awright that mentioned it20:40
Kronuz oh, thanks AAA_awright :)20:41
doener left20:41
abetaha joined20:43
yhager left20:43
Geralt joined20:44
marc_kdab left20:44
bmahe joined20:44
morganick joined20:44
rettub left20:44
phishy__ joined20:44
malfy joined20:44
Geralt Hi, how can I get the name of the branch that's currently checked out? I couldn't find a command that does that in man git, but I suppose that's a fairly common task20:45
jast Hi: the 'git' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git.html [automatic message]20:45
phishy_ left20:45
bronson_ left20:46
imMute Geralt: git branch20:46
patrikf git status will also tell you, iirc20:46
tizzo_ joined20:47
rettub joined20:47
Geralt oh I forgot to mention one important piece of information: I want to use it in a program, and git branch always prepends a * is there something that prints just the plain branch name?20:48
malfy left20:49
tizzo left20:50
tizzo_tizzo20:50
patrikf Geralt: git symbolic-ref HEAD20:51
malfy joined20:51
marcamilly joined20:51
patrikf note that a git repository doesn't have to have to be at a branch at all times20:52
Geralt patrikf: thanks, however there's now way to detect if a working directory is in detached head state or if it's not a Git repo at all, the exit code is in both cases 128.20:52
now=no20:52
chrisshattuck left20:53
ChanServ set mode: +v20:53
echrsin left20:53
marcamilly ok...awesome20:53
Maletor joined20:53
marcamilly so I just did a git pull from a remote repo20:54
but it destroyed a bunch of stuff I didn't want destroyed20:54
cbreak pull doesn't destroy anything20:54
patrikf Geralt: you might combine it with "rev-parse HEAD"20:54
marcamilly specifically, lots of images were deleted.....so I want to do two things: 1) I want to go back to the commit before, 2) pull down some of the changes to a new branch20:55
cbreak: you are right, but the changes in the last push do20:55
*did20:55
Geralt patrikf: that would work :-) Thank you!20:55
Chillance joined20:55
malfy left20:55
dmlloyd left20:55
cbreak marcamilly: if you have no local uncommitted changes, do git reset --hard HEAD@{1}20:56
marcamilly GREAT!20:56
patmaddox left20:56
panike left20:57
stliu left20:58
ptl joined20:59
_iron left20:59
marcamilly brilliant20:59
ok....so now that I know that the pull will be destructive to some of my local images21:00
but I want some of the changes to the code21:00
how do I proceed ?21:00
i.e., I want to keep the stuff that were added and not merge the 'deletions'21:00
spearce left21:00
yhager joined21:01
dmlloyd joined21:01
EricInBNE joined21:02
Geralt marcamilly: git pull --no-commit then edit the pending changes to your likings :)21:03
dev2 joined21:03
marcamilly ahh...interesting21:03
guardian left21:04
marcamilly so would it be the full path 'git pull origin master --no-commit' ?21:04
lresende_ joined21:05
dev2 I have done a few commits. Is it possible to revert to a previous commit? I can't remember what exactly I named it21:05
albel727 left21:05
marcamilly error: unknown option `no-commit'21:05
albel727 joined21:05
Lord_Khelben dev2: can't you see the sha1 from git log ?21:05
j416 dev2: "git log" will show you the commits in the branc you're on21:05
branch*21:05
ChUrP left21:05
friskd left21:06
dev2 oh thats great!21:06
malfy joined21:06
Geralt marcamilly: put the --no-commit after the pull not after the remote21:06
iband left21:06
Lord_Khelben dev2: then you can reset your tree to that commit. the commits you revert aren't lost, so if you mess up, reflog is your friend21:07
dev2 Lord_Khelben: beautiful, I was afraid of getting berrated for asking a stupid question :)21:07
lresende left21:08
lresende_lresende21:08
shruggar sitaram, are you there?21:08
Lord_Khelben have you pushed the tree somewhere ?21:08
if yes, then it is better to actually "revert" the commits otherwise people that try to pull will be annoyed :P21:09
hacim left21:09
marcamilly Geralt: sweet...so how do I step through this now and commit what I want ?21:09
dev2 Lord_Khelben: if thats in reference to me, its all good, I'm just working locally21:09
malfy left21:10
hacim joined21:10
bmahe left21:11
ChUrP joined21:12
lresende left21:13
banisterfiend joined21:13
vortec left21:14
montylounge left21:14
jga left21:15
hacim left21:15
phishy__ left21:16
hacim joined21:16
ChUrP left21:16
ISF_ left21:16
