IRCloggy #git 2008-03-31

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.

2008-03-31

→ CelticSoul joined00:01
← sverrej_ left00:02
← lcapitulino left00:03
← ZabaQ left00:04
← Sonderblade left00:07
jengelh thiago_home: ah... repacking something already packed isfaster than packing from the start00:08
what's why koffice took so long00:08
kelvie is there a way you can get git-merge to automatically resolve whitespace changes?00:10
there's a lot of whitespace/formatting changes on the branch I'm trying to merge :/00:10
→ drylightning joined00:15
← drylightning left00:15
jengelh mergeoptions or so00:15
→ wagle joined00:18
← antifuchs left00:20
→ tomtt joined00:21
tomtt if i have pushed a changeset to origin/master, but now from my local branch have undone that changeset and done further work, how can I 'resync' origin/master with my local master? i.e. how do I undo a pushed changeset?00:23
← devurandom left00:24
RandalSchwartz you just want to reset local master to be the same as remote/master ?00:25
checkout master00:25
git reset --hard remote/master00:25
(replace remote with origin, sorry)00:26
tomtt?00:26
tomtt RandalSchwartz: no to other way around: i want origin/master to be like local master, but there is a commit on origin that i have undone locally00:26
RandalSchwartz so you want origin/master to be like its parent?00:27
do you want to do it as a second commit?00:27
tomtt RandalSchwartz: yes.00:27
RandalSchwartz so you want the history to show "A -> B -> A'" where A' and A are the same content00:28
but with a new commit message to say "I screwed up?"00:28
tomtt RandalSchwartz: that was yes to the first one. I'd rather the history just show A (it is not a shared repos), although it would be useful to know how to do both :)00:28
RandalSchwartz if you have to rewind something you've already pushed, it'll screw anyone else who has now pulled that repo00:29
and because of that, you'll have to "force" the update as a "non-fast-forward"00:29
← aroben left00:29
RandalSchwartz if you've ever wondered why there's "+" on git-push, you're about to find out00:29
but before you do that, let me make sure that's REALLY what you want.00:29
→ dann_ joined00:30
tomtt RandalSchwartz: yeah in this case it is ok, i am the only one working on the repos. although there is another case where i will have to do the a>b>a thing.00:30
RandalSchwartz for a-b-a', look at git-revert00:30
that's trivial. :)00:31
tomtt RandalSchwartz: ok :).00:31
RandalSchwartz for this, get your "master" history to look right already.00:31
git-log on master should show the commmits in the right order.00:31
then just do a push, with +00:31
git push +master:origin/master00:31
the "+" is a reminder that you've been a bad boy. :)00:32
as in, please avoid doing this in the future.00:32
tomtt RandalSchwartz: it says: ssh: +master: nodename nor servname provided, or not known00:33
RandalSchwartz gah. I always get the syntax wrong on that00:34
tomtt :)00:34
RandalSchwartz you can tell how often I have to do the wrong thing. :)00:34
git-push origin +master:master00:34
there00:34
tomtt RandalSchwartz: you are a better man than me :)00:34
RandalSchwartz if only you knew. :)00:35
tomtt RandalSchwartz: yep that did it, cheers!00:37
wagle_home so the + is a weaker form of --force?00:39
← tomtt left00:42
← Sho_ left00:44
→ tjafk2 joined00:44
← nkallen left00:51
gitster + is a more selective form of --force. You can say "git push there +master next" to force only master while protecting next with the fast-forward check.00:51
→ dsal joined00:52
← tjafk1 left01:01
← wagle left01:04
→ wagle joined01:07
→ miraculo joined01:08
→ nkallen joined01:11
→ SuttoL joined01:11
miraculo i did a git reset HEAD^, commited the new changes and attempted to push upstream, but get push failed 'non-fast forward'01:13
dsal Yep, that would happen.01:14
miraculo anybody got any clues?01:14
oh... well can i fix it?01:14
dsal You broke your head. You can push -f it to fix it.01:14
miraculo i own the upstream repo so happy to overwrite...01:14
dsal It's generally a bad idea because it'll confuse your collaborators.01:14
Yeah, I'm not saying I haven't done it several times. :)01:14
miraculo lol01:14
halorgium push the old ref to another branch if you want to keep the commit around though01:15
miraculo just the fact you can do it is a plus for git. ;)01:15
no it's really an update to the same commit01:15
missed something...01:15
thank dsal that did the trick ;)01:16
dsal np01:17
← tpiep left01:20
← LordCrimson left01:22
→ LordCrimson joined01:22
→ miraculo_ joined01:24
← miraculo left01:24
dsal OK, most common question... I've got staged changes and unstaged changes. How can I revert my unstaged changes?01:27
Or just, a file. I usually do a ``reset --hard'', but that gets angry if I give it a path.01:27
kelvie hrm so there isn't a way to automatically resolve whitespace differences inside git-merge right?01:28
I guess that should be done at the mergetool level01:29
dsal ``git checkout -- some/file'' seems to do what I wanted. Weird.01:29
kelvie oh yeah01:29
dsal yeah, I'd expect that to be the job of your merge tool.01:29
kelvie I found that a bit odd too01:29
to reset to the index you check it out :P01:29
dsal Too many commands that do roughly the same thing, but have different edge cases...01:30
kelvie something like git reset --to-index would probably be nice01:30
I'm sure most of us look in the git-reset manpage to find out how to do that01:30
dsal Yeah. I looked in there for a while, then tried the checkout thing. I've never done a checkout with a partial path before...that just seems a bit foreign to me.01:31
(I mean, I used to use CVS many years ago, but since then...)01:31
← tcoppi left01:32
→ tcoppi joined01:32
← strangy left01:33
→ Manoj joined01:35
Manoj Hi folks. I had some questions on rebasing; and have a test repository on git://git.debian.org/~srivasta/test.git set up for this.01:36
The repo essentially has two branches, master and kernel01:36
so kernel bracnhed off master, some development happened on both branches, and then there was a merge. post merge, the development is contibuing on each branch.01:37
→ _bdiego joined01:37
Manoj Is it possible to rebase the kernel branch to be rebased off the tip of the master branch?01:37
jengelh that's what git-rebase is for01:38
dsal git checkout kernel ; git rebase master01:38
?01:38
jengelh git-rebase --onto master startingpointrefspec01:38
Manoj well, since there was a merge in the middle, git gets hopelessly confused01:38
jengelh then don't merge01:38
Manoj I was hoping that there would be a recipe to manually guide git through that01:38
jengelh or twitch the tree until it is fit for rebasing01:38
sounds like you want a graft to get 'rid' of that merge01:39
Manoj that twitching is what I want to learn how to do01:39
jengelh echo '<shaofmergecommit> <onlyoneparentid>' >>.git/info/grafts;01:39
→ miraculo joined01:39
Manoj I stupidly made the same commit on both branches01:39
jengelh then rebase the branch and delete the graft afterwards.01:39
done :)01:39
basically you regraft the tree so as that the merge is not a merge anymore01:40
Manoj ooh, grafts. I have not heard of that before01:40
jengelh but I am curious as to how you would rebase something merged anyway01:40
← miraculo left01:40
Manoj which parent do I choose?01:40
jengelh I guess just rebasing with -i does the trick01:40
← SuttoL left01:40
→ miraculo joined01:41
← miraculo_ left01:41
jengelh just skip anything that you don't like ... and figure it out as it comes01:41
Manoj well, I have commits that were made (this is a imported arch tree with archimport that are confusing (arch has no commit --amend)01:41
jengelh http://jengelh.hopto.org/000.png -- suppose you have a tree like that01:42
then you would (probably) just rebase fnsubst onto master without caring much about the merge there01:42
Manoj but when I tried rease -i'ing the mainline, it essentially plit off the two branches into disjoint ones, with duplicate nodes01:42
jengelh or better yet, do two rebases01:43
Manoj: i just noticed.01:43
try segmented rebasing then01:43
meh someone killed the web01:44
← [tla] left01:44
Manoj are there any pointers to segemted rebasing?01:44
jengelh no, it's something I just made up :p01:44
well the term01:45
jengelh tries to find imgupload01:45
jengelh http://img100.imageshack.us/img100/9319/000jz0.png01:45
given that, you would use something like `git checkout -b somename 64393; git rebase --onto master e94e916!`01:46
er, git-rebase --onto master d63a4a1..6439301:46
so you only rebase part up to the merge01:46
then you continue with checking out the next segment from c9c565d and rebasing c9c..26ae01:47
onto somename.01:47
Manoj Oh, thanks.01:47
jengelh hope that makes sense01:47
Manoj goes off to map shasuums to the image01:47
jengelh the shas are there :)01:48
← leethal left01:49
Manoj so I select the start point as the commit before the branch, and the end point as the last commit before the first merge, and rebase onto master. got it.01:50
← miraculo left01:50
→ miraculo_ joined01:50
← scrooloose left01:52
→ scrooloose joined01:53
→ miraculo joined01:54
→ wayneeseguin joined01:55
← dann_ left01:57
→ dann__ joined01:57
← kukks left01:57
← miraculo_ left01:58
→ miraculo__ joined01:58
→ miraculo_ joined02:00
← miraculo__ left02:00
← miraculo left02:00
← nkallen left02:04
→ SuttoL joined02:05
shenson_not_here → shenson02:07
→ dalin joined02:09
→ garrin joined02:09
→ adante_ joined02:14
← IRSeekBot left02:23
← SuttoL left02:24
→ spearce joined02:24
→ woffle joined02:25
woffle Hey02:25
jengelh waffle :)02:26
woffle I'm having strange issues with git-clone02:26
When I clone a repo at most the transfer rate is 15KB/s and there are often 5 minute long pauses where nothing happens02:27
It's not just my connection to a specific server either, happens everywhere I try it02:28
This isn't normal is it?02:28
loops woffle, no it's not normal, sounds like a local network issue though02:28
woffle hmm02:29
loops woffle, are you cloning over http ? svn ? git ?02:29
woffle git02:29
>> git-clone git://anongit.freedesktop.org/git/mesa/mesa mesalib-git02:29
loops woffle, ah, well that site can be slow.. lemme try it here02:29
← adante left02:30
adante_ → adante02:30
woffle Anyone want to give me a fast repo to clone?02:30
jengelh it's not slow ..02:30
well or maybe it is02:30
their cpu however is not contended :)02:30
loops it is slow at times, but not like what woffle is seeing02:30
jengelh the guys should repack every once in a while02:31
loops yeah, that might be all it is02:31
jengelh Receiving objects: 17% (26807/157684), 13.71 MiB | 1938 KiB/s02:31
no, repacking is just for deltification speedup02:31
woffle :/02:31
→ imyousuf joined02:32
woffle Don't suppose you could send me that :P02:32
jengelh woffle: perhaps it's because you are, literally, near the end of the world02:32
woffle I've been downloading for two hours and only just passed 11MB02:32
loops woffle, better to just sort out what the problem is for you02:32
woffle Ok02:32
I'm on a source based distro, so if it's dependency related I can change it quickly02:33
jengelh git://dev.computergmbh.de/mesa feel free to clone02:33
loops well that'll be a good way to test02:33
jengelh whoa02:33
more object to push -.-02:33
sometimes I just wish --tags was default02:33
now it's done02:34
woffle I'll give it a shot, thanks02:34
loops sigh.. i'm not even transfering objects yet.. sheesh02:35
woffle jengelh, slow from there also02:35
loops woffle, it really is probably a local issue to you02:36
jengelh well, 'here' is germany, roughly one third around the world02:36
woffle yeah02:36
loops woffle, and i don't think it will be a git issue per se; more likely a network thing02:36
woffle No other connections are effected though02:37
→ IRSeekBot joined02:37
← IRSeekBot left02:37
jengelh perhaps some odd ISP that slows down port 9418?02:37
loops or everything but port 8002:37
woffle Thats what I was thinking02:37
Any way of testing that?02:38
jengelh hm02:38
yes02:38
git-clone http://dev.computergmbh.de/mesa/02:38
woffle ok02:38
jengelh use something like iptraf to monitor speed02:38
woffle "Cannot get remote repository information."02:39
jengelh perhaps I need to run update-serversomething02:39
woffle git-update-server-info02:39
jengelh try again02:40
loops holy crap my clone of mesa is still only at 18% indexing02:40
woffle O_o02:40
loops i already have a local clone anyway :oP02:40
woffle Downloading via httpd at 53KB/s02:41
loops yeah..02:41
jengelh srsly... :>02:41
→ Ademan joined02:42
jengelh modprobe tcp_illinois perhaps.02:42
or tcp_hybla02:42
← schacon left02:43
loops woffle, once you have a clone, you can repoint it at freedesktop.org; speed really won't be an issue just to grab new updates02:43
woffle yeah02:43
Thanks both of you02:43
← wagle_home left02:44
