IRCloggy #git 2008-08-29

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.

2008-08-29

← irotas left00:00
→ supercali joined00:00
vuf it's tomorrow already00:00
→ m-vo joined00:01
← Maledictus left00:03
← krawek left00:03
spearce robinr: ping00:03
gah, not online.00:04
← parasti left00:04
doener spearce: I think he's known as gittish now anyway (got disconnected about 30m ago)00:05
spearce i think its quite late in his timezone anyway.00:06
davidfetter → guys00:08
guys → davidfetter00:08
Mikachu 2:0800:08
← cilly left00:09
← charon left00:09
← felipec left00:12
← deavid left00:12
← schacon left00:14
→ sbahra joined00:14
← jstad left00:15
← supercali_ left00:16
← polpak left00:17
→ supercali_ joined00:18
← thelsdj left00:18
← david_koontz_ left00:20
← kergoth`work left00:22
→ gretch_ joined00:24
→ francois__ joined00:27
→ nickolai joined00:31
← m-vo left00:31
→ pygi joined00:31
nickolai is there a way to keep certain files "branch-local" in git (such that they don't get merged between branches, to not create conflicts)? an example would be some kind of branch-specific config file that is versioned on each branch but shouldn't be merged between them (unlike the rest of the files)00:33
← HenrikV left00:33
← cmarcelo left00:34
sebell → sb-away00:34
← supercali left00:34
← francois__ left00:34
← mczepiel left00:36
→ xs_ joined00:39
xs_ hi guys.. i would like know how can i revert my project00:39
git revert .. and than?00:40
Eridius git revert produces a new commit00:40
you probably want git reset00:40
tpope what do you mean a project?00:40
← vintrepid left00:40
johannes nickolai: i'd try using a pre-commit hook checking for a merge and then dropping that file from the index or something. but never tried that00:40
xs_ i would make tha same like svn revert .. also delete local all changes .. so a restauration00:41
nickolai hm, that sounds interesting, i'll have to look more at the possible hooks in git. i was trying to use merge drivers, but unfortunately they have to be defined in .git/config rather than in .gitattributes, which means anyone that checks out this project will have to manually add things to .git/config to make things work.00:42
vmiklos xs_: man git-reset00:42
Gitbot xs_: the git-reset manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-reset.html.00:42
spearce nickolai: try looking at the "merge" attribute in gitattributes. you can set a per-file program to handle a merge.00:42
xs_ okay...sure...00:42
nickolai any ideas if i can define the merge drivers (as opposed to pathname-to-driver mappings) in a way that's tracked across clone?00:43
or more generally, i keep wanting to make certain .git/config settings part of the repo, such that anyone that clones the repo gets these settings as well, but haven't found a way to do this00:43
vmiklos spearce: i just wondered why does 'pu' in egit.git cause a problem with forks and it does not in git.git; do people base branches in the forks on egit.git's 'pu' or what is the difference?00:43
spearce nickolai: just put a script in the top level of the repository that you can run to add the driver definition into .git/config.00:43
vmiklos: yes, apparently people base stuff on pu every once in a while. :-|00:44
vmiklos uh00:44
xs_ uffff f.. thankss git..00:49
my rails app works ..again ..00:49
bts- if a committer to my repo made a few separate commits today, and we want to revert all of his commits for today (other people have commits in between) but also want to 'replay' his work later, should i use rebase interactive to combine the problematic commits into one, then export that commit as a patch (to save the work), then delete his commit, then re-commit that patch later on?00:51
i'm not 100% comfortable with what i'm about to do :) so i'm just doing a sanity check here00:51
← thannoy left00:53
bremner bts-: you could just keep a branch with the doomed commits and merge them back in later00:53
→ igorgue_ joined00:55
← igorgue left00:55
bts- bremner: the part i left out from this story is that i'm using git locally to clean up this mess, but our central repo is svn. so in that case, should i go with my initial suggestion, or achieve your suggestion by creating a branch with the doomed commits locally, then deleting (inverting?) those commits in master, and then commiting my inverse/commit-deletion?00:56
← danderson left00:57
bts- i guess i create a branch to save what's currently in the repo, then use rebase interactive to delete his commits, then commit?00:57
← zykes_ left00:58
bremner bts-: you original suggestion seemed ok, except for maybe losing history. You can also use git-format-patch to export the individual commits as patches once you have them isolated00:58
bts- in that case, should i use rebase interactive to combine the problematic commits into one?00:59
before i do the format-patch00:59
bremner bts-: you can, but you don't need to.00:59
← supercali_ left00:59
bts- ok00:59
great. thanks a lot for the help00:59
bremner good luck00:59
bts- :)01:00
← \ask-_ left01:01
← rdavila_ left01:02
→ zykes- joined01:03
← kanru left01:03
← dshep left01:03
← eddyp left01:05
← rmh3093 left01:06
← statim_ left01:06
→ aiur joined01:07
→ rmh3093 joined01:09
→ CodeOfficer joined01:10
← rmh3093 left01:10
→ HenrikV joined01:11
gretch_ hi guys, how will I undo a deleted file?01:14
j_engelh any way you like.01:14
gretch_ i'd used git rm01:15
Eridius have you committed?01:15
gretch_ not yet committed though01:15
Eridius git checkout HEAD -- file01:15
→ rmh3093 joined01:16
← cbreak left01:17
← mw left01:17
← xs_ left01:17
← joydivider left01:18
bts- bremner: so i created the patches to re-apply later. now i can use rebase interactive to just delete the lines corresponding to those commits?01:18
gretch_ tnks Eridius. but still it doesn't01:18
Eridius gretch_: what?01:19