malfy joined21:18
wtfnessfoocraft21:20
elmob left21:20
bigpeted left21:20
keyvan- left21:20
ChUrP joined21:21
j416 left21:22
hacim left21:22
matjas left21:23
malfy left21:23
Ryback_ left21:23
ChUrP left21:24
Kronuz left21:25
yhager left21:25
bmahe joined21:26
marcamilly Geralt: there ?21:26
markelikalderon left21:26
francogrex joined21:27
xiong left21:28
francogrex Hi I am on windows and downloaded the msysgit. How can I download all files in this project https://github.com/xach/linj using only the comand line?21:29
snearch left21:29
hacim joined21:29
Geralt marcamilly: sorry, I was afk. If you type git status you'll see a list of things that are about to be commited (you can see the diffs with 'git diff --cached') if you don't want something to be in this commit use git revert HEAD -- <filename> and it won't be in that commit. If you read the lines of git status that start with # it tells you always how to undo a change or include a change21:30
larsmoil joined21:32
montylounge joined21:33
tty1 joined21:33
francogrex left21:33
albel727 Geralt: git revert HEAD -- <filename>? you mean, git RESET?21:33
MattDiPasquale left21:33
chrischambers_ joined21:34
marc_kdab joined21:35
chrischambers left21:35
chrischambers_chrischambers21:35
Geralt albel727: oh...yes, I always have to look at the hint from "git status" before I unstage something...21:35
albel727 yeah, happens. =)21:35
alester joined21:36
Mage joined21:36
albel727 I guess, "READ what git tells you" is the best meta-advice out there =)21:36
malfy joined21:36
Geralt definitely :)21:37
bronson_ joined21:39
utu joined21:40
MattDiPasquale joined21:40
stew so read what #git tells you about reading what git tells you is the best meta-meta-advice21:40
malfy left21:40
warbee joined21:41
amondracorp left21:41
ManDay joined21:42
malfy joined21:43
ManDay Is there any git manpage that lists all manpages? In fact, I'd have expected git(7) to list all the manpages that come with git, but it still lacks at least a few21:43
wereHamster albel727: you can make it even shorter: READ (or RTFGO - Git Output)21:43
rgr joined21:44
rgr left21:44
stew RTFGO or GTFO21:45
ManDay same thought here21:45
psoo joined21:46
albel727 lol21:46
ManDay in case there is no such list, any chance git(7) will have it added in the next release?21:46
wereHamster ew, I didn't notice the similarity -.-21:46
ManDay: if you make a patch, sure21:46
ManDay wereHamster: uhm well, if I find a complete list in the first place nothing easier than that21:46
vgr joined21:47
ManDay but i guess one of (you?) guys is quicker at addint the line?21:47
wereHamster ManDay: find /usr/share/man/ -name 'git*'21:48
albel727 apropos git?21:48
fr0sty-away joined21:48
wereHamster some man pages have no corresponding git command21:48
cubuanic left21:48
wereHamster (gitignore, gittutorial etc)21:48
marcamilly hrmm Geralt so what happens if I want something that was deleted by another developer to not be deleted on my machine after the commit ? in other words, I run git status and I see a bunch of 'new file:' those are good and I want all of those....but then I see 'deleted: <filepath>' I don't want to commit those deletes21:48
ManDay yes. there were others the name of which icant recall21:48
marcamilly how do I exclude those ?21:49
ManDay funnily enough the wiki on kernel.org states "git(7) serves as an overview of git commands and man pages."21:49
I'd love that :)21:49
cking_nj joined21:49
marcamilly if I am not mistaken, your instructions work for the situation where I want to 'exclude' something that was added (like a file) to the repo21:49
but not quite sure how to handle a delete command - where it deletes files I want to keep21:50
Geralt: I hope that makes sense21:50
ManDay git: nothing appropriate21:50
cking_nj left21:50
ManDay just as for the apropos21:50
levifig joined21:50
radarek joined21:50
Geralt marcamilly: use git checkout -- <filename>21:50
marcamilly: wait that's wrong21:51
marcamilly hrmm21:51
i was wondering21:51
remember I am in between merges and commit21:51
i have done a git pull21:51
stamina_ left21:51
marcamilly --no-commit21:51
Geralt marcamilly: do as git tells and go with 'git reset HEAD <file>..." to unstage'21:51
marcamilly so now I have to commit only the things I want to21:51
git doesn't tell me anything :|21:51