jengelh I removed all our svn repositories yesterday .. now 60% of all http requests from bots end up in 404 :p02:44
← tommorris left02:45
→ IRSeekBot joined02:46
← IRSeekBot left02:46
shenson → shenson_not_here02:49
woffle Whats with the bot?02:50
← robinr left02:51
→ replicant joined02:52
→ antares_ joined02:53
← imyousuf left02:54
woffle jengelh, error: Unable to find feb356caff82e996ba0b898c02383fdfa3effc5f under http://dev.computergmbh.de/mesa02:55
jengelh I already nuked it :(02:55
since you said it was equally slow02:55
woffle heh02:55
→ SuttoL joined02:55
woffle I let it run on httpd02:55
http.02:56
That's alright though02:56
bah, freedesktop doesn't support http02:56
→ IRSeekBot joined02:57
← IRSeekBot left02:57
jengelh should I regrab it?02:57
woffle If you wouldn't mind02:58
→ IRSeekBot joined03:01
← IRSeekBot left03:01
jengelh Receiving objects: 81% (127799/157684), 38.14 MiB | 325 KiB/s03:02
woha... they're slow!03:02
→ desatan joined03:02
Manoj jengelh: I could not get the sha1..sha2 stuff to work during rebase, but your technique of creating test branches off some points at the end of a segment, and judicious use of rebase --onto allowed me to clean the brancjes. Thanks.03:04
jengelh woffle: restored.03:05
woffle Thanks jengelh, I'll let you know when I'm done03:05
← replicant left03:05
→ bugfux joined03:07
bugfux my terminal crashed in the middle of git rebase -i, and now when i try to run it again it says Interactive rebase already started03:08
→ pumpkin joined03:08
← pumpkin left03:08
gram git rebase --abort and start over03:08
bugfux oh nice thanks03:08
← eikonos left03:09
→ schacon joined03:09
← antares_ left03:09
jengelh that is, if starting over is acceptbale :)03:09
sometimes there is just tooooo many patches to warrant a startover03:09
(for everything there is a solution, so do not fear..)03:10
→ IRSeekBot joined03:13
← IRSeekBot left03:13
→ imyousuf joined03:14
← QuercusMax left03:16
← a-priori left03:18
→ nkallen joined03:20
→ yec joined03:21
→ desatan_ joined03:24
woffle jengelh, I got it, thanks for doing that03:26
If you hadn't I'd still be downloading tomorrow :P03:27
jengelh ok03:27
← threeve left03:28
← Ademan left03:28
→ Ademan joined03:28
← desatan left03:31
← priidu_ left03:34
← SuttoL left03:36
→ IRSeekBot joined03:41
← IRSeekBot left03:41
blbrown with git-gui can you set passwords for ssh connections; git push ssh://xyz@url.com...03:42
→ replicant joined03:43
spearce blbrown: no, use an ssh key.03:44
fujin ssh connections would probably be handled with ssh-agent03:44
→ IRSeekBot joined03:45
← IRSeekBot left03:45
fujin so yeah, use a key03:45
blbrown nooooooooooo!!!!!!!!!!1103:45
hehe03:45
→ SuttoL joined03:46
← bugfux left03:47
→ IRSeekBot joined03:49
← IRSeekBot left03:49
← csc` left03:49
→ SuttoL_ joined03:52
← SuttoL left03:52
← replicant left03:54
← desatan_ left03:54
→ IRSeekBot joined03:57
← IRSeekBot left03:57
← otavio left04:01
← Modius left04:08
→ IRSeekBot joined04:08
← IRSeekBot left04:08
← chrisb left04:09
← vbgunz left04:18
← _bdiego left04:19
→ IRSeekBot joined04:20
← IRSeekBot left04:20
← xjrn left04:26
→ xjrn joined04:26
← queso left04:28
← niki left04:33
→ k0001 joined04:38
← bobesponja left04:39
k0001 what do i have to set to change my repo description and owner ?04:40
Ramune ?04:42
you mean on some git hosting service or something?04:42
k0001 from gitweb; description Unnamed repository; edit this file to name it for gitweb.04:42
spearce on gitweb? description is in the file $GIT_DIR/description; owner is set by the user who owns the repository in unix. best way to fix it is to use project.list to list the projects, and their owners.04:43
k0001 or is this some gitweb related thing... hmm04:43
spearce, thanks i'll check that project.list thing04:43
← pasky left04:45
← MDCore_ left04:47
→ MDCore_ joined04:47
← SuttoL_ left04:48
→ SuttoL joined04:48
← igorgue left04:52
← gitguy left04:53
← SuttoL left04:54
→ MDCore__ joined04:57
MDCore__ → MDcore04:59
← MDCore_ left05:03
schacon does anyone know off the top of their head how the walker works? does it cache the downloaded objects somewhere? if it gets halfway through the fetch of the whole tree, how does it figure out where it left off?05:04
← MDcore left05:05
→ MDcore joined05:05
gitster it does not figure out where it left off. A loose object is written in the object store as it is downloaded in its entirety.05:07
Suppose you have history A--B--C--D on the remote side, and you have up to A.05:08
→ IRSeekBot joined05:08
← IRSeekBot left05:08
schacon so if local side sees it has D, it continues to walk the parents until it gets to one with no parent?05:09
gitster The walker first learns that the remote tip is at D, and then downloads D (commit). It looks at D and learns its tree D^{tree}. It downloads D^{tree}. And so on. If it gets interrupted after downloading D^{tree}, and gets restarted, then05:09
schacon it doesn't assume it has everything below it?05:09
gitster it again learns (by running ls-remote) that the remote tip is still at D, looks at its object store and it has D (commit), so it does _not_ download D, but it still locally walks D and knows it needs D^{tree}.05:10
Which it already has, so it looks at D^{tree} and learns what blobs and trees it needs.05:10
← dsal left05:10
gitster This continues until it gets to the commit it points at its own ref (A).05:11
schacon gotcha05:11
makes sense05:11
gitster The important thing is that its own refs are _never_ updated until it finishes fully.05:11
So even though it got commit D and all trees and blobs that D records, it does not update its ref to point at D, because the commit D is _not_ complete (it lacks its ancestors).05:12
→ IRSeekBot joined05:12
← IRSeekBot left05:12
schacon very helpful - I was working on clone, so I didn't think about stopping when it gets to the sha it's current ref is pointing at05:13
gitster When the remote side has packs, the principle is the same. You try to fetch loose, and you fail, and then you slurp the pack .idx files that has list of objects their corresponding .pack records. And you download the .pack file to obtain the missing objects.05:13
You were working on clone of git?05:13
schacon yeah - pure ruby implementation05:14
← brixen left05:14
Ramune when you have A-B-C-D on the remote, and you have A-B-D on local -- but thepack file covers A-B-C-D, does git get the pack file? or can you get it to send just the loose objects?05:14
gitster The contract between the refs and object database is that what refs point at are complete (everything reachable from them are in the object store).05:14
schacon haven't had to decipher C in some time05:14
→ IRSeekBot joined05:15
← IRSeekBot left05:15
pygi spearce, ping05:15
spearce pong05:15
pygi spearce, I think you have a response at gsoc webapp ;)05:16
← spearce left05:16
→ spearce joined05:17
pygi spearce, got my last sentence?05:18
→ IRSeekBot joined05:18
spearce i think so. if not there's always the log. :-)05:18
← IRSeekBot left05:18
spearce i just read over your reply.05:18
makes sense.05:18
pygi ;)05:18
I agree that things have to be generalized, but we have to think carefully how to do that05:19
not everyone needs supermodules, and a complex set of permissions, and an admin panel and stuff like that05:20
→ halorgium joined05:20
→ IRSeekBot joined05:22
← IRSeekBot left05:22
→ bentob0x joined05:27
pygi spearce, if you have any suggestions, I would be more then glad to hear them05:29
will also talk with Johan today once again specifically about this05:29
spearce pygi: i got nothing additional right now. but i'll think about it and forward any comments.05:30
pygi thanks05:30
→ IRSeekBot joined05:32
← IRSeekBot left05:32
→ IRSeekBot joined05:33
← IRSeekBot left05:33
→ cedricv joined05:35
→ priidu_ joined05:36
← MDcore left05:36
→ MDcore joined05:36
→ \ask joined05:36
→ IRSeekBot joined05:38
← IRSeekBot left05:38
→ Dodji joined05:40
← schacon left05:43
→ Modius joined05:47
→ devogon joined05:48
→ |vovkav| joined05:51
← |vovkav| left05:52
→ BabelO joined05:54
← nkallen left05:56
→ nkallen joined05:57
→ johan-s joined06:00
→ IRSeekBot joined06:05
← IRSeekBot left06:05
→ IRSeekBot joined06:06
← IRSeekBot left06:06
→ sverrej joined06:08
→ namenlos joined06:09
← nkallen left06:11
← halorgium left06:11
← Ademan left06:11
← Manoj left06:11
← mithro left06:11
← Morphous left06:11
← Mikachu left06:11
← context left06:11
← gebi left06:11
← TuxCM left06:11
← petteri left06:11
← bartman left06:11
← MadCoder left06:11
← Illocution left06:11
← duncanm left06:11
← derRichard left06:11
← tmm1 left06:11
→ mithro joined06:12
→ Illocution joined06:14
→ context joined06:14
→ Manoj joined06:15
→ nkallen joined06:15
→ halorgium joined06:15
→ Ademan joined06:15
→ Morphous joined06:15
→ Mikachu joined06:15
→ gebi joined06:15
→ TuxCM joined06:15
→ derRichard joined06:15
→ petteri joined06:15
→ bartman joined06:15
→ MadCoder joined06:15
→ duncanm joined06:15
→ kergoth joined06:15
→ Helios- joined06:15
→ Vaxholm joined06:15
→ nutrimatt joined06:15
→ PerlJam joined06:15
→ FastJack joined06:15
→ tmm1 joined06:15
← cmarcelo left06:16
→ IRSeekBot joined06:20
← IRSeekBot left06:20
→ xenoterracide joined06:21
← nkallen left06:24
← miraculo_ left06:25
xenoterracide git is supposed to have really good support for binary version control right? better than svn?06:25
← fdr- left06:25
spearce everything in git is a binary. but git is better suited to source files than large binary files, as git uses deltas to compress files and text files tend to delta better than large binary files. but still, git can delta binary files if they don't completely change in a commit.06:27
→ ZhangSC joined06:27
→ |vovkav| joined06:28
← |vovkav| left06:28
gitster IOW, random binaries are not suitable for any version control, and we do not suck worse than svn ;-)06:28
xenoterracide they are all pdfs... actually I doubt many of them change past initial commit at this point06:31
I accidentally wiped out my entire svn tree06:31
but I didn't like the fact that it was using 5G for 2.1G of data. especially in a checkout (didn't bother me as much in the repo itself)06:32
← dsaxena left06:33
spearce well, a git checkout is usually a lot smaller than an svn checkout, even though we have the full history. :)06:33
packed PDFs that don't change will take up no more space than the same PDFs in a ZIP file, but is typically a little bit smaller (lower per-file overhead and maybe some delta compression actually usefully applies cross-files).06:34
← \ask left06:35
xenoterracide cool what's the best thing to look at for getting started with git?06:37
gitster two videos at youtube, then our tutorials.06:37
pygi isn't sure those videos won't scare new people06:39
pygi :p06:39
gitster Ok, perhaps I also should advertise FLOSS#19 ;-)06:39
gram which videos?06:40
xenoterracide link to videos/tutorials?06:40
gitster everything starts from http://www.git-scm.org/gitwiki/ ;-)06:40
← pygi left06:40
→ pygi joined06:40
→ fdr- joined06:41
← sverrej left06:41
gitster http://www.git-scm.org/gitwiki/GitLinks -- scroll down to Talks section.06:42
gram ah, yeah06:43
i enjoyed the linus one06:43
i should watch the others06:43
xenoterracide thanks06:45
thiago_home likes how he hesitates when he's about to call us KDE developers stupid :-P06:46
thiago_home the fact that he didn't means it's high praise :-)06:46
gitster he?06:47
spearce: I've commented on some of the applications. overall I am not so impressed but I think there are interesting ones.06:48
thiago_home Linus06:50
→ deavid joined06:52
jengelh it's because gnome is even worse06:53
and apparently he has not discovered anything 'better'06:53
xenoterracide that's because there isn't anything better06:53
unless you go minimalist06:53
spearce gitster: thanks, i just read through those comments and they are right on target, so i appreciate you taking the time to do it.06:53
xenoterracide fluxbox, etc06:53
pygi gitster, heh, what are the interesting ones? :)06:53
spearce i agree that some of the applications are uh, less interesting. especially the ones that literally just copied the project idea from the Soc2008Ideas page on the wiki and pasted it into the application. :-|06:54
jengelh xenoterracide: linus is not just some 0815 user -- therefore "minimalist" would be totally ok06:54
gitster 0815?06:54
pygi spearce, btw. just talking with johan06:54
xenoterracide jengelh: yes of course. but it's all preference anyway. I wouldn't doubt he's tried them.06:55
pygi he's comfortable in acting as my mentor, and is happy me working on gitorious whetever it is for gsoc or not ;)06:55
spearce pygi: awesome all around.06:55
jengelh gitster: "Average Joe"06:55
pygi spearce, just discussing how we could improve gitorious, perhaps with ticketing functionality06:57