bts-: if you delete the lines, rebase -i will skip them01:19
← rmh3093 left01:19
bts- Eridius: but the help message says: # If you remove a line here THAT COMMIT WILL BE LOST.01:19
gretch_ the file is still deleted01:20
Eridius bts-: well yeah, that's what you just said you wanted, right?01:20
gretch_: does the file exist in HEAD?01:20
bts- Eridius: yeah. i want those changes gone. but i want to be able to reapply them later (i've already saved them with format-patch)01:20
Eridius: i should be good, right?01:20
Eridius bts-: yeah01:20
bts- Eridius: thanks!01:21
Eridius besides, if you remember what the sha1s are, you can just re-apply them with git-cherry-pick later too. It'll take about 30 days before they'll disappear01:21
(30 days is how long the default reflog is, and at the very least git-gc won't delete objects newer than 14 days)01:21
gretch_ it is in the HEAD, I ddn't committed the deleted file01:22
doener Eridius: isn't the reflog unlimited since 1.6.0?01:23
← jeffpc left01:23
Eridius doener: if it is, I didn't notice01:23
gretch_: then are you sure you typed the git checkout HEAD -- file properly?01:23
← digitalhobbit left01:23
Eridius just plain `git checkout -- file` won't work, since that uses the index01:23
bts- Eridius: hm when i tried to sync back to a central svn repo, i had to git svn rebase first, and apparently that made the commits come back01:23
Eridius bts-: trying to manipulate the history of a git-svn branch is dangerous01:24
and hard to do right01:24
bts- Eridius: i made patches for the commits i want gone.. can i use git am to apply them in reverse, then sync that back to the central repo?01:24
Eridius 60bce2bb8b3cd5ca56f8156cbca16abee151d81701:24
Gitbot [git 60bce2bb8]: http://tinyurl.com/6maepy -- Make default expiration period of reflog used for stash infinite01:24
→ paltman joined01:25
bts- that way i'm not rebasing or anything; i'm just doing a brand new commit that happens to be the inverse of some old commits01:25
Eridius bts-: what's your goal here?01:25
if you want to invert old commits, then don't try and rip them out of the history01:25
just use git-revert01:25
bts- someone on my team made a bunch of commits today that we want to undo. but we want to re-apply them later01:25
Eridius doener: is 60bce2bb what you were talking about? that's just for the stash01:25
bts-: did they make these commits in svn?01:25
bts- yes01:25
Eridius then you cannot rip them out of history01:26
bts- okay01:26
Eridius you have to use git-revert01:26
→ jonshea joined01:26
→ shenie joined01:26
→ _VVSiz_ joined01:26
gretch_ Eridius: sorry noob here. git checkout HEAD --file -> absolute path ryt?01:26
Eridius it's not --file, it's -- file (With a space)01:27
doener Eridius: I just read something along those lines on the list, where someone was relieved that reflog entries no longer expire by default. The above question was not (meant to be) rhetorical ;-)01:27
bts- ok so i'll use git revert to revert the series of commits. git will allow me to re-apply these commits later, right? i saved them with format-patch01:27
Eridius and relative path01:27
doener: as far as I'm aware the default for expiry hasn't changed01:27
but 3cb22b8efe adds per-reflog expiry configuration01:28
Gitbot [git 3cb22b8ef]: http://tinyurl.com/5pw367 -- Per-ref reflog expiry configuration01:28
Eridius bts-: throw away your format-patch'ed commits01:28
just use git revert and give it the commits you want to reevrt01:28
bts- ok01:28
then how do i re-apply the old ones?01:28
later down the line01:28
Mikachu revert the revert01:28
shenie hi there, if i have 2 repos with disjoint set of files. is there a way to merge repos #2 into #1 without a merge commit?01:28
bts- :) thanks everyone01:28
Eridius shenie: if you want to unify two histories, you need a merge commit01:29
shenie i see. that's my conclusion after trying various combo of rebase, merge pull etc.01:30
wanted to check if i missed something01:30
thanks for clarifying01:30
→ rmh3093 joined01:30
← paltman left01:30
← VVSiz_ left01:32
← spearce left01:33
← MK_FG left01:37
→ paltman joined01:37
→ afiestas joined01:40
→ mithro joined01:42
→ LostInTheWorld joined01:43
MaddHatter if you want the files without the history, just copy and git add01:44
bts- Eridius: thanks again. everything worked great01:44
Eridius great01:45
shenie MaddHatter: that's what i ended up doing now :) thanks01:46
← HenrikV left01:52
← yofel left01:52
→ krawek joined01:54
→ pjhyett joined01:55
→ sunblush joined01:55
← m4v left01:57
sb-away → sebell01:57
→ m4v joined01:58
→ shoe joined02:00
← loincloth left02:00
shoe I know about git log --left-right, but is there someway to get only the left or the right?02:01
In particular, I'm combining with --cherry-pick.02:03
Mikachu shoe: a sec02:04
http://git.mika.l3ib.org/?p=git.git;a=commitdiff;h=ffe43c3302:05
you can probably deduce how to add --only-right from there, if you're too lazy to reverse the a...b to b...a every time :)02:05
→ stillLotR joined02:06
sunblush → sunblush|away02:12
machrider what's the simplest way to rewind the head of my branch by one commit?02:13
← LotR left02:14
← sgrimm left02:14
← igorgue_ left02:16
Eridius machrider: git reset HEAD^02:17
machrider thanks, just figured that out02:17
Eridius if you want to update the working tree too (throwing away any local changes), git reset --hard HEAD^02:17
machrider i had ^HEAD.. duh02:17
Eridius haha02:19
j_engelh what the hell02:20
13 fucking tv channels showing politics02:20
shoe Mikachu: thanks, have you sent a pull request for that?02:20
→ loincloth joined02:27
→ \ask-_ joined02:31
← crab left02:34
← ijcd_ left02:34
← sbeyer left02:37
← \ask-_ left02:40
← whee left02:44
→ bobmcw joined02:45
Mikachu shoe: no02:45
→ voodootikigod joined02:49