Geralt as I said above I always follow the hints in git status if I need to do some tinkering with staging area21:52
marcamilly ok, how do I see the hints ? do I need an additional flag ? because this is not giving me any hints21:52
this is what I see: https://gist.github.com/80106521:53
ManDay Can anyone who knows submodules confirm that it's perfectly all right to create submodules "inplace" - meaning not backed up by a remote/bare repository. For example two submodules A and B in an otherwise completely added C (C contains A and B)?21:53
Geralt marcamilly: strange there are really no hints21:53
marcamilly lol....that's what I said...LOL21:54
:)21:54
thought I was being a n00b ?21:54
Geralt marcamilly: no, I supposed you were just missing it if the output was really huge21:55
marcamilly: to remove a change from the index use 'git reset HEAD -- <filename>'21:56
in fact some people add a git alias so that they can use the more intuitive 'git unstage <filename>'21:57
marcamilly so in my case, I would do something like 'git reset HEAD -- 02 Login Screen.jpg' or would it be the full path like this 'git reset HEAD -- public/system/images/2/original/02 Login Screen.jpg'21:57
?21:57
wereHamster Geralt: the HEAD is implicit, and unless you have a branch with the same name as the filename, git reset <filename> is just fine21:57
marcamilly: full path within the git repo21:58
marcamilly: and quote the path if it has whitespace21:58
marcamilly ok21:58
fatal: ambiguous argument 'public/system/images/1/original/Argentina-2010-World-Cup.jpg': unknown revision or path not in the working tree.21:59
Use '--' to separate paths from revisions21:59
Bass10 joined22:00
marcamilly i did this: git reset "public/system/images/1/original/Argentina-2010-World-Cup.jpg"22:00
goshawk left22:00
Pewpewarrows left22:00
dirthead joined22:00
CarlFK joined22:00
wereHamster marcamilly: it appears the file is not trakced22:01
tracked even22:01
marcamilly well I tried the suggestion Geralt had, and I get this: Unstaged changes after reset:22:01
M public/system/images/1/original/Argentina-2010-World-Cup.jpg22:01
Bass10 left22:01
marcamilly so was that successfully done ?22:01
SeanLazer left22:01
dirthead I have a commit in a bare repository that is causing Trac problems. How can I back the commit out of the bare repository. git-reset doesn't "make sense" in a bare repository.22:01
Bass10 joined22:02
CarlFK I just did "git checkout origin/master episode.html" to update that single file. and stepped on some changes to it. how can I revert back?22:02
wereHamster marcamilly: that's what to be expected after the reset.22:02
marcamilly great...so it worked22:03
so looks like you do need the syntax HEAD --22:03
ManDay Can anyone help me with submodule please?22:03
marcamilly 'git reset HEAD -- <full file path>' is what works for me....fyi22:03
wereHamster marcamilly: I doubt it. The <commit> defaults to HEAD in all forms.22:03
Reemo joined22:03
acid joined22:03
marcamilly hrmm.....well iono why it's working with the HEAD and wasn't without it22:04
but it is :|22:04
wereHamster CarlFK: git status22:04
Geralt marcamilly: that's ok git just removes that change from the index, but it's still in your working directory. First unstage all the "wrong changes" and then commit the "good changes" after that I'll tell you how to fix your working directory (because your git status doesn't :/)22:05
sako left22:05
Geralt btw. what does git --version say?22:05
CarlFK wereHamster: http://dpaste.de/6Uip/22:05
autojack joined22:06
wereHamster CarlFK: and what do you want to do?22:06
marcamilly 1.7.3.222:06
dbykov joined22:06
dbykov left22:06
autojack is there a way to get git to 'print' the contents of a file in the repo, given a path and possibly a specific branch?22:06
marcamilly I will continue adding these to unstaged, then ask for advice after22:06
autojack git cat-file seems like it should do what I want but I can't really figure it out.22:06
nadavoid left22:06
ben_alman left22:06
wereHamster autojack: git show <rev>:<path>22:06
Reemo Hey guys, i wondered if someone could recommend a good (opensource?) scm infrastructure (talking about bugtracking, ldap account management, etc.) build around git22:06
I tested gitorious on a local server, but found it too unflexible for my needs (especially the account and install procedures)22:07