spearce gitster: compared to last year's gsoc applications i think this years are generally better written, some quite a bit better written. however thus far nobody has offered up code, which was something that our top choices had done last year, and without us prodding them with a stick.06:59
pygi: connecting issue tracking systems to git has been a popular question, and often its "use Trac", but the git Trac plugin isn't considered mainstream yet from what i know. so having something more mainstream in a git centric tool like gitorious may be quite nice.07:00
pygi spearce, heh, only svn is really mainstream in Trac07:00
gitster what's the deal about github software? Are they very closed?07:03
spearce yea, i think they are trying to go commercial and charge projects hosting fees, etc.07:03
pygi nods07:03
pygi spearce, but they are saying that opensource will be free07:03
gitster sounds familier, eh?07:04
spearce yes, yes it does. :)07:04
pygi is away, but will probably update his application tonight, and poke you folks07:04
gitster Ah, we should prepare this year's April Fools.07:05
spearce heh. only got 20 hours or so remaining, depending on time zone. :)07:05
gitster The one we planned with Larry McVoy was sort-of fun.07:06
→ queso joined07:07
spearce oh, we should try to plan one with the SVN folks this year. :)07:07
← queso left07:08
→ nud joined07:08
spearce "git announces working directory change to match svn format" :)07:10
"git-gui adds support for CVS; creates migration tools to convert git to CVS"07:11
hmm, the last one wasn't as funny. oh well.07:11
gitster Not really.07:13
bd_ "git 1.5.5 adds automatic rebasing for better darcs compatiblity"07:15
→ Airwulf2k joined07:15
Airwulf2k hey07:15
← ben_h left07:15
Airwulf2k how can I copy a file with git like 'svn cp ...'?07:15
spearce just use cp07:15
then git-add the copy07:15
loops git/svn to merge.. All the beauty of the Git command line, with all the power of the svn central repo model07:16
gitster yuck.07:16
loops Best one ever was the perl/python merge where they used : to open a block and } to close it07:16
→ leethal joined07:16
Airwulf2k ah ok git doesn't track file histories or?07:17
spearce loops: "git to adopt svn's bdbfs as new storage format, disk usages expected to increase 1000%, git developers purchase hard disk maker stock"07:17
loops lol07:17
gitster so anybody tackling any outstanding issues?07:17
spearce Airwulf2k: we do, but we do it after the fact, so you don't need to explicitly tell us what you are doing.07:17
bd_ "git to move to InnoDB for backend storage"07:18
gitster Isn't Inno effectively dead?07:19
bd_ Better(?) than MyISAM07:19
→ ben_h joined07:20
gitster It would be interesting to have a curl library call to download foreign object from within read_sha1_file() -- that would be a good April Fools material.07:20
loops yeah.. that'd make the 3 guys who got it cry with laughter ;o)07:22
spearce we had a serious patch along those lines for the native git protocol not too long ago. but making it use RFC 1149 _only_ and not allowing local object caching would be a nice touch. :)07:23
→ FunkeeMonk joined07:24
→ Catfish__ joined07:26
→ MDCore_ joined07:26
bd_ spearce: I don't know, IP over percussive carrier ought to work just as well.07:27
spearce but its not covered by an IETF RFC so its probably not as widely implemented.07:28
→ athought joined07:28
→ aggieben_ joined07:28
bd_ Eh, push it through ISO. They'll approve anything these days.07:29
spearce yes, so long as it is fast-tracked.07:29
bd_ I hear they'll be adding the 'revolving door track' to their product lineup sometime in Q4.07:29
← mstormo left07:35
← johan-s left07:36
→ mstormo joined07:36
→ yborisov_ joined07:36
← LordCrimson left07:36
← MDcore left07:45
→ sverrej joined07:56
← ben_h left07:57
← aazar left07:57
← FunkeeMonk left07:58
← Catfish__ left07:58
← fdr- left07:58
← k0001 left07:58
← bryanl left07:58
← hliusv561 left07:58
← elijah left07:58
← SpComb left07:58
← ag left07:58
← masterdriverz left07:58
← ska-fan left07:58
← compbrain left07:58
← bla left07:58
← Ilari left07:58
← normalperson left07:58
← Thumper_ left07:58
← mojombo left07:58
← frsk left07:58
← Ramblurr left07:58
← bdiego left07:58
→ Ramblurr joined08:00
← dalin left08:00
→ Ilari joined08:01
→ compbrain joined08:01
→ ska-fan joined08:01
→ masterdriverz joined08:01
→ hliusv561 joined08:01
← mithro left08:02
← spearce left08:04
→ jenka joined08:04
→ Catfish__ joined08:05
→ normalperson joined08:08
← Ramblurr left08:10
← compbrain left08:10
→ compbrai1 joined08:10
→ _Ramblurr joined08:10
→ mojombo joined08:11
→ frsk joined08:11
→ SpComb^ joined08:11
→ Thumper__ joined08:11
→ fdr- joined08:11
→ k0001 joined08:11
→ bryanl joined08:11
→ bdiego joined08:11
→ Thumper_ joined08:11
→ bla joined08:11
→ SpComb joined08:11
→ Ramblurr joined08:11
→ compbrain joined08:11
→ ag joined08:11
→ elijah joined08:11
← compbrain left08:11
← k0001 left08:11
← Ramblurr left08:11
→ ag_ joined08:11
_Ramblurr → Ramblurr08:11
← fdr- left08:11
→ k0001 joined08:11
→ fdr- joined08:12
← bryanl left08:13
→ johan-s joined08:13
← Thumper_ left08:14
← ag left08:15
← eternaleye left08:18
← jim__ left08:19
← fdr- left08:20
← k0001 left08:20
← ag_ left08:20
← elijah left08:20
← SpComb left08:20
← SpComb^ left08:20
← bla left08:20
← Thumper__ left08:20
← mojombo left08:20
← bdiego left08:20
← frsk left08:20
→ eternaleye joined08:21
→ fdr- joined08:21
→ k0001 joined08:21
→ ag_ joined08:21
→ mojombo joined08:21
→ frsk joined08:21
→ SpComb^ joined08:21
→ Thumper__ joined08:21
→ bdiego joined08:21
→ bla joined08:21
→ elijah joined08:21
Airwulf2k does git has something like svn:keywords to write revision information into sourcecode?08:21
→ wagle_home joined08:23
→ markelikalderon joined08:24
mugwump Airwulf2k: yes, see the documentation about gitattributes08:25
however, there are good reasons to make that something that happens during the build phase of your project, not in the source08:26
tokkee Airwulf2k: http://git.or.cz/gitwiki/GitFaq#head-4a0afe71a2bb7734777a8b0b345e0308aefdbd4008:27
← cedricv left08:33
→ mithro joined08:35
← mstormo left08:37
→ mstormo joined08:37
→ wagle_ joined08:44
← wagle left08:45
← markelikalderon left08:56
→ railk joined08:57
→ ebel joined09:01
ebel → ebel_zombie09:02
← pete__c left09:04
← Manoj left09:04
→ Manoj joined09:04
→ pete__c joined09:04
← yann left09:04
→ yann joined09:10
← woffle left09:12
← MDCore_ left09:17
← johan-s left09:18
→ johan-s joined09:18
→ njero joined09:19
njero Hi #git. I have a repo locally. It has branch master and branch jeff/apzu. In jeff/apzu I have many glorious and wonderful changes that I want to merge to master. I say git checkout master; git merge jeff/apzu and it tells me already up to date... but the magical changes aren't copied. Any idea what I am doing wrong?09:20
dsop njero: what does git-cherry jeff/apzu master says?09:21
njero dsop: it shows 5 guids09:22
dsop, should I cherry pick those?09:23
dsop njero: no, but actually git merge or git rebase should do the job09:23
jengelh GUIDs!?09:23
njero er, are those not guids?09:24
uuids?09:24
dsop just try git rebase jeff/apzu master09:24
njero what have you?09:24
dsop they are sha1s09:24
just commit sha1s09:24
njero eek that makes more sense09:25
jengelh GUIDs are an evil thing from Windows and come in the form of XXXXXX-XXXXXXXXXXXXXX-XXXX or so09:25
← piclez_ left09:26
dsop njero: you might just wrong order of your parameters in git merge, but actually the git rebase stuff should do it09:26
njero dsop: tried the rebase and it told me "Switched to branch master, master already up to date"09:27
dsop hmm than the changes from jeff/apzu should already be in master09:27
njero yeah, but they don't seem to be09:28
uggg09:29
I did somethng stupid09:29
I just killed all of the magical changes09:29
← ZhangSC left09:35
← ruphy left09:36
dsop njero: what happend?09:37
njero dsop I did git checkout jeff/apzu; git merge jeff/apzu master (which I thought might push the changes in...) but instead it just deleted all of the files in jeff/apzu that were not in master09:40
I tried to git revet <that commit> which it claimed worked but did not restore the files09:41
*revert09:41
← nutrimatt left09:42
dsop njero: whats the status of git log? Does it reflect your changes? So that only the changes are not in the working copy?09:45
njero dsop: I see a commit for "Merge branches j/a and master into j/a09:46
But that is the only one listed09:46
dsop ??09:46
I cannot imagine the current state of your repository09:46
njero lol09:47
I blame myself09:47
I think I have a backup of the changes somewhere... should I kill this and start it over?09:47
this repo had the sad lifecycle of us learning git on it09:47
dsop actually I don't know if your repository might contain your changes. Well you can for sure start over if you want, but usually git doesnot forget commits and it's just about, how to get what you want from it.09:48
njero right, I would much rather learn how to fix it... git has been treating me really good up until this morning09:49
dsop njero: what does git log on your jeff/apzu branch give you? does it contain your changes?09:49
njero ahh, you mean somewhere down before I stared the stuff this morning?09:50
dsop: yes09:51
← saintdev left09:52
dsop njero: so whats the current problem, you have your changes in your branch jeff/apzu?09:52
but not in master?09:52
njero dsop, the changes are in neither branch now09:52
dsop hmm what does git fsck --unreachable give you?09:53
njero dsop, gives me nothing09:54
dsop or do you haev the sha1 id of your latest commit (with your changes)?09:54
maybe gitk --all helps you to visual what you have done09:54
njero looking09:55
njero thanks for all of the help dsop09:55
→ dkagedal joined09:56
→ rrregis joined09:57
→ pdherbemont_ joined10:00
dkagedal What is the difference between a "dangling commit" and an "unreachable commit" in the output from fsck?10:01
→ miraculo joined10:01
← pdherbemont left10:03
njero dsop: I am running out of time so I gave up and decided to merge the files from my backup again and let git go on10:04
Can I ask a couple branching merging general question to make sure I don10:05
dsop k10:05
njero don't mess up again?10:05
thanks10:05
→ smarnach joined10:06
SRabbelier Gah, yesterday, almost no msgs on the mailing list at all, I g tot bed, I get up, BAMF, 38!! msgs 0.o10:09
→ gerhardlazu joined10:11
→ pdherbemont joined10:11
gerhardlazu I'm trying to do a git-svn dcommit and all I'm getting is fatal: git-cat-file d8856b02245cd2754d012b7b45fcac34472358ca: bad file . How do I fix this?10:12
njero This is my current cheat sheet I have been working on http://pastebin.com/m32682b7c dsop, can you check lines 55-60?10:12
wagle_home dkagedal, i have no idea.. but try running git-log on both of them.. maybe dangling means parents dont exist...10:13
... unreachable would mean that the commit isn't on a git-rev-list from any branch or tag10:13
← railk left10:15
wagle_home note that git-gc (?) ---prune doesn't think that pseudo-branches like FETCH_HEAD pins anything down10:15
dsop hmm njero looks good for me, but I don't have time to take a deatiled look at the moment10:16
njero dsop, np thats good enough10:17
→ redondos- joined10:17
→ Catfish___ joined10:20
← pdherbemont_ left10:20
← redondos left10:20
redondos- → redondos10:21
→ SuttoL joined10:22
gerhardlazu I have re-traced my steps back, I have added 2 folders into my current project using git-svn fetch. When I'm trying to do a git-svn dcommit, it compains because the 2 folders have their own .git file. How do I go about this?10:22
xenoterracide silly question but does git work with winblows yet? it does work with non linux *nixes right?10:23
→ Industrial joined10:24
Industrial Git on windows is kind of anoying in that I can not clone inside a folder in windows explorer. I have to go one level up and then right click on the folder I want it in and then fill in the target directory myself (I actually want it one level down of that dir)10:25
this is with Git GUI10:25
xenoterracide oh... the last time I heard git didn't work on windows. but I'm just investigating it for use10:26
Industrial works fine here10:26
xenoterracide I chose to learn svn first. because of cvs's annoying prevalence.10:26
does git require the registry on windows? or is it portable (installable to flash drive)10:27
jengelh another buzzword?10:27
Industrial jengelh: ?10:28
jengelh http://en.wikipedia.org/wiki/Portable_application#Portability_on_UNIX-based_systems10:28
xenoterracide http://portableapps.com/10:28
jengelh well it seems like "portability" is a buzzword for the Windows environment...10:29
← Catfish__ left10:29
xenoterracide windows is gay but I have to tolerate it10:29
broonie jengelh: He means does it rely on having registry keys set up or can you just copy binaries around and run them without explicitly insalling on a given system.10:29