shoe Mikachu: I have a related question you might know the answer to... sometimes --cherry-pick doesn't work as I expect. Instead it lists apparently identical commits on both the left and right sides.02:52
Mikachu could be that the diff is slightly different02:52
→ schacon joined02:53
→ EmilMedve joined02:55
← tongueroo left02:57
shoe Mikachu: oh, I see, does the detection of equivalent commits not respect the --unified=2 setting in git log --left-right --cherry-pick --unified=2 a...b ?02:57
Mikachu no idea02:57
zjason_ → zjason02:57
shoe because, with --unified=2 there is no difference in the diffs, but there is a difference in the third context line.02:58
→ spearce joined03:00
← sebell left03:00
→ ajonat joined03:01
← aiur left03:03
← spearce left03:04
litage in an ignore-file, what's the difference between tmp/**/* and tmp/*/* ?03:05
Mikachu shoe: it probably uses git-cherry which in turn uses git-patch-id iirc03:05
← schacon left03:05
Mikachu which doesn't take any options03:05
← LiamH left03:06
→ sgrimm joined03:06
→ schacon joined03:10
→ aiur joined03:11
nDuff litage, ** recurses03:11
→ hobbsc joined03:12
nDuff litage, ...so while tmp/*/* is exactly two levels deep, tmp/**/* may be more.03:12
hobbsc anyone recognize this error: http://dusk.pastebin.com/m63bed48b03:13
nDuff litage, (IIRC, that's not a regular POSIX glob rule, but an extension used by several "modern" shells such as zsh... and presumably in git as well, if you've cause to ask about it here)03:13
hobbsc potentially related to this: http://dusk.pastebin.com/m73a63fdf03:13
with gitosis03:13
→ rdavila joined03:13
← lea_w left03:14
← bts- left03:17
← jonshea left03:18
← EmilMedve left03:24
← sbahra left03:27
← flaguy left03:29
← schacon left03:30
→ johnw joined03:34
→ statim joined03:35
← kukks left03:40
→ EmilMedve joined03:41
← johnw left03:42
→ johnw joined03:42
← mithro left03:46
→ idahoev1 joined03:47
→ MK_FG joined03:50
→ mithro joined03:50
→ digitalhobbit joined03:51
→ bbs joined03:55
litage nDuff: thanks for the explanation about '**'. much appreciated03:56
→ imyousuf joined03:56
→ offby1` joined03:58
← eikonos left03:59
← drewr left04:00
← voodootikigod left04:02
← rafaelbandeira3 left04:02
→ perezd88 joined04:02
← eternaleye left04:02
→ gberdyshev joined04:04
← gberdyshev left04:05
→ schacon joined04:05
→ eternaleye joined04:05
← zachinglis left04:05
← eternaleye left04:10
← markkalderon left04:11
← offby1 left04:12
← MattDiPasquale left04:14
→ eternaleye joined04:15
← hobbsc left04:15
← perezd88 left04:17
← schacon left04:18
→ capisce joined04:23
→ spearce joined04:27
← tjafk1 left04:27
← xenoterracide left04:27
→ tjafk1 joined04:27
ben_h what's a neat way to look back through the commit history of a particular file? I'm playing with 'git diff @{n} path/to/file', but that seems to only go back to a certain point04:30
relatively small amount of the total history.04:30
spearce git log --follow -- path/to/file04:31
ben_h niiiice04:31
spearce your other problem is @{n} is the wrong way to look at history. its only the local history, not the project history. you really meant HEAD~1, HEAD~2, HEAD~3, HEAD~n. but log --follow is easier.04:32
ben_h ahh ok04:32
:)04:32
yeah that makes sense.04:32
explains the truncated history04:32
→ mediogre joined04:35
← rdavila left04:35
← capisce_ left04:35
← rmh3093 left04:36
← madewokherd left04:38
← G_SabinoMullane left04:40
→ priidu joined04:41
→ schacon joined04:44
→ iggy joined04:51
→ dsal joined04:55
← dsal left04:55
→ eikonos joined04:57
← eikonos left05:00
← mithro left05:05
→ barn3y joined05:05
→ rmh3093 joined05:11
→ mithro joined05:12
→ pipegeek joined05:13
pipegeek does git-checkout -f throw away the index as well as the working tree?05:14
spearce yes05:15
pipegeek thought as much.05:17
spearce the change flows from the commit into the index, and then from the index into the working tree05:18
→ igorgue joined05:19
MaddHatter Does Johannes Schindelin hang out here?05:21
or more to the point, is msysgit.git still how to get that source, or is it all in git.git itself now?05:21
mstormo MaddHatter: Git on Windows is merged into main now. However, msysgit still has some more patches which hasn't made it into mainline yet.05:23
← gretch_ left05:23
→ gretch_ joined05:23
MaddHatter How does that work?05:24
mstormo MaddHatter: btw, Johannes Schindelin (Dscho) goes by the nick "gitte" here..05:24
(not on atm)05:24
MaddHatter good to know, thanks05:24
gretch_ → gretchen05:24
mstormo MaddHatter: How does what work? git on windows, or ?05:24
MaddHatter Does msysgit have a bunch of patches that have to be applied to the git.git clone before compiling... or?05:24
(kind of like freebsd ports, I guess?)05:25
mstormo No no, you can _use_ the main git on windows. Just that msysgit has a few more patches to handle some specific Windows "issues" which haven't made it into git main yet. But the gist of it has05:26
MaddHatter I'm trying to get at how do I get the most bleeding edge msysgit available, presumably with the code from whatever's HEAD in git.git, and compile it / maybe work on it?05:26
okay.05:26
So if I somehow manage to find something I can fix, does the commit go to git.git or to msysgit?05:26
(Not that I'll actually be able to do anything, but I figure I'll at least take a looksee)05:27
mstormo MaddHatter: I guess you're really looking for this repo, to stay on the "bleeding edge" http://repo.or.cz/w/git/mingw/4msysgit.git05:27
MaddHatter: That has been a discussion actually, where to send them. If it's a problem that manifested itself for you on Windows, but it not necessarily a Windows specific issue, it goes to the git mailing list. If not, I suggest sending it to the msysgit list first, so it can be discussed in a smaller targeted audience first :-)05:29
MaddHatter okay. thanks for the help.05:30
mstormo sure, np05:30
← aiur left05:31
→ aiur joined05:31
→ cedricv joined05:37