CarlFK wereHamster: get the version of episode.html as it was before the checkout22:07
Geralt that's strange your git version if newer than mine so git reset without HEAD should definitely work and I don't understand why you don't see any help messages in your git status output22:07
benishor_ left22:07
wereHamster CarlFK: did you have local changes which you haven't added to the index yet?22:07
ManDay submodules, anyone? :-/22:08
yhager joined22:08
wereHamster ManDay: if you come across some, run screaming and kicking.. submodules are a pain in the ...22:08
SeanLazer joined22:09
autojack wereHamster: ideally this would be something I could run from any host, any directory, without having to check out a given repo and cd into it. is that possible?22:09
CarlFK wereHamster: not sure really. episode.html had some changes that were specific to the box it is on that I never intended on pushing to master. so I was just hack-n-slashing on it.22:09
wereHamster autojack: no22:09
sonnym left22:09
ManDay wereHamster: I'm afraid so, but I think it's still the better way than creating an ugly mix of symlinks and making sure two complete repos dont interfere22:09
autojack rats.22:09
wereHamster CarlFK: if not, then you are out of luck22:09
CarlFK then I made some changes on my primary copy, pushed to master and plunked on top of my hacked ver. now I want the hacked ver back so I can merge22:09
wereHamster ManDay: google 'gitslave'22:09
autojack suppose I could pull it over http from the web interface22:10
wereHamster autojack: sure..22:10
autojack I have a test script that needs to parse the config files from two different repos before it can run, that's all.22:11
but maybe checking out the branch is my best bet.22:11
CarlFK wereHamster: is there some way to tell if anything was added to index?22:11
ManDay wereHamster: you see, it's even worse than that. In fact I got DirA and within DirA I got DirB throught DirE - I want DirA to be a repository which manages Everything BUT DirD and DirE and *one* (1) other repository to manage DirD and DirE22:12
sounds like fun, eh?22:12
is there perhaps any way I could configure a repository so that it does NOT follow nor account for any symlinks?22:13
wereHamster CarlFK: you could go through all unreachable blobs and see if one matches your file22:14
autojack left22:15
thermal joined22:15
SeanLazer left22:15
melvinram left22:15
theoros_ joined22:15
ChillaxChillaway22:15
theoros left22:16
desaiu joined22:17
pheaver left22:18
pheaver joined22:18
Maletor left22:19
yhager left22:19
yhager joined22:19
necrodearia I want to set up a one minute crontask to pull from a git repository. How can I prevent the need to enter a password each `git pull` attempt?22:20
jxpx777 joined22:21
bremner necrodearia: you understand that pull is meant to be interactive, right?22:21
tizzo left22:21
necrodearia bremner, Yes, but I am setting up a development environment that I want to pull from a local git repo so that I do not have to manually pull each time I want to view the website related to the dev branch of the repo.22:21
bremner necrodearia: can't you use hooks instead?22:22
janis joined22:22
necrodearia bremner, I have no idea. How can I do that?22:22
bremner necrodearia: well, the usuall answer is not to have the live files under version control, but just a snapshot generated by git-archive in a commit or post-update hook.22:23
avalys joined22:23
avalys is there any way to have "git status" return only the status of files under the current directory (and subdirectories)22:23
jxpx777 left22:25
marcamilly ok Geralt I got it22:25
rbq joined22:25
necrodearia bremner, Does that mean it is not possible to automate preparing latest snapshot of git repo branch to be used in a live development environment?22:26
marcamilly Geralt: this is how it looks: https://gist.github.com/80106522:26
rbq left22:26
bremner necrodearia: not at all. It just means that pull is not the right tool for deploying websites (and neither is push).22:26
Geralt marcamilly: if you've unstaged all you want to unstage you can do a 'git commit' now22:27
marcamilly just 'git commit', not 'git commit -am "<message>"' ?22:27
SIGBUS left22:28
bremner somebody should write a "if you must use git as a deployment system, here is how" howto22:28
Geralt marcamilly: don't use -a or all work will be gone!22:29