jengelh because as far as unixoids go, it's mostly portable all the way10:29
xenoterracide because firefox on windows uses the registry natively I can't put it on a flash drive10:29
portable apps has fixed this10:29
Ramune xeno: the mainline git relies on cygwin10:29
jengelh nonsense - i have pristine firefox 2.x running wihtout any registry fluff on windows too10:30
← Industrial left10:30
jengelh well, _almost_.10:30
xenoterracide jengelh: can you put it on a flash drive and carry it around (non portable apps version)10:30
jengelh Windows Firefox uses Windows's definition of $HOME, and hey, that's good :>10:30
because, well, on UNIX it does just the same, it uses $HOME/.mozilla10:30
xenoterracide I don't like it because I don't like my schools dumbass env10:30
Ramune xeno: there's a mingw port, but i never used it, so can't comment. it *should* be standalone.10:30
xenoterracide jengelh: yes but you can define home to be anyway10:31
anywhere*10:31
jengelh exactly10:31
xenoterracide it's pretty easy to install an app where you want and tell it $HOME where you want10:31
jengelh that's why "portable apps" seem so redundant10:32
xenoterracide not so easy on windows10:32
try installing mainline firefox on a flash drive10:32
take it to another syste10:32
see how it works. same for pidgin and a few others10:32
→ Yuuhi joined10:32
xenoterracide or anything on portable apps10:32
but it's offtopic10:32
I just defining my definition of portable10:33
fyi. my school doesn't use roaming profiles so the home directory is local. and it get's wiped every night10:34
Ramune xeno: on a side note -- you can run ff from a usb drive. google for it and you'll find the right link. :)10:35
but git itself doesn't rely on the registry. dunno about cygwin.10:35
or mingw.10:35
xenoterracide thanks.10:36
jengelh xenoterracide: looks like a glitch in firefox. It should probably use the HOMEPATH environment variable instead of using the registry for finding the home dir. BTW, GNOME has the same stupid problem. (Who would have thought.)10:36
← Catfish left10:38
Catfish___ → Catfish10:38
ebel_zombie → ebel10:38
xenoterracide jengelh: it's a glitch in thinking... and more than just gnome has the problem. I'm hoping kde-win doesn't end up with the same problem.10:39
but it's really only a problem if you are constantly on computers you can't control10:39
on my own system's I don't worry about it10:40
I only started looking for a solution recently. then looked at vm's but eventually found portable apps10:40
→ pasky joined10:40
xenoterracide of course I could probably tweak an installation for an hour to make these things work. but who wants to do that for 20 apps10:41
Ramune get it done once, then script it :)10:42
xenoterracide I think that's basically what portableapps has done10:44
they haven't changed enough firefox source to cause branding issues10:45
and mozilla is kinda anal about those things10:45
anyways all I needed to know was that I can do my work wherever I need to10:47
← imyousuf left10:47
← smarnach left10:51
→ artmi joined10:51
artmi if i use git+ssh, how can i change what `user-name' i want to use on the remote host?10:51
when i push something10:52
Catfish artmi: You're pushing to something like host.com:repo.git ?10:52
artmi yeah10:53
Catfish Try [email@hidden.address]10:53
artmi well ssh://host.com/...10:53
sbeyer hm I always do git push pkqs:foo/bar.git while pkqs is a host in my .ssh/config10:57
(a host I use very often)10:58
jengelh define a remote.10:58
sbeyer jengelh, of course ;)10:58
→ clear joined11:02
xenoterracide since git has no central repo I just make one in my home dir right?11:04
→ Catfish__ joined11:05
← Catfish left11:05
Catfish__ → Catfish11:05
jengelh every repo is a winner :)11:06
artmi i just can't seem to get my ssh key correctly it seems :/11:06
"No refs in common and none specified; doing nothing."11:06
jengelh on first push you need to specify master11:06
artmi oh :P11:07
that works11:07
i thought it actually "automatically" assumed master on first push11:07
jengelh yes I ofte think the same11:09
dkagedal wagle_home: no, if a parent was missing, I should get "missing commit"11:18
deavid Hello :)11:20
← kmap left11:20
deavid If I "push" a branch to a repo wich has a Working Copy of that branch..., then, the Working Copy will get updated too?11:21
broonie deavid: no11:21
deavid I know that I should do push only on a bare repo11:21
← dkagedal left11:22
deavid broonie: and what if I execute a "git reset --hard" on that repo?11:22
there is other way to update a remote WC?11:22
dsop hmm maybe hooks?11:23
deavid dsop: yes, I'll take a look to the hooks. But I don't know how I should do that manually.11:24
→ dkagedal joined11:24
deavid I'll try to make a test first and see what happens to the remote Working Copy11:25
→ kmap joined11:25
broonie reset should do the trick; you might wish to consider having a separate repository with a working copy for whatever you need that for, though.11:26
→ bryanl joined11:26
deavid I have a separate repository, wich is a Bare repo. But I need that, because I want a folder automatically updated with the source code.11:28
← pete__c left11:31
jengelh sounds like a contradiction11:31
deavid :-P11:32
for example11:32
think in a web server with php code11:32
→ Fullmoon_ joined11:32
deavid I can place a Git repo, as the php code for apache11:32
I want do: "push" and get the Git folder updated11:33
I've another git that is a bare repo.11:33
→ langenberg joined11:34
→ charon joined11:34
xenoterracide can I cancel a commit once I've started it? I just realized I have svn cruft in this dir that shouldn't be in the repo11:37
it's prompting me for commit message11:37
dato xenoterracide: quitting the editor without saving the file should abort the commit11:38
Mikachu or saving an empty file11:38
deavid In git commit, if you don't give message, it shoud abort the commit.11:38
dato xenoterracide: if you already saved it, you can just let git commit, and then use git-reset11:38
← clear left11:38
Fullmoon_ I always get a bunches of changes because filemode changed upstream, typically from 755 to 644 or backwards. I dont care. Can I ignore this?11:38
dato oh, empty, cool11:38
← anttih left11:38
xenoterracide thanks11:38
→ clear joined11:38
xenoterracide I didn't want to be running cleanup this early11:38
:P11:38
→ tomtt joined11:41
tomtt i'm kinda curious how the global uniqueness of the sha1 is guarantueed. any pointers?11:42
xenoterracide is the # commented stuff in the commit kind of a template for a commit message?11:42
Mikachu tomtt: it isn't guaranteed, it's just very very likely11:43
deavid tomtt: http://www.schneier.com/blog/archives/2005/02/sha1_broken.html11:43
xenoterracide tomtt: there is no hash (that I know) that guarantee's global uniqueness11:44
and if they do they are probably lying11:44
or there is more too it than that11:44
tomtt Mikachu: ah that would explain it :). So is there a small chance that two commits in a repos do have the same sha111:44
langenberg I'm trying to deploy a rails application on a host that also has git installed. It's git version 1.4.4.4 while deploying I get an error: 'fatal: Needed a single revision'. Has there been an change between 1.4 and 1.5 that changes the behaviour of 'git fetch origin'?11:44
deavid tomtt: maximum, a probabiliti of 1 / (2^69)11:44
Mikachu tomtt: very small, yes11:44
← sbeyer left11:45
tomtt deavid: not 1 / (2^320)?11:46
so what is git's mechanism if this would happen: will that sha1 then refer to the most recent one?11:47
← _esac left11:47
deavid tomtt: There are some SHA-1 collisions at 2^69, or that says in this url: http://www.schneier.com/blog/archives/2005/02/sha1_broken.html11:47
jengelh the sha is not that strictly for cryptography11:48
deavid there aren't any hash that guarantees 2^X, where X are the number of bits of the hash. That is impossible.11:49
tomtt deavid: i see :)11:49
← clear left11:50
xenoterracide didn't I say that? in a less sophisticated manner11:51
gerhardlazu after I do a commit, can I go back and change that commit?11:53
deavid xenoterracide: yes; basically is the same thing. ;)11:54
xenoterracide so what about all this stuff that's commented out in this commit message? should I uncomment it? is it supposed to be a template? or is it just for my knowledge11:55
jengelh knowledge is power.11:55
hence, with git you have the power.11:55
dato xenoterracide: git will ignore commented lines11:56
xenoterracide curses at synergy always on wrong screen11:56
xenoterracide dato: I know that I'm just wondering if it is intended as a template for messages11:57
deavid gerhardlazu: Have you seen the "amend commit" option on git GUI ? :-)11:57
dato xenoterracide: ah, sorry. it's just fyi11:58
gerhardlazu deavid: no, not really. is that the gitk?11:58
→ Man-erg joined11:58
gerhardlazu deavid: I'm on OS X btw11:58
← ebel left11:58
gerhardlazu can't I do that through the terminal?11:59
← borior left12:02
xenoterracide can git do automagic commits?12:06
→ Sho_ joined12:06
deavid gerhardlazu: I never seen Git on OS X; In Linux & Windows there is a git-gui command to make commits easier12:06
← miraculo left12:07
deavid but, you can see the man page for git-commit, I think there is some help about --amend option12:07
but i never used that option on command line12:07
gerhardlazu deavid: yes, but that is only for the last commit. I'm looking right now into rebase --interactive12:07
← langenberg left12:09
deavid gerhardlazu: ups. Then I don't know how to do that.12:10
→ QuercusMax joined12:10
gerhardlazu it's allright, google to the rescue :)12:10
dato xenoterracide: automatic how?12:10
→ jim joined12:11
Thumper__ → Thumper_12:13
xenoterracide dato: like I save a file it gets commited and I don't think about it12:15
obviously I might need a little scripting...12:15
→ harinath joined12:17
→ antares joined12:21
← QuercusMax left12:21
→ BabelO__ joined12:22
← charon left12:23
← Airwulf2k left12:25
← BabelO left12:27
← antares left12:28
→ QuercusMax joined12:34
→ kukks joined12:38
→ chrisb joined12:41
← QuercusMax left12:45
→ deavidsedice joined12:53
→ roylee17 joined12:58
← tomtt left12:59
→ lcapitulino joined13:02
→ theefer joined13:07
→ [tla] joined13:08
→ CIA-50 joined13:11
← Fullmoon_ left13:11
→ aggieben joined13:12
theefer Hello Git people13:13
If that's of interest to anyone, I gave a short presentation on Git at the latest BarCamp in Lausanne, and I put up my slides here:13:14
http://exodus.xmms.se/~theefer/git-presentation-barcamp.pdf13:14
It's free for use, etc, naturally. (evil keynote sources available on demand, i.e. I don't have them here atm)13:14
→ rafl joined13:18
mstormo theefer: "+Linus" "-Linus" :-D13:18
ehamberg Heh, made me smile as well :)13:18
theefer Does anyone disagree with this? ;-)13:18
mstormo theefer: You were missing "+Junio"13:19
[tla] Hi. I have svn cloned a repo and all the history is in .git/. I'm missing the vital piece of information on how to check it out! The last revision from svn was 69. If I try "git checkout -b master" I get the error: "fatal: You are on a branch yet to be born". How do i checkout the data into git?! Thx...13:19
mstormo IMO ++13:19
yborisov_ → LordCrimson13:20
mstormo theefer: Very nice presentation though, to explain the inner workings13:20
theefer mstormo: I probably missed a lot of things and people, indeed. Thanks.13:20
jengelh were not you asking for email addresses once upon a while?13:20
→ langenberg joined13:21
→ aazar joined13:24
← LordCrimson left13:24
[tla] ^help pls... is this because my import from svn has no valid HEAD?13:26
Ilari [tla]: what does 'git branch -r' list?13:28
[tla] Ilari: git-svn13:29
→ Fullmoon_ joined13:29
Ilari [tla]: 'git checkout -b master git-svn'.13:29
← aggieben_ left13:30
← namenlos left13:30
→ destruct_ joined13:31
→ etnt joined13:34
→ juanjoc joined13:35
← mithro left13:36
← mstormo left13:37
→ mithro joined13:37
← deavidsedice left13:37
→ mstormo joined13:37
→ cmarcelo joined13:39
→ aroben joined13:40
→ charon joined13:41
→ raxor joined13:42
← destruct left13:42
→ ben_h joined13:42
← duderonomy left13:42
→ aroben_ joined13:43
→ dbtid joined13:44
← aroben_ left13:45
dbtid how can i ask git to tell me what the various revisions in my database are?13:45
→ aroben__ joined13:45
dbtid i did a pull, then a merge, and i'm wondering what the state of a particular file is.13:45
← aroben left13:45
aroben__ → aroben13:45
← eternaleye left13:46
cehteh dbtid: git blame ?13:47
dbtid really?13:48
kind of cynical :)13:49
← njero left13:49
→ G_SabinoMullane joined13:51
Ilari dbtid: Also 'git log -- path/to/file' lists commits that modify that file. s/git log/gitk/ to see it in graphical form.13:52
dbtid thanks13:52
Ilari dbtid: You can also list multiple files there, and they are properly unioned.13:52