→ vintrepid joined05:37
← paltman left05:45
← vintrepid left05:50
← rmh3093 left05:52
← mithro left05:53
MaddHatter ug. do I really have to have a google login to participate with msys :-/05:54
rtyler MaddHatter: bugmenot? :p06:02
← spearce left06:03
→ badserii joined06:04
→ Weaselweb joined06:06
→ Ademan joined06:09
← numbah left06:11
← uau left06:12
→ eddyp joined06:12
→ dancor joined06:12
dancor gitweb or cgit06:13
→ uau joined06:14
← k776 left06:15
→ Aikawa joined06:18
→ eternaleye_ joined06:19
← bronson_ left06:20
← eternaleye left06:20
← ciaran left06:25
→ copawaves_ joined06:29
copawaves_ how do I revert an accidental modification of a file?06:30
→ bentob0x joined06:31
→ Paf joined06:31
dancor copawaves_: git-reset if you haven't already pushed it anywhere06:31
thiago_home git checkout HEAD filename if you haven't committed yet06:33
copawaves_ beautiful, thanks!06:33
→ mczepiel joined06:37
← aquabu left06:37
→ bma_ joined06:38
bma_ → bmalee06:38
← LostInTheWorld left06:43
→ pjhyett_ joined06:46
→ SimonXu joined06:50
← sunblush|away left06:51
SimonXu Hi! "git reset --hard" failed, the error message was "fatal: exec read-tree failed. fatal: Could not reset index file to revision 'HEAD'." Any idea?06:51
johnw invalid permissions?06:52
volante_ can you execute 'git-read-tree'?06:52
johnw no r/x bit on .git?06:52
← eternaleye_ left06:54
→ eternaleye joined06:54
SimonXu git-read-tree: command not found06:55
← schacon left06:57
← digitalhobbit left06:58
→ ciaran joined06:59
→ floppyears joined07:00
floppyears hi guys07:00
given a sha1 how can I diff the current head and a given sha1 ?07:00
johnw git diff HEAD sha107:01
MaddHatter git diff <sha1> HEAD07:01
← krawek left07:01
MaddHatter or git diff <sha1>.. if you prefer07:02
← pjhyett left07:02
floppyears THANKS MaddHatter07:02
MaddHatter johnw too07:02
SimonXu I installed git-1.6.0.1 manually, there's no git-read-tree installed.07:03
MaddHatter did you try git read-tree?07:03
SimonXu git read-tree is fine07:04
but it did not output anything07:04
MaddHatter and this surprises you?07:05
johnw SimonXu: it's in /usr/local/libexec/git-core07:05
MaddHatter I was trying to find the readme where they talked about how git-* was deprecated07:05
johnw i miss the git-* commands07:06
→ digitalhobbit joined07:06
→ ph^ joined07:06
johnw because I was fond of typing git-<TAB> to remind myself of what was available07:06
SimonXu I know the reason now. I installed git in /home/simon/local/usr, and I added "/home/simon/local/usr/bin" to my $PATH, it seems that what I did is not enough07:06
Mikachu johnw: oh please don't start this discussion again... :)07:06
MaddHatter there's a completion script that will complete commansd for you, isn't there? (I haven't tried it)07:06
johnw Mikachu: I missed those discussions previously07:06
Mikachu johnw: it's not previously, it's now, on the ml07:07
johnw MaddHatter: It doesn't complete *all* of them07:07
MaddHatter nothing a patch or two couldn't fix then, right? :-)07:07
johnw well, I can add /usr/local/libexec/git-core to my PATH, so I have no complaint07:07
Mikachu in the zsh completer you can select which groups of commands you want completed..07:07
johnw in fact, doing so now...07:07
Mikachu changing __git_porcelain_commands to __git_all_commands near line 1643 should do it07:09
even better is probably $__git_all_commandlist (no surrounding $() then)07:09
SimonXu johnw: I have added ~/local/usr/libexec/git-core to my $PATH, but "git reset --hard" gives the same error07:11
johnw and if you type: type git-read-tree?07:11
Mikachu SimonXu: how did you compile git? did you give the prefix only to make install? you have to give it to make too07:12
as it says on line 8 in INSTALL07:13
MaddHatter I thought prefix was usually given to configure...?07:13
→ SimonXu_ joined07:13
Mikachu MaddHatter: configure isn't the recommended way to install git07:13
← ajonat left07:13
Mikachu as INSTALL also says07:13
SimonXu_ Mikachu: How to configure git to be installed in ~/local/usr properly? and work properly?07:14
Mikachu SimonXu_: read lines 8 and 9 in INSTALL07:14
→ nadim_ joined07:15
← nadim left07:15
← BBHoss left07:15
← eternaleye left07:16
→ ciaran_ joined07:17
← SimonXu_ left07:17
→ eternaleye_ joined07:17
← ft left07:17
→ ft joined07:17
← fbond left07:17
→ fbond joined07:17
floppyears hi guys07:18
how can I view the log of commits that have happened between a given sha1 and the current HEAD07:18
Mikachu git log agivensha1..07:18
→ SimonXu_ joined07:18
Mikachu with the dots07:18
→ BBHoss joined07:19
→ lol_h joined07:19
→ drizzd_ joined07:20
floppyears thanks Mikachu07:20
← Weaselweb left07:21
→ redondos joined07:21
→ mtodd joined07:23
→ bartocc joined07:24
bartocc hi07:25
I'd like to pull a branch from a fellow developper07:25
we are on the same subnet, and I can access he's box with ssh07:25
SimonXu_ Mikachu: I recompiled and reinstalled git, now I still get the same error: fatal: exec read-tree failed.07:26
bartocc I tried to add a remote like "git remote add friend [email@hidden.address]07:26
Mikachu SimonXu_: what make commands did you use?07:26
→ snitko joined07:26
bartocc but then, a git remote show friend fails with git-upload-pack does not exist ...07:26
what am I missing ?07:26
Mikachu bartocc: ask SimonXu_ ;)07:27
bartocc: probably you have a semi-old git on one of the machines07:27
SimonXu_ Mikachu: make prefix=/home/simon/local/usr all07:27
Mikachu: and make prefix=/home/simon/local/usr install07:27
bartocc that's right, I've got Git 1.6.0.90.g436ed on my box, and 1.5.5 on the other07:27
SimonXu_ Mikachu: I don't have root priv on this machine07:27
Mikachu SimonXu_: well, that should work then.. i don't know why it can't find it07:28
← SimonXu left07:31
← drizzd left07:31
→ charon joined07:31
→ Weaselweb joined07:31
SimonXu_ Mikachu: Thanks07:32