rcpeterson left22:29
marcamilly ok22:29
i did that and now see a textmate window open22:29
atrofast left22:30
Geralt marcamilly: if you don't use any flag git opens an editor with a default message that contains a summary of the merge, I suppose you want to keep that message (and maybe add a comment of your own)22:30
marcamilly I guess it is asking for a message22:30
Geralt marcamilly: write your commit message then save&exit and your merge is commited :)22:31
marcamilly done22:31
i still have those unstaged stuff22:31
GaVrA on windows i have tortoisegit which enables me to use 'git gui' and when i have some conflict when merging i can use some handy tool to see that conflic. but now im on ubuntu and just cant figure out what to do22:31
marcamilly what do I do with those ?22:31
kenneth joined22:31
zorzar_away left22:31
bremner GaVrA: man git-mergetool22:31
jast GaVrA: the 'git-mergetool' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-mergetool.html [automatic message]22:31
Geralt if you don't want the changes in your working directory anymore use "git reset --hard" and they'll be gone with no way to get them back, so check twice before you do that22:31
bartj left22:32
avalys left22:32
marcamilly GREAT!22:32
that worked22:32
*phew*22:32
thanks meng22:32
wow....what a long exercise22:32
zorzar joined22:33
kenneth hey—i've got a git hook that does stuff. i'd like to do that stuff in the background and return immediately to the user. i tried doing ( do whatever ) & exit but it doesn't seem to work :/22:33
ideas?22:33
Geralt marcamilly: yeah, but the more you exercise the faster you'll become :)22:33
not too long ago a merge took me ages toot22:33
-t22:33
marcamilly cool22:34
necrodearia bremner, So I imagine to use hooks properly that I must configure the hooks with gitolite, right?22:35
marcamilly so to merge this with my other branch, all I do is checkout that branch and do 'git merge <new branch>'22:35
right ?22:35
wereHamster yes22:35
ormaaj left22:35
marcamilly great!22:35
janis left22:36
Agos joined22:36
joel_falcou left22:37
yhager left22:38
ISF joined22:40
rgoytacaz joined22:40
rgoytacaz how do I update a git project to its latest version?22:40
SimonP86 left22:40
wereHamster rgoytacaz: git pull22:41
rgoytacaz oh22:41
right22:41
tgunr joined22:41
nadavoid joined22:41
GaVrA bremner: thanks man, managed to work it out ;)22:42
bremner GaVrA: \o/22:42
HG` joined22:43
k04n joined22:43
k04n left22:43
GaVrA would it be offtopic to ask what mergetool is easiest to use?22:43
bremner GaVrA: no, not offtopic. I find kdiff3 ok. Some people like meld. Lately I use emerge more, but I already know emacs.22:44
GaVrA at first i found meld a bit confusing22:45
i mean i still do, so im thinking is there some easier tool for newbies :D22:46
bremner try kdiff3? there is a version kdiff3-qt which doesn't depend so much on kde22:46
saschpe left22:47
alvaro_o left22:47
GaVrA i type 'git mergetool --tool=kdiff3' to use that one?22:47
alvaro_o joined22:47
bremner think so22:47
Tuller joined22:48
ManDay Is there a command to "commit -a" only the changes within a particular directory?22:49
Tasser joined22:49
Tasser how to pull onto a new branch?22:49
kenneth i've got the following post-receive hook… anybody know why it ain't working?22:50
git push unfuddle master &22:50
exit22:50
tstclair left22:50
kenneth erm22:50
bronson_ left22:50
kenneth http://pastie.org/150744222:51
wereHamster kenneth: and how is it failing?22:51
bronson_ joined22:52
kenneth the stdin is not being passed on to the email script, which makes it fail22:52
ayust you may need to nohup the 'git push' if you're backgrounding it22:52
thermal left22:52
ayust oh22:53
chrisshattuck joined22:53
bronson_ left22:53
ayust have you tried having the script read stdin into a variable, and then sending that explicitly to the emailer's stdin?22:53
Noldorin_ left22:54
GaVrA is it wise to go and alter past commits? i mean when i created git repo i still had entire zend framework in that project. so for few commits it was inside the project, all ~2000 files of zend framework... after we have moved zf to php's include path, but im thinking that my repo is now a bit 'heavy' couse of those initial commits that had zf. what should i do? i was reading something about alterign past commits and simply removing entire zf folder from each comm22:56