dbtid out of curiosity, are all the commits stored in the local database when the fresh version is retrieved, or does git go back to the repository to ask?13:53
Ilari dbtid: Copies are stored in local database.13:53
gerhardlazu I think I've messed my git repository :( I've done a hard reset that took me 1 step back. Can I go forward that step?13:54
ag_ → ag13:54
dbtid i thought so13:54
destruct_ → destruct13:54
Ilari gerhardlazu: 'git reset --hard HEAD@{1}'?13:54
gerhardlazu Ilari: what will that do?13:54
Ilari gerhardlazu: Do hard reset to point that was current before last operation.13:55
gerhardlazu Ilari: phew, YES, that did it13:55
I almost panicked there :)13:55
though I've lost half day's work13:56
ok13:56
this is my problem13:56
3 commits ago13:56
I've done something stupid13:56
I want to go back there, and erase that13:56
Mikachu man git-rebase, see 'interactive rebase'13:56
gerhardlazu because I'm doing an git-svn dcommit13:56
danderson (wouldn't be lost anyway - you could use the forensics commands to find the commit object, and repoint the branch head)13:56
gerhardlazu and SVN craps out13:56
→ threeve joined13:57
gerhardlazu Mikachu: when I do git-rebase --interactive HEAD-4, I get fatal: Needed a single revision13:57
Mikachu yes?13:58
~, not -13:58
gerhardlazu git rebase --interactive doesn't do anyting13:58
ohhhh13:58
→ threeve_ joined13:58
gerhardlazu Mikachu: that's it! cheers bro ;)13:58
dbtid Ilari: i have 2.6.25-5, and i'm looking at updating to 2.6.25-rc7. there's a particular file i'm looking for changes in... can i do that easily?13:59
Ilari dbtid: 'git diff 2.6.25-rc5 2.6.25-rc7 -- path/to/file' (or whatever the tagnames are).14:00
dbtid wonderful14:00
i haven't retrieved rc7, yet.14:01
does that need to be done, first?14:01
Ilari dbtid: 'git fetch origin'14:01
dbtid so that will... ?14:01
gerhardlazu Mikachu: I've also done a git commit --amend and I do not want the last commit in my logs. how do I delete it?14:01
Ilari dbtid: Update the local mirror of remote repository nicknamed 'origin' (the default set up by git-clone) and fetch any tags pointing to revisions fetched.14:02
dbtid wonderful14:02
then when i want to merge everything i just retrieved, i'd do 'git merge'?14:02
← daved_ left14:02
Ilari dbtid: Yes, 'git merge <revision>'.14:02
→ McInnes joined14:02
dbtid ok14:02
gerhardlazu Mikachu: found it. removing that line in the rebase will drop that commit. that is seriously cool mate. SVN only dreams about this :)14:02
← threeve_ left14:03
dbtid grr 'bad revision 2.6.25-rc5'14:04
Mikachu dbtid: v14:04
dbtid oy14:04
ty14:04
Mikachu also, git tag -l14:04
dbtid right14:04
can git diff work on a directory?14:05
Mikachu try14:05
dbtid ok14:05
← leethal left14:05
dbtid yeah, duh, sorry; i guess i should have asked if any flags were necessary14:06
thanks to all who answered questions!14:06
← vmiklos left14:07
etnt how can I quiet git-status from telling me about non-tracked files ?14:10
Fullmoon_ → Fullmoon|mobile14:10
shenson_not_here → shenson14:11
roylee17 how to clone a repo's with it's remote repo? So that I can 'git log origin/origin/branchX', for example ?14:11
→ antares joined14:11
← antares left14:12
Ilari roylee17: There is no direct way, but you could copy the remote branches manually (using git fetch with suitable arguments)...14:13
Mikachu i think you can do something like git fetch origin 'refs/remotes/*:refs/remotes/origin/remotes/*'14:13
roylee17 thx alot , that's what I want14:14
← threeve left14:15
roylee17 does git already has an command or option to to the same things for short?14:16
← pflanze left14:17
dbtid ok, i did 'git merge v2.6.25-rc7' and it told me 'path/to/file: needs update'14:20
does that mean there's a conflict with what i changed and something that came in?14:20
or just that i'd changed it but hadn't done an add?14:20
← SuttoL left14:21
Mikachu the latter14:22
dbtid ok... to generate diff then do i just do 'git diff path/to/file'?14:22
yes14:22
→ ceez joined14:23
dbtid ok, so say i made some logging changes and i just want to revert the file... git X path/to/file... is it checkout? fetch?14:23
→ niki joined14:24
→ chris2 joined14:26
→ threeve joined14:27
→ Tidus2 joined14:28
[tla] Ilari: thx for help. the branch checked out ok but i don't have the latest revision of all the files. I "know" this because I checked out the latest rev from svn (using svn) and when I compare the svn and git repos I see that later files in svn are not present in git. Any ideas what could cause this?14:30
→ vmiklos joined14:31
Ilari [tla]: 'git svn fetch' and 'git svn rebase'?14:32
dbtid Ilari: how do i revert a file?14:33
Ilari dbtid: 'git checkout <version> -- path/to/file' to revert to practicular version, 'git checkout -- path/to/file' to revert to last staged version.14:35
← k0001 left14:35
dbtid what's 'staged'?14:36
last pulled?14:36
Ilari dbtid: as in 'git add'.14:36
dbtid thanks14:36
→ matt_n_at joined14:40
matt_n_at i have a remote repository and a local branch tracking changes. i branched off of the local branch and checked in a bunch of changes. the i switched to my tracking local branch and git pull'd. i switched to my work local branch and git rebase tracking. that killed all my checkins. can anyone help me get them back?14:43
→ vbgunz joined14:43
→ krh joined14:45
→ intripoon joined14:46
→ FunkeeMonk joined14:46
→ aortona joined14:46
→ bobesponja joined14:46
intripoon hi14:47
where can i specify the language used in the git gui?14:47
Ilari intripoon: It is determined by the standard i18n settings (LC_ALL, LC_MESSAGES, LANGUAGE, etc...)14:48
intripoon I'm on windows ...14:48
Ilari intripoon: No idea then. :-/14:49
intripoon hmm14:49
→ EmilMedve joined14:49
→ ebel joined14:49
[tla] Ilari: interesting ... no changes are being made to the svn repo and yet running repeated "git svn fetch" keeps pulling new data from svn. Why did the initial git svn fetch not get all revisions in one go?14:50
aortona anyone know of a decent idiot proof git gui wrapper for os x?14:50
→ Mitz` joined14:50
Ilari [tla]: No idea...14:50
aortona it's not for me...14:50
← johan-s left14:51
Ilari aortona: Wrapper?14:51
aortona well, front-end14:51
← matt_n_at left14:51
aortona whatever you want to call it14:51
Ilari aortona: If you mean GUI for git, git-gui.14:51
aortona yes, i looked at that, it seems to be missing quite a few features14:51
unless i just can't see them14:51
intripoon My gui always starts in german. While I understand german, I don't know what git commands the translated commands refer to. Like what is the pull equivalent?14:52
aortona tagging, for example14:52
checking out old revisions14:52
Ilari aortona: You can check out old revisions...14:52
Also, git-gui is missing rebase support... There's also qgit, but I never figured out how that works...14:53
← Mitz` left14:54
← ebel left14:55
aortona Ilari: i can't seem to work out how to check out old revisions14:55
there's an option to specify "Revision expression" in branch -> checkout14:55
but it doesn't seem to accept anything i give it (nor is it actually clear what it means by "revision expression")14:56
i tried the first eight or so characters of a commit id and just got a strange tcl/tk error14:56
Ilari aortona: First 8 chars of commit ID should be valid for smaller projects. For larger ones, you need more than that.14:57
aortona this is just a small test project with four or five commits14:58
Ilari aortona: What was the error message?14:58
aortona let me trigger it again...14:58
← langenberg left14:58
← mithro left14:59
aortona http://rafb.net/p/6apB5w77.html15:00
Ilari aortona: I just tried that, and it doesn't throw any errors (one warning notice yes).15:00
aortona it seems to leave the program in an inconsistent state too15:01
i don't appear to be able to do anything other than quit15:01
it's fine upon reopening15:01
→ KirinDav joined15:01
Ilari aortona: What it shows as current branch after restarting?15:01
aortona HEAD15:02
so it has actually checked out the revision correctly15:02
maybe just a simple bug somewhere15:02
Ilari aortona: What git version, what git-gui version?15:02
aortona the history shows the latest revision to be the one i selected15:02
← raxor left15:03
Ilari aortona: Git-gui version can be seen in about box, for git version, 'git --version'.15:04
aortona git-gui version 0.9.2.7.g3bae15:04
git version 1.5.4.315:04
→ piclez_ joined15:04
aortona sorry, i have to run across the lab to execute commands on that machine at the moment15:04
recabling...15:04
gerhardlazu after some re-basing, I'm trying to commit but am getting this fatal: cannot lock HEAD ref15:07
any ideas?15:08
→ duderonomy joined15:08
→ kumbayo joined15:09
← mattf left15:10
→ sbeyer joined15:11
← Tidus2 left15:14
shenson → shenson_not_here15:15
venox how can I merge two branches while keeping a file?15:16
I mean, I have edited this file on both branches, but while merging the two branches, I want to keep the file from the branch I'm working on15:17
→ schacon_ joined15:18
gerhardlazu venox: git stash15:19
you can also name your stashes15:20
git stash BranchA-File etc.15:20
→ saintdev joined15:21
← McInnes left15:24
← FunkeeMonk left15:24
← aazar left15:26
venox gerhardlazu: but git help stash doesn't say anything about creating a stash of just one file15:27
gerhardlazu venox: yes, you're right, it will save all your uncommited changes15:28
you can commit everything except that file...15:28
numbah resolve the conflict with git checkout OtherB -- file15:28
and add it, commit15:29
venox hmm, the problem is that I need to keep the file with different contents on each branch15:30
(forever!)15:30
→ gluegadget joined15:33
← charon left15:34
gerhardlazu venox: doesn't git do manual merges as well? you could choose what not to merge maybe...15:34
← sverrej left15:35
gerhardlazu I have a problem here with git-svn. I have 2 submodules, when I do git-svn dcommit, I'm getting fatal: git-cat-file d8856b02245cd2754d012b7b45fcac34472358ca: bad file15:35
→ shoe joined15:36
← Yuuhi left15:36
Ilari gerhardlazu: What does 'git-cat-file -t d8856b02245cd2754d012b7b45fcac34472358ca' give?15:37
gerhardlazu unable to find15:37
← shoe left15:37
→ saintdev_ joined15:40
→ jlh joined15:41
← krh left15:41
→ eqvinox joined15:41
→ seanius joined15:42
→ leethal joined15:43
← garrin left15:45
→ aroben_ joined15:45
→ Oejet joined15:50
jlh In a fresh git repo (with or without commits), running "git fetch" fails with:15:52
fatal: 'asd': unable to chdir or not a git archive15:52
fatal: The remote end hung up unexpectedly15:52
← aroben left15:52
aroben_ → aroben15:52
jlh this is quite confusing. where did it try to chdir to? what is not a git archive? and there exists no remote end, since the remote doesn't exist15:52
← saintdev left15:53
→ McInnes joined15:54
Pieter yeah... that's what it says15:55
asd doesn't exist15:55
→ DenisG joined15:55
DenisG hi there15:56
I've got two git repositories without a common ancestor15:56
I'd like to merge them15:56
is this possible with git?15:56
Pieter yes15:56
saintdev_ → saintdev15:56
Pieter just merge them15:56
DenisG how?15:56
jaalto CAn git tag names be treated as commit ids. Like if I use HEAD~1, I can use <tag name>~1 ?15:57
Pieter git pull /path/to/repo master15:57
DenisG okay15:57
I did that already15:57
Pieter jaalto: yes15:57
DenisG but the directory is empty15:57
jaalto Pieter, ok - thanks15:58
Pieter DenisG: which directory?15:58
DenisG the one into which I pulled15:58
Pieter why did you pull it then?15:58
DenisG (it was empty before, and is still empty now)15:58
the issue is this: I have a repository to which I can't commit15:58
which is svn *cough*15:59
so I checked it out using svn-git and worked on it15:59
now I would like to create a new svn repository from my changes15:59
I created a new svn repository and checked out from the new one using git-svn, too16:00
SRabbelier gah, gsoc deadlien extended, more room for competition! :P16:00
DenisG and now I want to get the changes from the old git-svn repo to the new git-svn repo so that I can commit them into the new svn repo16:00
;(16:00
eqvinox SRabbelier: the gsoc deadline always is extended :)16:00
they do that on purpose16:01
to get the slackers in a panic16:01
SRabbelier eqvinox: mhhh, heh16:01
← ben_h left16:01
DenisG anyways. Why is the new git repository empty now?16:01
eqvinox but, er, where do you see that? faq still says 6h till deadline16:01
Pieter DenisG: if you want to create a new repo with the same contents, why not import the same remote?16:02
and then rebase it?16:02
DenisG hm16:02
there are changes in the new repo that aren't in my local git-svn16:02
eqvinox SRabbelier: where'd you see the deadline extension?16:03
→ dsaxena joined16:03
SRabbelier eqvinox: #gsoc16:03
and the announce mailing list16:03
→ krh joined16:03
DenisG meaning I would have to merge manually, which I don't want16:03
eqvinox ah... thanks :)16:04