→ icwiener joined07:33
← bentob0x left07:33
← ciaran left07:34
gretchen hello07:34
how can we handle externals in git?07:35
→ atom joined07:37
johnw submodules07:37
atom hi everyone... This question does not truly belong in this channel, but as in #gentoo noone's answering, and there's a good chance someone here will know...07:38
does anyone know where all the git-* programs have gone with 1.6.0?07:38
I just have git now07:38
johnw look in /usr/local/libexec/git-core07:39
atom johnw: that directory does not exist.07:40
johnw try /usr/libexec then07:40
git --exec-path will tell you where they are07:40
atom johnw: yeah, that's where they are, sure enough. So, I just add that to my path and that'll be the end of it?07:40
thanks.07:41
johnw that's what I did07:41
← ciaran_ left07:41
Auris- atom, why would you want to?07:42
atom Auris-: um... because I like to use the git-* syntax instead of git * syntax?07:42
Auris- okay.07:42
bartocc can I "git remove" a folder but have git leave the folder and its content intact ?07:44
johnw git rm --cached -r folder07:45
← Ademan left07:45
johnw the index is what gets committed, not your working tree07:45
machrider i pulled some work in from a remote repository, so now my repository has references to a couple remote branches07:46
i'm probably not going to use them again07:46
do you just delete them like normal branches?07:46
bartocc thx johnw07:48
johnw machrider: do you want to delete all references to the remote?07:48
→ BabelO joined07:48
machrider yeah07:48
← BBHoss left07:49
dwmw2_gone → dwmw207:49
johnw git remote rm NAME07:49
machrider cool, thanks07:49
← ehc left07:50
← eddyp left07:50
← pygi left07:51
← digitalhobbit left07:53
← ben_h left07:55
← gretchen left08:00
→ mako joined08:04
→ lea_w joined08:06
→ gretch_ joined08:07
→ Ademan joined08:09
← ph^ left08:11
← mczepiel left08:14
→ redondos_ joined08:14
→ [tla] joined08:19
← redondos left08:21
redondos_ → redondos08:21
→ hein2 joined08:22
← Sho_ left08:22
hein2 → Sho_08:22
← SimonXu_ left08:24
→ elmex joined08:26
→ koke joined08:27
→ offby1`` joined08:28
→ zerok joined08:35
← Gof left08:39
← Illarane left08:41
→ Fullmoon joined08:42
← offby1` left08:42
→ hurikhan|Work joined08:42
→ lea_w1 joined08:42
← lea_w left08:42
lea_w1 → lea_w08:43
← Paf left08:44
← Fullmoon left08:46
→ gittish joined08:47
← barn3y left08:54
→ Kbyte joined08:58
← charon left08:58
→ danten joined09:05
← iml left09:17
ciaran29d I want to put a repo up on github publicly but I’d like to not show the revision history (but keep it locally). Is that possible?09:17
← copawaves_ left09:17
← CodeOfficer left09:19
← eternaleye_ left09:20
→ eternaleye joined09:20
← clairvy left09:23
→ sf joined09:24
→ chris2 joined09:25
→ kbingham joined09:26
→ zerok_ joined09:28
zerok → Guest505709:30
zerok_ → zerok09:31
→ markkalderon joined09:34
→ ereslibre joined09:40
← litage left09:42
← Ademan left09:42
← lol_h left09:42
→ stableable joined09:44
stableable Hi - when trying to checkout a submodule I get an error of: "reference is not a tree". How could this come about? What went wrong to cause it?09:45
→ RubyHunt joined09:47
RubyHunt do you have to have git installed locally to download a file from github.com ?09:48
→ Yuuhi joined09:49
← snitko left09:49
→ lanfeust21 joined09:51
← shenie left09:54
← bobf left09:55
← elmex left09:56
→ yofel joined09:56
tango_ ciaran29d, what's the purpose of putting the repo on github if you're not showing the history?09:57
ciaran29d tango_: well… the files??09:58
johnw you have no better way of just giving people a tarball?09:58
tango_ ciaran29d, why github, I mean.09:58
ciaran29d tango_: I just don’t want to show the *current* history, not all future history…09:59
it has passwords and stuff in that I don’t want visible09:59
← aiur left09:59
tango_ ciaran29d, github and similar services are there to host git repositories, not checked out working directories09:59
ah09:59
in that case you want to rewrite your history removing that old stuff09:59
and then publish it on github09:59
→ parasti joined10:00
tango_ ciaran29d, iow, start a new repository from the point where you want history to be published10:00
ciaran29d well, that’s exactly what I want to do, to avoid rewriting stuff :) if it comes to that I will probably just nuke the repo and make a new one10:01
tango_ I must say that the expression "future history" is wonderful 8-D10:01
ciaran29d hehe :)10:01
nn → _nn10:03
→ HenrikV joined10:03
→ elmex joined10:04
← markkalderon left10:09
→ Ademan joined10:10
Pieter http://snippets.dzone.com/posts/show/599610:18
heh10:18
why doesn't the guy just git archive HEAD?10:18
_nn → nn10:19
ciaran29d lol10:19
Tv that's not quick, but it sure is dirty10:19
parasti that's a terminology issue... what's the difference between "tree" and "treeish" anyway10:22
→ Ace_NoOne joined10:22
Ace_NoOne how can I quickly check when I last pushed to origin?10:23
that is, if I wanna commit --amend, I first need to make sure it's not pushed yet10:23
← sf left10:24
doener Ace_NoOne: git fetch origin && git log origin/master..master10:24
Ace_NoOne: that shows only your local commits10:25
parasti: treeish is anything that can be resolved to a tree10:25
parasti: AFAIK10:25
parasti: eg. a commit can be resolved to its top-level tree object10:25
Ace_NoOne doener: git log origin/<branch> is what I needed - thanks!10:26
not sure what the ..<branch> does though10:27
doener "a..b" == all commits in b that are not in a10:27
Ace_NoOne ahh thanks10:27
← mediogre left10:35
→ pmf joined10:39
→ deavid joined10:40
← MK_FG left10:42
← ft left10:44
→ sbodo_w joined10:45
→ markkalderon joined10:46
→ cilly joined10:50
→ boink__ joined10:51
← Gitzilla left10:55
← pjhyett_ left10:55
← pmf left10:59
← imyousuf left11:00
ciaran29d I hate when I am writing a commit message on the command-line with -m'foo' and then my commit message needs an apostrophe :(11:00