bentob0x left22:56
psundaram joined22:56
psundaram left22:56
ayust it depends; is your repo shared among multiple people?22:56
kenneth ayust: hmm, no. let me see if i can figure out how to do that22:56
eijk left22:56
ayust rewriting shared history can be problematic22:57
fr0sty-away left22:57
morganick left22:57
GaVrA ayust: 3 people22:57
eijk joined22:57
nullvariable left22:57
ayust are you familiar with the issues that 'git rebase' can cause, GaVrA ?22:57
jxpx777 joined22:57
GaVrA i was also thinking about just to go and delete that git repo and start new one22:57
fr0sty-away joined22:57
GaVrA ayust: no i am not22:58
ayust okay22:58
timepilot left22:58
GaVrA but i did read a bit about it22:58
ayust well, you *can* rewrite history, but when you push that history, everyone else's history will no longer match up22:58
GaVrA ok22:58
ayust which means that everyone else will have to either reset back to your new version (losing any changes they might have made after you last pulled)22:58
eijk left22:58
ayust or rebase their changes on top of the new, rewritten history22:59
GaVrA ok how about this22:59
ayust (which can vary in difficulty)22:59
GaVrA we are all on last commit22:59
i rebase history22:59
then it should be ok when they pull?22:59
linusoleander left22:59
ayust they'll need to "git fetch && git reset --hard origin/master"22:59
instead of "git pull"22:59
since their local history won't match up with what's on origin anymore22:59
GaVrA and it would work?22:59
ayust so git pull would get confused22:59
spearce joined22:59
ayust that would work, yes. just make sure that everyone is on the same commit first.23:00
GaVrA righto23:00
ayust (and that everyone knows they need to reset afterwards)23:00
GaVrA and the process of changing history can be done how?23:01
ntoll joined23:01
ManDay left23:01
ayust GaVrA: man git-filter-branch23:01
jast GaVrA: the 'git-filter-branch' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-filter-branch.html [automatic message]23:01
kenneth alright, ayust so now i've got it working with your trick, but the backgrounding still doesn't seem to work…23:01
http://pastie.org/150744223:01
kuhrt left23:02
rettub left23:02
kenneth i thought it should, with the &. but the git hook doesn't actually return until the scripts are done. (i still see the stdout of those scripts in the git returned value)23:02
nullvariable joined23:03
rettub joined23:03
Textmode joined23:03
ntoll left23:03
GaVrA ayust: thanks man, bookmarked it, gonna think about what to do... cheers!23:03
tizbac left23:04
jfelchner joined23:04
ayust kenneth: not sure what to tell you; that should background23:05
utu left23:05
elbod joined23:06
LiamH left23:06
GaVrA i was talking with friend of mine about converting some mercurial repo to git repo, does that hmm create some overhead or something? i mean like unnecessary files or something23:07
mikemeyer left23:08
shruggar left23:08
mikemeyer joined23:08
kenneth ayust: i just tested it by "backgrounding" a script that has sleep 5; echo hello and the hook waits for the script to finish and output before returning23:10
i don't quite get it23:10
wereHamster GaVrA: not if the conversion is done properly23:10
webPragmatist hey all23:10
i have a repo that we lost the origin to but i have a bunch of branches that are only on origin23:10
jfelchner left23:11
webPragmatist how can i drop those/23:11
wereHamster webPragmatist: git remote prune23:11
webPragmatist and the name of the branch23:11
?23:11
Textmode left23:11
wereHamster webPragmatist: man git-remote23:11
jast webPragmatist: the 'git-remote' manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-remote.html [automatic message]23:11
webPragmatist reading23:11
CarlFK1 joined23:12
Textmode joined23:12
CarlFK left23:12
webPragmatist wereHamster: i did git remote prune origin23:13
but there are still red branches listed under remotes/origin23:13
adnap how can i see all the versions of a particular file? i don't want to see a version that shows all the different changes. i want to see each instance of the file for each respective commit that modifies the file.23:14
rgr joined23:14
wereHamster webPragmatist: they will always be red23:14
red means it's a remote trackign branch, not that it doesn't exist in the remote23:14
webPragmatist how do i drop those?23:14