→ antares joined16:04
Pieter DenisG: if you want to keep that history in your new svn, you'll probably have to rebase anyway16:04
DenisG hm, okay16:04
but I don't want to merge any changes16:04
← piclez_ left16:04
DenisG manually, that is16:04
← antares left16:05
DenisG I'd prefer to just overwrite the files16:05
is there an option for that in git?16:05
← athought left16:06
→ yann_ joined16:06
McInnes DenisG: There is the "ours" merging strategy.16:06
See man git-merge16:07
compbrai1 → compbrain16:07
DenisG hm, okay16:07
thanks so far16:07
← yann left16:07
→ ronny joined16:08
ronny yo16:08
→ sverrej joined16:08
ronny are there any plans for tracking merges o git-svn ?16:08
← sverrej left16:08
→ maio joined16:11
maio hi, is there some tutorial how should one setup company's hosting of git repository(es)?16:12
DenisG man is this ugly16:13
I can't easily clone the svn repository16:13
maio git-svn clone URL16:13
sounds easy enough :)16:13
DenisG nah, I need a new svn repo :(16:14
McInnes svnadmin dump/load?16:14
DenisG I don't have shell access to the old svn access16:14
McInnes svnsync?16:14
DenisG and the machine on which I want to create the new svn repo does not have access to the old repo, since it's in a protected network16:15
McInnes svnsync to a machine in that network, then svnadmin dump?16:15
DenisG so I have to copy the repo to my laptop and from there to the new server16:16
yeah16:16
ugley16:16
← Fullmoon|mobile left16:17
→ DenisG_ joined16:20
→ Fullmoon_ joined16:21
gerhardlazu I have this .dotest directory in the root, how do I resolve it? I think that's there because of the conflict..16:23
← jenka left16:23
Mikachu rebase and am use the .dotest directory16:23
← nud left16:24
gerhardlazu ok, I've tried a rebase16:24
and it said to remove the .dotest if I'm not in the middle of another rebase16:25
→ alley_cat joined16:25
gerhardlazu Mikachu: I seem to have a problem though... git-svn dcommit doesn't like submodules AT ALL16:26
any ideas how I can go about it?16:26
Mikachu no surprise there16:26
nope16:26
→ SuttoL joined16:26
gerhardlazu crap16:27
← SuttoL left16:28
→ vintrepid joined16:31
→ mattf joined16:32
→ charon joined16:33
mattf i just did a git-fetch and discovered 3 branches i was tracking have changes. i can run "git diff --stat BRANCH origin/BRANCH" and see the differences for a single branch. is there a way, in a single command, to see all the differences in all branches?16:35
← Fullmoon_ left16:36
← KirinDav left16:36
← DenisG left16:36
→ ebel joined16:37
→ shafty joined16:41
→ DenisG joined16:44
yann_ → yann16:46
→ QuercusMax_ joined16:48
← ronny left16:50
← Oejet left16:51
← DenisG_ left16:53
→ markelikalderon joined16:54
→ p4tux joined16:54
→ joevandyk_ joined16:56
← jlh left16:57
gerhardlazu if I add a folder to gitignore, why is git-svn dcommit still trying to commit it?16:58
tpope dcommit just pushes the commits you have to the server16:59
it doesn't retroactively edit them to respect your very latest ignores17:00
gerhardlazu tpope: I have created the .gitignore and added the ignores first17:00
and then added the submodules17:00
but it's still trying to commi them17:00
commit*17:00
tpope so you're mixing submodules and svn?17:00
good luck buddy17:00
also, gitignore only applies to untracked files17:01
it won't retroactively stop tracking files you already added17:01
nor will it cause changes to tracked files to be ignored17:01
gerhardlazu hmmm. Is there no way of having submodules under git but preventing them from being commited when I do a dcommit?17:03
← yec left17:04
→ jim_c joined17:05
→ KirinDav joined17:05
→ timocratic joined17:09
← deavid left17:09
→ arekm joined17:11
arekm hello, how to update .git/description on a remote server (repository was pushed there initially) without shell access?17:12
gram pushed how?17:14
ie what protocol did you use instead of ssh17:15
gerhardlazu how do I resolve a git conflict?17:15
arekm gram: ssh but it (savannah.gnu.org) doesn't allow interactive shell17:16
gram well, i would think you should still be able to use that to overwrite files17:16
like: ssh host "echo foo >project.git/description"17:17
or maybe they support sftp?17:17
tpope not with git-shell, but scp/sftp might work17:17
might17:17
gram hmm... didn't know about git shell17:18
→ asdx joined17:21
→ Eludias joined17:24
fonseca tokkee: Just saw "pending upload mail" .. FYI, tig-0.10.1 has a stupid bug fixed in commit b1744cbe9a17:24
← redondos left17:24
Mikachu fonseca: when i say 'tig log' i get "`tig log' has been deprecated", but what should i use instead?17:26
← Manoj left17:26
Mikachu git log|tig doesn't work at all17:26
→ Manoj joined17:26
fonseca Mikachu: I didn't think anybody used it. ;)17:26
Mikachu: I can de-deprecated17:27
Mikachu what do you use?17:27
fonseca I just use plain `tig`17:27
Mikachu if i run tig and press L, it gets the diffstat for everything which is a bit slow17:27
fonseca Yes, true.17:27
Mikachu i used to have tig as the git pager, and when using 'git log' it would look exactly like tig log17:28
but it seems tig doesn't like stdin anymore17:28
but tig -h says you can do it17:28
fonseca You can configure it with TIG_LOG_CMD:17:28
TIG_LOG_CMD="git log" tig17:28
Really?17:29
Mikachu really what? :)17:29
" or: tig < [git command output]"17:29
if i have to guess17:29
fonseca Hmm, I might have broken check for whether inside a git directory or not.17:29
Mikachu if i say git log|tig i get the same screen as just running tig17:29
fonseca I was refering to handling of stdin.17:30
Mikachu so did i answer everything?17:31
→ markelik_ joined17:32
fonseca Mikachu: Yes, I will try to look into it.17:33
s/try to//17:33
Mikachu okay17:33
"there is no try, only do"17:34
fonseca Heh, sandt.17:34
BTW, do you know a good cafe with internet access in Lund?17:35
Mikachu no, sorry17:35
← Fullmoon left17:39
← EmilMedve left17:39
aroben I can't seem to run the tests on next17:40
* error: cannot run git init -- have you built things yet?17:40
(yes, I have built things)17:40
guess I'll try on master17:40
← vbgunz left17:41
aroben same thing happens on master17:41
this is on Cygwin, btw17:41
anyone running into this?17:41
← arekm left17:49
← sbeyer left17:51
← markelikalderon left17:51
← ebel left17:52
BabelO__ → BabelO17:52
fonseca Mikachu: tig's pager mode should be fixed now ...17:53
Mikachu nice work :)17:53
aroben works fine on OS X17:54
oooh17:54
I know what's wrong17:54
I have a space in my path17:54
→ sbeyer joined17:55
Mikachu ah, i have the repo.or.cz url as origin17:55
yeah it works now17:56
fonseca I will dedeprecate tig log also, if you use it.17:57
Mikachu i will probably use it less now that git log|tig works17:58
fonseca OK.17:58
[tla] Ilari: thx 4 help today :) l8r17:59
← [tla] left17:59
fonseca Mikachu: Reminds me, thanks for reporting this.18:00
Mikachu sure, thanks for fixing it :)18:01
→ eikonos joined18:01
← markelik_ left18:01
→ ruphy joined18:03
→ no_maam joined18:05
no_maam hi18:05
any good documentation, which gives you some advices for best practices when using git, and having used svn before?18:05
Mikachu no_maam: git.or.cz18:06
no_maam there are some documents which describe which commands you have to use in git instead of svn commands, but I couldn't really find something which describes how git is used most efficiently18:07
DenisG so I want to merge a git-svn repo with another git-svn repo18:10
what would i use for that?18:10
git-push or git-pull destroy the svn commit history, it seems18:10
Ilari DenisG: AFAIK, can't be done.18:10
→ eternaleye joined18:11
Ilari no_maam: One fairly important thing to is to know when to rebase and when to merge, as both have their upsides and downsides.18:11
DenisG wtf18:11
:(18:11
gram speaking of which, when is it better to use rebase vs merge18:12
DenisG this just sucks18:12
gram i just end up rebasing all the time because i'm the only developer on my projects18:12
→ johan-s joined18:15
Ilari gram: If others have seen (even partially) both branches, merge, otherwise rebase. That's the rule of thumb.18:16
Lemurnomicon rebase is rewriting changes as if they occured on top of a different set of history18:17
(and you don't want to muck with published history)18:17
Ilari And the history others haven't seen is rebased on top of what others have seen.18:17
← McInnes left18:18
Ilari I.e. Take "does not fast forward" errors seriously, and don't just force the operations without thinking twice.18:18
→ DenisG_ joined18:19
→ njero joined18:20
gram ok, good, that's been my policy so far18:21
← gerhardlazu left18:21
Ilari The basic workflow when working on projects is to branch topic branches that each are used to develop one feature, and kept up to date by rebasing (but not published, except maybe as patches, or with very strong warnings attached).18:21
When they are finished, they are rebased on top of main branch for one final time, and then merged into main branch (merge will be trivial). Then the topic branch is often deleted.18:22
→ nbrunson joined18:23
Ilari When receiving pull request, first fetch the branch, look if it is acceptable, and if it is, merge it.18:23
gram right18:23
i've tried to do the topic branch thing but haven't found any really good information on how others do that18:24
i end up fumbling trying to keep multiple repositories in sync with things being rebased all the time18:24
Ilari Only really obivious stuff should be committed directly to main branches. Also, if you have multiple main branches, you should do changes to stabilest branch and merge towards unstablest one.18:24
→ Beket joined18:25
← chris2 left18:25
gram multiple main branches? like a master and a next?18:25
Ilari gram: Actually, in git.git it is maint, master and next (then there's pu as well).18:26
shenson_not_here → shenson18:29
← z3ro left18:29
gram maint is for patch releases and such?18:31
Ilari gram: Current maint corresponds to 1.5.4.X versions.18:31
gram i new thought stuff was commited on next and then master was fast-forwarded when everything is stablish18:31
s/new//18:31
Ilari gram: The changes that apply to maint are rare, but when they happen, the branches are merged towards least stable branch.18:32
gram: They are mostly bugfixes.18:33
gram right, so the patches for 1.5.4 go to maint and then get pulled into next if they are not there?18:33
← DenisG left18:33
← yann left18:33
Ilari gram: Actually, they go to maint, then maint is merged into master and master is merged into next.18:34
gram oh, interesting18:35
i think i need to clone git.git and play around with it a bit18:36
Ilari gram: If one looks in git.git history, there are fair amounts of maint-into-master and master-into-next merges...18:36
dato oh, wow, gram.18:36
gram hey!18:37
← harinath left18:38
→ yann joined18:38
← arun left18:38
→ etnt` joined18:42
← dkagedal left18:48
← ulim_ left18:51
→ ulim joined18:51
← sgrimm left18:51
← threeve left18:52
→ markelikalderon joined18:55
→ threeve joined18:55
← Beket left18:59
→ threeve_ joined19:01
← threeve left19:03
→ threeve__ joined19:03
← ulim left19:03
→ deavid joined19:05
← dbtid left19:06
madduck git-rev-parse(1) says '"r1..r2" is equivalent to "^r1 r2"' and git-diff says "r1..r2" is equivalent to "r1 r2"19:07
who's right?19:07
xyzzy42 how can I get git-log to follow a copy? i.e. show me the log of a file before it was copied?19:08
madduck --find-copies-harder19:08
xyzzy42 it's not working...19:09
madduck then you modified it?19:09
make a minimal test case...19:10
xyzzy42 I didn't do anything, it's an existing repository that I want to see the history of a file in19:10
it's --follow that I needed19:11
→ strangy joined19:13
→ doener joined19:17
← Catfish left19:18
madduck ah, the manpage says that they're not the same.19:19
dancor are there conditions under which an outstanding change would prevent me from changing to branch A, but when i check it in to a new branch B and then go to A and pull from B there is no conflict19:22
← p4tux left19:23
dancor is like the merge strategy from outstanding changes more conservative than the normal merge strategy19:23
← threeve__ left19:23
dancor maybe i did an up from svn in between that somehow cleared things up19:24
Pieter I think you can only switch branches if files you have modified aren't modified on the new branch19:25
← threeve_ left19:26
→ langenberg joined19:26
thiago_home madduck: both are right19:26
madduck yeah, i just noticed...19:27
→ threeve joined19:27
madduck dancor: git-stash allows you to merge...19:28
kelvie_ is it possible to revert a hunk of a file in a commit?19:28
or to have git-format-patch create a reverse patch19:28
madduck only manually19:28
you can revert the entire commit19:28
kelvie_ ah and cherry pick it back19:28
and squash afterward :P19:29
madduck you can git-rever --no-commit and then use git-add -p maybe?19:29
kelvie_ oh19:29
even better :)19:29