← raz left11:03
boink__ → raz11:03
Bombe ciaran29d, that's why I prefer to use double quotes on the command line. :)11:04
→ FoobarWidget joined11:04
FoobarWidget are 'git fetch' and 'git remote update' the same thing? if not then what are the differences?11:04
Pieter and then you need an exclamation mark ;)11:04
→ cbreak joined11:05
ciaran29d exactly ;)11:05
Pieter FoobarWidget: the latter will fetch all remotes11:05
Bombe :)11:05
FoobarWidget what about 'git fetch <specific remote name>' ?11:05
Pieter that'll fetch just a single remote11:05
Bombe Besides, one could always write -m'foo'"'"'s a bitch!' :)11:05
FoobarWidget so that's the same as git '<specific remote name> update'?11:05
ciaran29d Bombe: that’s true :)11:05
hm, is there a flag I can add to commit to force EDITOR to open even though I have specified a message with m? Then I could edit it11:06
Pieter FoobarWidget: 'git remote update <remotename>' perhaps, but I don't know if that works11:06
FoobarWidget yeah that's what I mean11:06
Bombe Also, I don't need exclamation marks in commit messages. If something pisses me off I walk up to the developer that exclamation mark would be aimed at and smack him in the head! :)11:06
ciaran29d hehe11:06
FoobarWidget I'm reading a tutorial about git submodules11:06
and it says that instead of doing a 'git pull' in the submodule,11:06
one should do 'git remote update origin && git merge origin/master' in the submodule11:07
Pieter ciaran29d: -e11:07
FoobarWidget so I was wondering whether that isn't the same as 'git fetch origin && git merge origin/master'11:07
ciaran29d Pieter: sorry, I should have just rtfm :)11:07
Pieter FoobarWidget: yes11:07
ciaran29d: yes ;)11:07
FoobarWidget ok11:07
ciaran29d but that’s perfect, I can just stop when I hit an apostrophe and use -e11:07
Pieter can't you just escape it?11:07
FoobarWidget in the latest git versions, 'git status' uses the pager to display its output. is there any way to disable that, but only for 'git status'?11:08
Pieter FoobarWidget: yes11:08
FoobarWidget: but latest git does not use a pager for status, afaik11:08
FoobarWidget: that was latest git -111:08
← doener left11:08
Pieter FoobarWidget: see man git-config, there's a <command>.pager config variable11:08
Gitbot FoobarWidget: the git-config manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-config.html.11:08
FoobarWidget ok. thanks :)11:09
it doesn't list 'status.pager' but I'll try anyway11:09
Pieter it's pager.<command>11:11
FoobarWidget it seems that git 1.6.0.1's 'status' doesn't use the pager11:11
I was on 1.5.6.311:11
so it's ok now :)11:11
← johnw left11:11
up_the_irons FoobarWidget: they took it out11:11
FoobarWidget: i liked it, but apparently there was a blacklash ;)11:12
FoobarWidget good decision. it was very annoying and prevented me from copy & pasting filenames properly11:12
→ doener joined11:13
up_the_irons FoobarWidget: bah, w/e ;)11:13
parasti next thing we know, git status is on the news, apparently being the cause of global warming11:14
err, git status + pager11:14
up_the_irons haha11:15
→ ft joined11:19
→ botanicus joined11:19
→ BarryCarlyon joined11:20
BarryCarlyon Hey guys, I wanna merge master into the branch base, yes I know I should be working on base, the merge doc says "Make sure you are on one of the to-be-merged branches and merge the other one now" so do I need to checkout base and then merge?11:21
← ft left11:21
bremner BarryCarlyon: git checkout base && git merge master11:22
→ ft joined11:22
BarryCarlyon Lovely thanks bremner thats what I thought just decided the check first, its my first merge :-)11:23
→ standel joined11:29
← gretch_ left11:31
← FoobarWidget left11:32
← Ace_NoOne left11:36
→ mediogre joined11:36
← EmilMedve left11:38
← Ademan left11:39
← cedricv left11:41
→ yofel_ joined11:47
← samppa left11:52
→ diegoviola joined11:55
→ Ademan joined11:55
← mnemoc left11:55
→ MK_FG joined11:56
← sbodo_w left11:57
← yofel left12:01
→ heavysixer joined12:03
← loincloth left12:03
→ bryanl joined12:03
← [tla] left12:05
→ sunblush joined12:07
← Sho_ left12:09
→ lorandi joined12:10
← parasti left12:11
← vbgunz left12:13
→ Beket joined12:15
→ parasti joined12:16
← parasti left12:16
→ parasti joined12:16
← heavysixer_ left12:18
→ akahn joined12:19
→ DrewR joined12:19
→ pygi joined12:20
→ vbgunz joined12:20
← idahoev1 left12:23
→ mw joined12:26
→ andres joined12:27
Pieter ciaran29d: I think your svn-id email change made valid email-addresses disappear too12:27
ciaran29d :(12:27
→ EmilMedve joined12:28
→ charon joined12:28
→ sf joined12:32
→ madewokherd joined12:34
→ irotas joined12:34
→ snitko joined12:35
cbrake_away → cbrake12:37
→ qrush joined12:39
→ stouset joined12:42
← gk left12:44
← Oompa left12:45
← qrush left12:48
→ qrush joined12:49
→ voodootikigod joined12:51
← madduck left12:57
→ madduck joined12:57
→ Speeple joined12:58
← stableable left12:58
Speeple hi, I'm trying to get up and running with Git12:59
I can retrieve the repo from my production server12:59
but then I wish to make changes on my local dev machine to the source code, then commit them to the production12:59
j_engelh you commit local. always.13:00
Speeple commit local?13:00
I did13:00
git commit -a13:00
then git push13:00
but I didn't see the changes on the simple web page test I did13:00
→ aroben joined13:00
j_engelh do not push to non-bare repos.13:00
Speeple non bare?13:01
Tv Speeple: http://git.or.cz/gitwiki/GitFaq#head-b96f48bc9c925074be9f95c0fce69bcece5f6e7313:01
Speeple how do I publish the local changes to the remove production machine then?13:01
parasti Tv: see faq non-bare :P13:01
Gitbot Tv: You shouldn't push to a non-bare repository. See http://git.or.cz/gitwiki/GitFaq#non-bare13:01
Tv oh smarter anchors, that's nice13:01