wereHamster webPragmatist: git branch -r -D ..23:15
webPragmatist should i do this?23:15
because i can't check them out23:15
wereHamster why do you want to delete them?23:15
teuf adnap: git log filename will show you all commits that touched this file, dunno if that's helpful to you23:15
bousquet joined23:15
floppyears hi guys23:16
webPragmatist wereHamster: because i can't check them out?23:16
wereHamster webPragmatist: you are not supposed to check them out23:16
floppyears there's a --git-dir option, is there a similar option to specify a remote git repository?23:16
adnap teuf: kind of. i want to see what the file looked like for each of those commits though.23:16
Dave^| left23:16
elbod left23:16
webPragmatist wereHamster: well they don't even exist though anymore/23:16
gusnan joined23:16
wereHamster webPragmatist: then delete them23:16
SimonP86 joined23:16
Lord_Khelben floppyears: have you checked the git-remote manpage ?23:16
adnap teuf: rather than a single file that shows all the modifications with +/-23:17
CarlFK1 left23:17
Lord_Khelben you can specify a url for a remote repository ofcourse23:17
hacim left23:17
floppyears thank you Lord_Khelben . I didn't know the right man page :)23:17
webPragmatist wereHamster: i tried git branch -r -D remotes/origin/enterprise-1.9.1.123:17
teuf adnap: git show sha1:filename will show you the file at this revision23:17
wereHamster webPragmatist: git branch -r -D origin/enterprise-1.9.1.123:17
Lord_Khelben floppyears: you give it a name and use that from then on. "origin" is such a repository23:18
Geralt left23:18
adnap teuf: thanks23:18
floppyears Lord_Khelben: I don't have a local copy of the repository, the repository is in another server.23:18
Lord_Khelben floppyears: then you can clone it, and the remote will be setup automatically23:18
zombor left23:19
Lord_Khelben floppyears: run git clone url_of_repository23:19
MattDiPasquale left23:19
adnap teuf: how can i modify the file to look exactly as it did for a particular commit?23:19
wereHamster adnap: git checkout <commit> -- <file>23:19
floppyears Lord_Khelben: I'm trying to setup redmine, but we have a bunch of webheads, and we are trying to avoid having a bunch of copies/clones of repositories (one per webhead) :(23:20
theoros_ left23:20
theoros joined23:20
theoros left23:20
theoros joined23:20
psoo left23:20
adnap wereHamster: is this a reasonable way to resolve a merge conflic? just change it so that it did in a commit that i like?23:20
wereHamster: *looks like it did23:20
wereHamster if that's what you want..23:20
Lord_Khelben floppyears: git is a distributed vcs and everybody works on his local clone. i don't know if there is another way23:20
floppyears: i am not an expert though23:21
arkonova left23:21
floppyears thanks Lord_Khelben , the man page is not shedding any light23:21
wereHamster floppyears: a tutorial would though23:22
radarek left23:22
Lord_Khelben floppyears: git supports shared repositories and many other tricks. wait to hear from someone with more knowledge, so he can recommend you some way of working23:23
webPragmatist wereHamster: so a normal repo that is newly cloned is *master , remotes/origin/HEAD -> origin/master, remotes/origin/maste ?23:23
spaceonline joined23:23
webPragmatist wereHamster: and a headless repo should just have master?23:23
the "origin" in this case23:23
unless there is a branch of some sort23:24
wereHamster depends on how you created it23:24
webPragmatist but that's pretty typical/23:24
sonnym joined23:24
adnap if i have already checked out some different branches, is git checkout the right way to switch between them, or is there another way?23:24
wereHamster webPragmatist: what is typical?23:24
webPragmatist wereHamster: typical meaning if someone else looks at this theres limited wtf factor23:25
adnap sorry for all the noob questions23:25
webPragmatist adnap: yes23:25
wereHamster webPragmatist: I assume by 'headless' you mean bare. And how a bare repo looks depends on how you created it. WHich you haven't told me...23:25
webPragmatist wereHamster: well, i created it using --mirror… which may not have made sense… as I said before I just have one branch since we lost our original headless repo23:27
morganick joined23:27
webPragmatist we just have what i was working on23:27
wereHamster .. and what is it you are trying to do?23:27