→ cannonball joined19:31
→ sverrej joined19:31
wagle_ → wagle19:33
wagle rhetorical minds need to know why make info still uses docbook2-texi command19:33
← paulproteus left19:34
→ authentic joined19:35
← devogon left19:35
Pieter Hmm19:38
→ sgrimm joined19:39
wagle_home (no current rpm's supply it)19:40
← sgrimm left19:41
Pieter why can I only do "git archive <tree-ish>" in a subdir if that subdir actually exists in that tree-ish?19:41
← authentic left19:42
DenisG_ meh19:43
my git is screwed up19:43
I upgraded my perl and now it stopped working19:44
:(19:44
Pieter I find that hard to believe19:44
perhaps git-svn stopped working?19:44
Ilari Also, some other commands depend on perl (such as git remote).19:45
← njero left19:45
← Ademan left19:46
→ Ademan joined19:46
Pieter I guess there are only three important ones19:46
← aortona left19:46
DenisG_ hm, maybe git-svn did19:46
Pieter who uses git-relink anyway?19:46
Ilari git-add--interactive.perl git-cvsexportcommit.perl git-cvsserver.perl git-send-email.perl git-archimport.perl git-cvsimport.perl git-relink.perl git-svn.perl19:47
Pieter yeah, add -i, remote and send-mail, the others are just for fun19:47
← priidu_ left19:48
Ilari (that list of 8 commands is from newest master, remote doesn't depend on perl anymore)...19:48
Pieter ah, I was on 1.5.319:49
I had made the same list :)19:49
right, only 2 commands then19:50
DenisG_ hallelujah19:53
finally19:53
got the git-svn repo moved into the svn repo19:53
it just took, like, 8 hours!19:53
→ sgrimm joined19:54
← dsaxena left19:56
← sbeyer left20:01
← leethal left20:02
bdiego → midga20:04
midga → midgar20:04
→ ebel joined20:05
→ sbeyer joined20:06
Lemurnomicon so is it possible to graft a file and its history from one repository? the specific case I have is one perl file which was maintained under RCS that I want to merge in with the rest of our perl repository.20:07
→ flujan joined20:07
→ antares joined20:08
← pombreda left20:10
wagle if all else fails, build a sequence of patch files, and a corresponding sequence of files with the commit messages in them.. using script, commit them one at a time... if you care about the commit and author name/timestamps, look into git-filter-branch (or look at its script)20:11
← shafty left20:13
Lemurnomicon I was thinking I could just try pulling in from a repository with just the file in question onto a branch and then cherry-pick the HEAD over onto master20:13
→ MDCore_ joined20:13
MDCore_ → MDCore20:13
wagle main thing is creating a commit with the desired commit name/date, author name/date, commit message, and file contents.. if you want to create the commit first, then move it to master, that seems reasonable20:17
intripoon can I push to a repository like http://... ?20:18
→ pietia joined20:20
fujin sure, if it's running webdav20:20
it's incredibly awesomly slow though, compared to ssh+git or git://20:21
← bentob0x left20:21
intripoon ok thx20:21
Pieter well, http is nice for an initial import of a large repo20:21
→ pombreda joined20:22
Ilari fujin: Don't push over git://20:26
intripoon why not?20:26
Ilari intripoon: No security.20:26
intripoon and if there is a vpn?20:27
← jmspeex left20:27
Ilari intripoon: And firewalls to firewall that daemon from unauthenticated users? And good deal of effort expended to keep unauthorized users out?20:28
intripoon is git-daemon a seperate binary or is it included git?20:28
Ilari intripoon: Included in git (but it is its own binary).20:28
intripoon is it available for windows?20:28
Pieter I think there was a problem with git-daemon on msysgit20:29
but it might run in cygwin20:29
Ilari /usr/bin/git-daemon: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), for GNU/Linux 2.6.8, dynamically linked (uses shared libs), stripped20:29
intripoon I don't have it in the dir where all the other git-* are20:29
→ ebel_ joined20:30
← flujan left20:30
← ebel left20:30
→ leethal joined20:30
ebel_ → ebel20:30
Ilari intripoon: For me, it is with other executables (as one of the selfstanding ones, not hardlinked to git).20:31
→ flujan joined20:32
intripoon hmm ok thx. If Pieter is right, I'll try another version20:32
← Sho_ left20:33
Ilari intripoon: Well, maybe push over git:// could be apporiate in some closed LAN. But the costs to security otherwhere probably outweight the benefit of not having to deal with SSH.20:33
← kumbayo left20:34
intripoon ok20:34
→ mario_ joined20:36
← pygi left20:36
mario_ → pygi20:36
→ warthog9 joined20:38
← Shamgar left20:43
→ Catfish joined20:44
← DenisG_ left20:44
fujin Ilari: I push over git:// where it's applicable20:48
Ilari Well, one can ignore security. But wheither that is good thing to do is seperate matter... :-)20:48
← leethal left20:51
gebi gitosis makes pushing over ssh equally simple20:51
← MDCore left20:52
→ MDCore joined20:52
→ leethal joined20:52
→ Yuuhi joined20:53
← leethal left20:53
tokkee Mikachu, fonseca: "git log | tig" may be used instead of "tig log" - how about adding alias support to tig?20:54
fonseca: Are you going to release some 0.10.x which includes that fix?20:54
tpope beware, git-svn makes it possible to break your svn repo when inept colleagues add commit a ".svn" directory20:55
Lemurnomicon ouch20:56
bronson Ya, that would be a good thing for git-svn to prevent...20:56
Ramune hmmm...20:58
i wonder what an svn repo w/ a .git directory will do to git-svn...20:58
← ceez left20:59
← pietia left20:59
tpope well committing .svn makes it impossible to check out again20:59
← alley_cat left20:59
gebi Ramune: you can't commit a foo/.git directory in git, i needed it last week21:00
→ Oejet joined21:00
← lcapitulino left21:01
→ Morphous_ joined21:03
← MDCore left21:03
→ MDCore joined21:04
→ Sho_ joined21:05
← mattf left21:06
→ vbgunz joined21:07
← QuercusMax_ left21:08
← Eludias left21:09
→ dsaxena joined21:11
← deavid left21:11
→ mattf joined21:17
→ TuxCM- joined21:19
intripoon I tried pushing via git:// and get the following error: "! [remote rejected] master -> master (hook declined)" what does that mean?21:19
← Morphous left21:20
Mikachu it means the pre-receive (i think) hook found something questionable about your push and refused it21:20
besides, pushing via git:// isn't really a good idea21:20
Ilari intripoon: Also, update hook may reject changes.21:21
Doesn't some Windows versions of git errorneously enable all hooks by default (when none should be)?21:22
Mikachu yeah21:22
→ paulproteus joined21:23
PerlJam cygwin git is broken that way.21:23
IIRC21:23
intripoon cygwin is broken as well?21:25
a few hours ago someone said the non-cygwin version is broken as well :/21:26
how do I disable hooks?21:26
fujin chmod -x21:26
intripoon or is there some information availbale which version is not known to be broken?21:27
Ilari intripoon: Rename the files in hooks/ directory.21:27
intripoon where is that dir?21:27
Ilari intripoon: In server repository.21:27
→ thannoy joined21:28
← paulproteus left21:28
fonseca tokkee: Yes, aliases would be a nice extension.21:29
intripoon oh renaming worked thx21:29
fonseca tokkee: Similar to the 'run requests' key bindings.21:30
intripoon how does git handle large files over git://? does it transmit the whole file or does it do some clever rsync-like algorithm?21:30
fonseca tokkee: alias slowlog !git log --stat21:31
tokkee fonseca: Yep - something like that.21:31
fonseca tokkee: I won't be releasing a 0.10.2, but you might get a 0.11 (with blame view and other bugs) :)21:31
tokkee fonseca: Ah, nice - do you have a timeframe for that?21:32
fonseca tokkee: What timeframe would you prefer?21:32
Is it bad luck to release the 1st or April?!21:33
PerlJam I don't know about bad luck, but you might have a hard time getting people to take it seriously :)21:33
Ilari intripoon: The algorithm isn't quite as smart as the fetch over git:// algorithm (it has problems in case of non-fast-forward pushes).21:33
intripoon no that was friday 13th.21:33
isn't push and pull the same thing jsut with opposite directions?21:34
fonseca intripoon: No.21:34
← MDCore left21:34
intripoon what's the difference?21:35
fonseca push and fetch is/are.21:35
→ MDCore joined21:35
fonseca pull = fetch + merge.21:35
tokkee fonseca: Well, anything in a not to far future should be fine :-)21:35
intripoon ok then fetch and push. how come fetch is smarter than push?21:36
← TuxCM left21:36
fonseca tokkee: OK, will do it this week.21:36
→ sverrej_ joined21:36
tokkee fonseca: Great :-)21:36
← sverrej left21:36
Ilari intripoon: Because it often needs to deal with non-fast-forward updates, whereas in standard Git usage model, non-fast-forward pushes don't come often...21:37
fonseca intripoon: I think the wording was inherited from BK.21:37
intripoon what's a non-fast-forward update/push?21:37
Ilari intripoon: Or actually, there is some nasty corner-case in algorithm that decides which commits to push. I don't remember exact conditions that are needed to trigger it.21:38
← vintrepid left21:39
Ilari intripoon: If it gets triggered, lots of unnecressary (remote already has) commits might be sent.21:39
intripoon so it's not a good idea to push you changes to some main server every now and then?21:40
→ matta_ joined21:41
Ilari intripoon: That case shouldn't trigger the issue (IIRC, it required shared repo that is updated from multiple repos, and local mirrors not up to date).21:41
intripoon how do you use it over the internet? I guess every participant is behind some firewall and don't allow access to everybody. So they have to have some server where everyone pushes their changes. no?21:41
loops intripoon, Git does not require the use of "Main" or central server. Of course you may use Git in that way if it makes sense in your case.21:41
matta_ quick question. can you git reset only a file?21:42
loops intripoon, that's one solution, you could also send patches over email etc.21:42
Pieter matta_: git reset -- path21:42
→ adante_ joined21:42
matta_ Pieter: awesome, thanks21:42
Pieter: git reset -- path --soft HEAD^ would work too, right?21:43
Ilari matta_: No, paths together with --soft are nonsensical.21:43
intripoon I guess, I'll just try to put my projects in there as see how long it takes to push :)21:43
matta_ Ilari: gotcha, thanks!21:43
intripoon .. after some work in them21:44
→ markelik_ joined21:44
Ilari intripoon: To get effect equivalent to 'git reset --hard -- path/to/file' (which isn't allowed because it would be dangerous), use 'git checkout HEAD -- path/to/file'.21:44
← markelikalderon left21:44
Ilari Oops, that was to matta_21:44
matta_ Ilari: oohh checking out a file would overwrite my changes? nifty21:45
← antares left21:45
intripoon you nearly confused me ...21:45
Ilari intripoon: Based on my experiences, object transfers are usually bandwidth-bound, unless you have fast LAN.21:45
intripoon :)21:45
loops matta_, also notice that the --hard HEAD^ must come before the -- <path>21:45
intripoon Ilari, that's why I'm asking for efficient pushes21:46
← rrregis left21:47
Ilari intripoon: AFAIK, fetching before pushing eliminates the corner-case and set of commits to be sent is computed correctly. And one would need to force the push anyway to trigger the issue.21:48
→ mithro joined21:48
intripoon so actually, if I a push works without being forced, the transfer algorithm is the same as for fetch?21:49
Ilari intripoon: No, but it would compute the same set of commits/objects to transfer.21:50
intripoon: And the bitstream format used for actual object transfer is identical.21:50
→ paulproteus joined21:52
Ilari intripoon: Ah, the issue is actually sightly worse and can also pop up when push creates new branch (again, fetch-before-push should be workable workaround).21:52
loops Ilari, how do you first fetch a branch that doesn't exist?21:52
Ilari loops: It isn't about fetching that branch. It is about knowing relationships of remote commits.21:53
loops relationship of which remote commits, where is the knowledge stored?21:53
intripoon hmm21:53
Ilari loops: When it has copies of those commits, it knows the relationships.21:54
loops copies of which commits? the branch doesn't exist.21:54
Ilari loops: The client of course has the commits that will make up that branch.21:55
loops Ilari, but it doesnt have to fetch them, it already has them21:55
Ilari, what does it get by fetching before pushing a branch that doesn't exist ?21:55
→ railk joined21:56
loops Ilari, i don't know the internals as well as you, i'm just curious21:56
Ilari loops: It avoids obscure corner case which causes push to send possibly excessive amount of unnecressary stuff.21:56
loops: Or at least should do that.21:57
loops Ilari, i believe you. I just don't understand _what_ you could possibly fetch that would help when pushing a new branch21:57
← adante left21:57
adante_ → adante21:57
→ amr joined21:57
Ilari loops: The commits that remote branches currently point at.21:58