now i just need to convince my ff3 way-less-than-awesome bar to complete that version13:02
→ tetraedr joined13:02
→ LiamH joined13:02
Tv seriously, my reaction to that question is alt-2 control-l faq control-c alt-1 control-v enter ;)13:02
tetraedr hi13:02
I'm trying to pull changes from repo and git says on one of the files:13:03
fatal: Entry 'app/controllers/user_controller.rb' not uptodate. Cannot merge.13:03
how do I fix this?13:03
Speeple hmm so I must do "git reset --hard" to realise the changes?13:03
← madewokherd left13:03
← Beket left13:04
Tv tetraedr: you can't merge changes if you haven't committed13:04
tetraedr: that would risk losing your changes, and git refuses to do that13:04
tetraedr Tv, hm, will try now13:04
← uau left13:04
← cilly left13:04
← grahal left13:05
→ mlins joined13:05
← igorgue left13:06
← mlins left13:07
→ mlins joined13:07
Speeple so is there a tutorial for making things like from a local repo on a remote server?13:07
Tv Speeple: easy way out: have a central bare repo, pull from that to your production13:08
← mlins left13:08
→ mlins joined13:08
Speeple can't the bare repo be the production?13:08
pygi local repo is your production repo13:09
← mlins left13:09
Speeple hmm13:09
→ mlins joined13:09
Speeple I mean the source code in the repo running on the production server13:10
which is remote13:10
I'm doing this to move the code FROM local machines after having a hard drive failure recently that shit me up (thought I lost source code)13:10
I think I might be mistaken in my knowledge of what Git is13:11
mlins → mlins_13:12
mlins_ → mlins|away13:13
mlins|away → mlins13:13
Speeple I am wrong right?13:14
what I was thinking is having the directory which Apache servers webpages from as a Git repo13:14
parasti Speeple: either way your production needs a checkout -- typically the path from your local repo to another non-bare repo is local -> push to bare -> pull from bare to non-bare13:14
← madduck left13:15
Speeple OK, looks like I need to read up on the theory a little more13:15
parasti but I think you _can_ get by with just two repos, except you have to deal with some inconsistencies13:16
→ bobesponja joined13:16
parasti if the branch you push to happens to be the one checked out13:16
on the remote end13:16
→ kumbayo joined13:17
← mlins left13:17
tetraedr Tv, thanks, it works :)13:17
→ mlins joined13:17
Speeple well I was hoping to use Git as a centralised repos of my source13:17
→ _slacker_ joined13:18
Speeple so I can can get the up-to-date source on my desktop13:18
and then on my laptop when on the move13:18
→ uau joined13:18
→ Oompa joined13:21
← EiNZTEiN left13:21
bremner Speeple: sure no problem. Many people do exactly this. But the easy way is to have a bare repo for each project for interchange on your desktop (to repeat).13:22
→ Pistos joined13:23
Speeple OK cool13:23
→ mithro joined13:23
Pistos Sometimes lately when I merge or pull, it says "error: Entry 'somefile.ext' not uptodate. Cannot merge." When I git stat, then try again, it succeeds. Why is this?13:23
Speeple I've got into this over the past two days, lot's of info I have to digest13:23
→ madduck joined13:24
← stouset left13:24
← EmilMedve left13:24
→ Ryback_ joined13:26
← halfline left13:27
← zerok left13:27
← badserii left13:27
Pieter Pistos: there was a bug in the new git-merge13:27
→ halfline joined13:27
Pistos I see.13:27
Has it been fixed in repo?13:28
Pieter yes13:28
Pistos Okay, pulling now, thanks. :)13:28
→ ebel joined13:29
→ chombee joined13:29
→ vbabiy joined13:29
← qrush left13:30
chombee Slightly confused, I moved a directory with git mv then committed and pushed the change to a central git repo. Then elsewhere I pulled the changes from the central repo into another repo. Now in this last repo, the moved directory is there twice in the working copy, in its old position and its new position, and the old one is called untracked by git status. So I just deleted it from the working copy. But what happened?13:31
doener probably you had untracked files in that directory13:31
chombee No, it doesn't look like I did (looking at the directory now in the wastebasket)13:33
No, it doesn't look like I did (looking at the directory now in the wastebasket)13:33
Dreamer3 how do i revert a failed merge?13:33
git reset HEAD . ?13:34
doener revert or abort?13:34
Dreamer3 git checkount . /13:34
abort13:34
doener git reset --hard13:34
Dreamer3 i'm in the middle of it13:34
parasti ORIG_HEAD?13:34
doener (no arguments required, but if you insist on giving one, just HEAD)13:34
Dreamer3 cool13:34
:)13:34
awesomeness13:35
chombee s there a reliable way to convert a bzr repo (including history) to git, or vice-versa?13:35
s there a reliable way to convert a bzr repo (including history) to git, or vice-versa?13:36
vmiklos git-bzr13:36
http://github.com/pieter/git-bzr/13:36
→ EmilMedve joined13:36
vmiklos it requires git-1.6.0, though (it will not work with previous versions)13:36
→ FunkeeMonk joined13:38
Pieter git-bzr is cool ;)13:38
← Guest5057 left13:41
← EmilMedve left13:44
charon wow, i managed to write a four-paragraph blurb in a 0/N series leader and leave the "*** SUBJECT HERE ***" unchanged. i guess i should go visit a doctor regarding that hole in my foot.13:45
Pieter hah13:45
bremner charon: or get work in the spam industry13:46
doener maybe write "*** LOOK HERE ***" next to the whole13:46
s/whole/hole/13:46
gah13:46
→ kukks joined13:46
bremner **FIX ME** on hole in foot13:46
Pieter is waiting for a patch series to git@vger with "*** SUBJECT HERE ***" as subject13:46
← Pistos left13:47
→ G_SabinoMullane joined13:47
charon it already made the roundtrip to me again, i wouldn't have noticed otherwise13:47
← Speeple left13:48
pasky tries to summon dscho to irc13:49
← voodootikigod left13:49
Pieter that's just like summoning the spanish inquisition.. they won't come if you expect them13:49