noahcampbell left23:27
webPragmatist well i've already done --mirror from my working copy23:28
i think i created everything properly23:28
i'm just trying to get the origin back up23:28
with just version history from the master branch23:28
disappearedng joined23:28
disappearedng left23:28
disappearedng joined23:28
wereHamster I'm confused as to which repositories you have, which ones were cloned from where, which branches they have etc.23:29
webPragmatist all i have is my working copy which only had the master branch23:30
artefon left23:30
webPragmatist and that's al it23:31
wereHamster ok. And what is it you are trying to do?23:31
elbod joined23:31
webPragmatist create a remote repo from this23:31
normanrichards left23:32
webPragmatist so at the very lseast we have changes that occured on that branch23:32
wereHamster webPragmatist: create the remote repo (git init --bare) and push into it from your existing repo (git push <url> master)23:32
webPragmatist how i did it is just the backwards way of doing it?23:32
wereHamster pretty much, yes23:32
webPragmatist git clone --mirror23:32
okay well i have another one i need to do that i'll do with push :)23:33
--bare won't copy the remotes?23:33
wereHamster init --bare creates an empty repo23:33
webPragmatist oh can i clone without remote?23:34
wereHamster no23:34
don't use clone23:34
forget it23:34
create an empty repository on your server with git init --bare and then push whatever you want into it23:34
webPragmatist okay23:34
necrodearia bremner, in repositories/site.git/hooks/post-receive I placed `git archive --format=tar HEAD | (cd /var/www/ && tar xf -)` and this does almost exactly what I want. The only thing it does wrong is stores files from master branch. If I try executing `git checkout design` before that outputs: remote: fatal: This operation must be run in a work tree23:35
tvw left23:35
necrodearia How can I arrange for `git archive --format=tar HEAD` to output from a branch other than master?23:35
webPragmatist wereHamster: do i have toclear the cache because when i push it just says no changes?23:37
or remove the branch maybe?23:37
bronson_ joined23:37
wereHamster webPragmatist: pleaste pastebin the output23:37
desaiu left23:37
dfr|mac joined23:38
webPragmatist Everything up-to-date?23:38
wereHamster is that everything?23:38
and also give me the command you ran23:38
opalka left23:38
Evious left23:39
MikhX left23:39
jfkw joined23:39
webPragmatist alright so… i did git init --bare on the remote … then i modified my origin on my working copy to the updated origin…. and did git push origin master23:39
d0k left23:40
wereHamster webPragmatist: did you run init --bare in an exissting repository on the server?23:40
webPragmatist no23:40
radarek joined23:41
morganick left23:41
wereHamster webPragmatist: create a new, empty directory on the server. cd into it. run 'git init --bare'. In your local repo, run this: 'git push <url> master' where <url> is an url (such as user@host:path.git) and _not_ origin23:41
webPragmatist okay23:42
wereHamster pastebin the whole output23:42
necrodearia or if anyone else is able to assist, my question above is available.23:42
smolyn left23:42
wereHamster necrodearia: replace HEAD with your branch23:42
... so not obvious23:43
banisterfiend left23:43
Veejay left23:43
Evious joined23:44
hasrb joined23:44
jrmuizel left23:44
thiago_home left23:44
webPragmatist wereHamster: worked thanks23:44
necrodearia wereHamster, ooh, thanks23:45
necrodearia offers a bitcoin to wereHamster for helping ^_^23:45
rseifert left23:46
ThomasLocke left23:46
coderdad joined23:48
MikhX joined23:49
bremner catches up on backlog a bit too late ;)23:50
beatak left23:50
aalex_laptop left23:51
jmil joined23:53
jxpx777 left23:55
jxpx777 joined23:55
chrischambers left23:55
cannonball left23:56
webPragmatist left23:56
esc left23:56
keyvan- joined23:57
radarek left23:57
TrentonDAdams joined23:58
Reemo left23:58
TrentonDAdams when using git-svn, can I branch off master without "tracking" remotely, and where dcommit does nothing?23:59
esc joined23:59
TrentonDAdams I'm thinking I'd like to track master, but nothing else. Use git what it's good for, merging, branching, etc, and then merge back to master without messing around.23:59

Logs Search ←Prev date Next date→ Channels Documentation