loops: http://marc.info/?l=git&m=120599996504783&w=221:59
loops Ilari, ah. actually now i get what you're driving at.21:59
Ilari loops: fetch has HAVE negotiation, while push doesn't.21:59
loops Ilari, then nothing you can fetch will change how much is pushed22:00
← matta_ left22:01
← Yuuhi left22:01
shenson → shenson_not_here22:02
Ilari loops: Client can do (and probably does) object listing with positive refs being refs to sent and negative refs being all known commits that server has refs pointing to. And fetching can add new elements to that second set, which can decrease number of objects sent.22:03
loops Ilari, right, it's just not done over the wire like in the fetch case22:03
Ilari, thanks, i learned something.22:04
Ilari There is patch in pu that adds protocol extension which extends push protocol with commit negotiation.22:05
loops what made me ask is that gitster recently said on the list somewhere that push into a repo is just the reverse of fetching from that repo.22:07
← amr left22:09
← meyering left22:09
→ meyering joined22:09
loops ie.. B --push---> A = A <--fetch-- B but i guess there are some subtle (and probably unimportant to his point) differences22:09
← pasky left22:10
Ilari loops: Yes, there are subtle differences. The actual datastream format is the same (well, actually some datastream features are used on fetch, but are discouraged on push).22:10
→ pasky joined22:10
→ paulproteus_ joined22:11
→ ekneuss_ joined22:11
→ efaistos_ joined22:11
→ gregf joined22:12
← strangy left22:12
← paulproteus_ left22:13
gregf I have a gitignore file that ignores * by default, and I specify the few items i want to track. I have a directory with some files and lots of subdirectories in it with more files. So I thought doing !code/* would get all of those but had no luck. Tried a few other things like !code/** etc.. any way to accomplish this nicely?22:15
threeve gregf: I had this problem just yesterday. Try !code/, !code/*, and !/code/** (yes, all of them together).22:17
gregf threeve: not the pretiest I guess but if it works i'll be happy, thanks.22:17
threeve gregf: unfortunately, this does not scale to infinite subdirectories. I will be posting to the list about it soon.22:17
gregf threeve: k i'm on the list so i'll be watching for it.22:18
thanks22:18
threeve (soon meaning, hopefully this week, btw)22:18
← Oejet left22:19
→ fiacaid joined22:19
→ duncanmv joined22:19
← railk left22:20
→ dionoea_ joined22:20
← efaistos left22:21
← ekneuss left22:21
← dionoea left22:21
← jim_c left22:21
→ arun joined22:21
← paulproteus left22:23
Pieter hmm22:23
git clone'ing on a repo with a high --depth is really slow :)22:23
← threeve left22:25
← panagos left22:26
gregf no easy way to do this by the looks because from what i see i can't pick those files up after with git add . so i would need to add them each manaully. Maybe i'm better off using gitignore the normal way I thought this might be faster before.22:27
← dann__ left22:28
doener gregf: did you try !code, without the trailing slash?22:28
→ ebel_ joined22:29
← Manoj left22:29
gregf ah he mentioned !code/, so maybe that is it22:29
← jim left22:29
doener hm, no22:30
→ QuercusMax_ joined22:30
← ebel left22:30
gregf yeah adding that to my gitignore, run git add . then git status i don't see anything new in code/ so assuming that is not helping22:30
is the code, trick something fairly new i'm back a few versions on git22:31
1.5.3.722:31
← duncanmv left22:31
fiacaid A question on merging: I have two branches on a large software project, "master" and "test". The "test" branch has some good ideas I'd like to merge back to the "master" branch....but other bad ideas. Any good way to do this?22:32
All inclusions/exclusions are on a per-file basis22:32
gram cherry-pick?22:32
thiago_home fiacaid: cherry-pick the good ideas22:32
jengelh put the cherry back.22:32
gram looks like git-filter-branch could be useful if it can remove commits22:34
ebel_ → ebel22:34
thiago_home gram: it can22:35
doener !code/22:35
oops22:35
gregf: this seems to work: /*\n!code/22:36
fiacaid Hmm, cherry-pick might work. Two related questions: (1) How do I get a summary of commits in "test" but not in "master"? (2) What if I want only part of a commit (yes, I know, sloppy committing....I've chastised myself already)22:37
Pieter fiacaid: git cherry-pick -n22:38
doener gregf: the starting slash probably makes it actually use the relative path starting at the top level repo dir, so that the inverted rule matches22:38
Pieter then add only the parts you want and commit22:38
and 1: git log !master test22:38
eh22:38
^master test22:38
fiacaid Got it, thanks Pieter!22:38
doener or for a summary: git shortlog master..test22:38
← langenberg left22:39
← QuercusMax_ left22:39
← xjrn left22:39
→ langenberg joined22:39
gregf doener: add that right to gitignore?22:39
← mithro left22:40
→ dionoea joined22:40
doener gregf: yeah, that were the only two lines I had (translate \n to newline ;-))22:41
← dionoea_ left22:41
gregf nods22:41
spuk- ;2322:43
← mahound left22:45
gregf doener: ok got it with that, thanks a lot!22:45
→ BabelO_ joined22:48
← BabelO left22:48
← flujan left22:50
jaalto Is it possible to exclude some dorectoried at git-archive call?22:50
→ strangy joined22:51
← langenberg left22:51
→ SuttoL joined22:51
← cmarcelo left22:54
→ dionoea_ joined22:54
→ rui_ joined22:55
← dionoea left22:55
→ pete__c joined22:56
← peeja left23:00
→ mithro joined23:02
→ mariusSO joined23:06
→ ben_h joined23:10
← krh left23:12
→ tsc_ joined23:13
tsc_ Any msysgit folks around? I'm getting "fatal: pack too large for current definition of off_t" when I try to clone a repo that is >2Gigs with msysgit23:15
thiago_home compile with a 64-bit off_t23:15
→ peeja joined23:15
tsc_ where do I do that? Can 32bit windows handle that?23:16
thiago_home yes, Win32 API can handle it23:17
← cannonball left23:17
thiago_home if msys can do it, I have no idea23:17
→ Fullmoon joined23:18
← thannoy left23:18
intripoon the cygwin version, is it compiled with 64-bit off_t ?23:18
→ paulell joined23:19
← duderonomy left23:19
Pieter tsc_: perhaps you can set a max size on your packs?23:20
that might solve your problems23:21
tsc_ Don't know about cygwin. I'm trying the clone again with cygwin, but it will be a few minutes...23:21
← gregf left23:21
intripoon what about files > 2gb?23:22
do they work?23:22
thiago_home probably not23:22
← Catfish left23:22
tsc_ Pieter: pack.packSizeLimit says it doesn't affect git:// so I ignored it, but I am using ssh, so I'll give that a try too. Thanks for the idea23:22
thiago_home but it sounds like a bad idea to put files of more than a gigabyte inside version control management23:23
← mstormo left23:23
Pieter tsc_: I think that affects transmission, not storage perse23:23
thiago_home: yeah, and doubly so on windows :)23:23
fiacaid How can I print the commit log for only a single specified commit (and not its history)?23:24
tsc_ Yeah, it's from an svn repo at work.23:24
Pieter tsc_: you might want to repack it23:24
git-svn doesn't produce very nice packs23:24
fiacaid: git show <commit>23:24
or git log -123:25
→ FireAuthBypass joined23:25
intripoon is there a way to exclude *.obj and some other filetypes but add everything else?23:25
tsc_ Yeah, it's been repacked aggressively and clones fine to a linux machine. I'll set the packsizelimit and repack again.23:25
Pieter intripoon: echo "*.obj" >> .gitignore && git add .?23:25
fiacaid Thanks Pieter!23:25
loops jaalto, no, but you can do git archive HEAD | tar --delete path/ > archive.tar23:25
→ csc` joined23:26
→ QuercusMax_ joined23:26
intripoon Pieter, I guess that means notepad .gitignore and typing the excluded stuff inside on windows...23:26
Pieter intripoon: can't you work in the bash shell?23:26
it's shipped with msysgit and defaults on cygwin23:27
intripoon on windows? that really feels strange23:27
loops intripoon, using a decent shell on windows seems strange? agreed.23:27
intripoon, on the other hand, it lets you get work done23:27
Pieter If only windows had a decent terminal23:27
one that has, oh i dunno, a scrollback buffer23:28
and real copy-past stuff23:28
intripoon writing those things into a textfile with that bash cmd is not much faster or lsower than typing them in notepad23:28
← juanjoc left23:28
intripoon cmd has a scrollback buffer23:28
Ramune or the older versions, where MORE.COM would talk to CON: -- so ssh into a windows box, run foo/?|more -- and watch your ssh connection hang as CON: wasn't redirected :)23:28
intripoon you use rdp for that on window23:29
→ roysjosh joined23:29
← ebel left23:29
intripoon all those linux like tricks suck on windows23:29
Pieter "ssh into a windows box" heh23:30
loops intripoon, you can open an editor on Linux too if you prefer that way, but often the shell is faster23:30
asdx lol23:30
Ramune interipoon: foo/?|more is how it was supposed to work on windows, too -- excpet it did the equivalent of opening /dev/tty, when it wasn't redirected.23:30
asdx windows sucks23:30
← FireAuthBypass left23:30
Pieter Windows is ok if you don't have work to do :)23:30
Ramune i'm assuming git has no code paths were it wants to talk to the main console.23:30
asdx Pieter: it doesn't even work for pr0n :p23:31
Ramune pieter: hence its use for gaming machines.23:31
Pieter Ramune: right23:31
intripoon I actually expected some windows flaming when I was typing the post that mentions I'm using windows ...23:31
Pieter Ramune: it's a nice differentiation too -- when working on windows, you know you're not doing any real stuff, so you quickly feel guilty about being there23:31
loops intripoon, it's more like sympathy23:31
Ramune hmm... odd thought, representing git objects as boxes, packs as rooms...23:32
← etnt` left23:32
Ramune and making git-repack launch a sokoban UI23:32
loops heh23:32
Ramune hey, if you have Doom kill -9 games, why not a git repack game?23:33
Pieter I'd like to see that with --depth=20023:33
hmm23:33
that reminds23:33
[pdebie@backup ruby.git]$ time ../projects/git/git repack -a -d -f --window=1000 --depth=20023:33
it's done :)23:33
real 64m39.912s23:33
bd_ What are the expected semantics of a $GIT_EDITOR with spaces in it? Command line args for the editor, or what?23:33
bd_ is trying to get the tests running properly when the checkout path contains a space - it's already turned out two real bugs23:34
→ cmarcelo joined23:34
Pieter bd_: there's a weird semantics for that23:34
loops bd_, afaik GIT_EDITOR is mean to point to an executable only, I don't think it is passed to a shell and executed as a command.23:35
bd_ loops: Except it is, actually.23:35
loops bd_, is it?23:36
bd_ loops: take a look at launch_editor in builtin-tag.c23:36
← strangy left23:36
→ igorgue joined23:37
loops bd_, but notice that editor is passed only as a single argument23:38
Pieter at least on OS X, giving it arguments works23:38
loops bd_, ahh, it's split first23:38
bd_ loops: line 57, if it finds $, space, \t, or ', it calls sh23:38
anyway, so far git ls-remote and git rebase --merge have issues working on paths with spaces in them23:39
loops bd_, why the hell does "tag" have all this special editor handling.. odd23:40
→ EmilMedve joined23:41
bd_ loops: don't ask me :)23:42
Pieter Hmm23:43
I didn't get any difference in pack size between --window=250 and --window=1000 :)23:43
pasky 78.175.220.4 - - [01/Apr/2008:01:12:31 +0200] "POST /m/pwproj.cgi?name=git-homepage HTTP/1.1" 200 873 "http://repo.or.cz/m/pwproj.cgi?name=git-homepage" "Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 6.0; SU 3.011; SLCC1; .NET CLR 2.0.50727; .NET CLR 3.0.04506)"23:44
78.175.220.4 - - [01/Apr/2008:01:12:35 +0200] "GET /w/repo.git?a=blob;f=cgi/pwproj.cgi HTTP/1.1" 200 14073 "http://repo.or.cz/m/pwproj.cgi?name=git-homepage" "Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 6.0; SU 3.011; SLCC1; .NET CLR 2.0.50727; .NET CLR 3.0.04506)"23:44
meh23:44
some wannabe cracker :/23:44
Pieter get him!23:46
fujin loads up jupenet23:46
← aroben left23:46
loops bd_, look at commit 5e2de4f9, it lays out expected semantics23:48
bd_ *nod*23:50
Ramune pasky: ? what's odd about those requests?23:50
← rui_ left23:51
bd_ will finish it tomorrow23:51
pasky he has reset password for repo.git, git-homepage.git and few other projects spuriously23:55
while requesting pwproj.cgi source23:55
call me paranoid... :)23:55
Ramune reset the password? er... huh?23:55
i thought the web interface was read-only?23:55
pasky it never was :)23:56
Ramune oh.23:57
← venox left23:57
Ramune now i'm tempted to install it and play to see how one can reset passwords.23:58
pasky :))23:58
Ramune trying to convince ppl here to move to git -- but since it's a security-based company, having something like that would kill the migration.23:58

Logs Search ←Prev date Next date→ Channels Documentation