→ qrush joined13:49
vmiklos pasky: he said he'll stay away from kbd till september13:50
← botanicus left13:50
→ stouset joined13:51
pasky ouch13:51
Pieter two more days to go!13:51
pasky I want to know where on the earth did he get ssh.exe from13:51
s/want/need/13:51
because I can't find it in stock msys :(13:52
Pieter doesn't j6t know?13:52
→ pmf joined13:52
pasky who is j6t?13:52
Pieter Johannes Sixt13:52
pasky ah, slow me13:52
:)13:52
grr, and slow my internet connection13:52
well, I'll write a mail and see if anyone replies13:52
pasky patiently waits for his net to come back up13:52
johannes funny nick13:53
pasky yeah, sixt :)13:53
gittish the pope?13:53
→ warlock_mza joined13:53
→ warlock_ joined13:53
← warlock_mza left13:53
Pieter another johannes?13:53
johannes gittish: the Pope is Bendikt, nowadays ...13:53
Pieter: yeah, a different one13:53
just to make it complicated :-)13:53
← Oompa left13:54
gittish not spellt VI either, that would make things even more confusing :)13:56
Pieter "I use the editor 'six'"13:56
→ ronny joined13:57
ronny hi13:57
gittish there was a strip in Userfriendly13:57
ronny is there any git command that gets me all informations git status supplies, but is more easy to parse ?13:58
Tv ronny: ls-files13:58
ronny trying to add git support to an ide, but getting the current tree state is kinda hellish13:58
Tv ronny: if you're programming, you need to use plumbing, not porcelain13:59
ronny hu ? (i dont get that one, im not a native speaker)13:59
well, im allready using git ls-files13:59
gittish Tv: so that's what he is asking for..13:59
ronny i seem to miss some params, as i dont get infomation about a new file file i just added to the index14:00
← tetraedr left14:00
ronny im using ls-files --cached --deleted --others --stage --unmerged --killed --modified -t14:01
← sunblush left14:02
ronny and it shows up as new file in git status14:02
bbl14:03
→ EiNZTEiN joined14:04
← diegoviola left14:04
→ EiNZTEiN_ joined14:04
← chombee left14:05
Pieter ronny: try without --unmerged14:07
← lea_w left14:07
gittish Pieter: not the same14:08
→ EmilMedve joined14:09
← snitko left14:10
gittish git diff HEAD --summary|grep create mode14:11
→ macgerhard joined14:11
→ spearce joined14:11
macgerhard during a rebase, I've got this /opt/local/bin/git-am: line 497: /opt/local/bin/git: Argument list too long14:14
→ zachinglis joined14:14
← irotas left14:15
→ Oompa joined14:15
→ irotas joined14:17
→ bogolisk joined14:17
→ sciz joined14:17
akahn I have a git repository on my machine, what do I need to do so that a co-worker can get that repository on his machine?14:17
Pieter do a 'git clone yourmachine:path/to/repo' on his machine14:18
given that you have an ssh daemon on your machine14:18
← mako left14:18
akahn Pieter: i don't need to start any other daemon process, as long as I have ssh on?14:19
Pieter yes14:19
→ alley_cat joined14:19
akahn sweet!14:19
and he won't need a password or anything?14:19
← Aikawa left14:19
Pieter yes, your password14:20
→ Aikawa joined14:20
akahn Pieter: oh, okay. is there a way to allow him to clone without my password?14:21
← mtkd left14:23
bogolisk hi, what's the "best" way of fast-forward branch without check it out?14:24
abenea akahn: you can create a user for him on your machine :)14:24
macgerhard ok, following my earlier problem, I've done a git rebase --abort and now I've "lost" the commit that caused problems when rebasing14:25
→ emmanuel_ joined14:25
macgerhard where can I find this commit?14:25
akahn abenea: is that what people commonly do?14:25
→ mtkd joined14:25
→ voodootikigod joined14:26
abenea akahn: I think it's better if you have a http server (gitweb? I'm not a git user so I don't know the details)14:26
Pieter akahn: you can run a git daemon14:26
bremner akahn: you can use gitweb.cgi14:26
Pieter akahn: or, you can use public hosting, like repo.or.cz14:26
← bbs left14:26
Pieter akahn: or you can use a shared server where everybody has an account or uses the same account14:27
akahn Pieter: okay. this is within the office so i think it makes sense for me to just host it from my workstation and he can clone14:27
Pieter: is running a git daemon as simple as git-daemon & ?14:27
Pieter akahn: yes14:27
akahn and then once he has made changes to the repo, he can start git-daemon and i can run git pull james:/Users/james/ourrepo ?14:28
macgerhard ok, I've found a dangling commit that is the actual merge that went wrong14:28
Pieter akahn: but you'll have to create a git-daemon-export-ok file in your .git directory14:28
macgerhard Pieter: any, ideas?14:28
akahn Pieter: why is that?14:28
Pieter akahn: if you just run git daemon, you'll have to specify the full path to the repository, so otherwise someone can clone other (private) repos too14:29
→ leitao joined14:29
Pieter akahn: take a look at man git-daemon for more information14:29
Gitbot akahn: the git-daemon manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-daemon.html.14:29
akahn okay, thanks14:29
→ juanjoc joined14:30
akahn bremner: how do i get gitweb.cgi going?14:31
← pmf left14:32
macgerhard I've ran git fsck, went through all the dangling commits14:33
← lanfeust21 left14:33
macgerhard but none of them is my lost one14:33
← alley_cat left14:34
macgerhard where did my lost commit go?14:34
Pieter macgerhard: try 'git reflog' and see if it's in there14:34
→ alley_cat joined14:35
→ sunblush joined14:35
sf and use a ssh key14:35
← alley_cat left14:35
→ aiur joined14:35
macgerhard Pieter: nope, doesn't seem to be in my reflog14:36
that's weird!14:36
I can see my rebases though (the ones that went wrong)14:37
emmanuel_ hello guys. I have what I think is a newbie question, can you help me?14:38
Mikachu no, because you didn't ask the question ;)14:38
macgerhard emmanuel_: no need to ask if you can ask14:38
ahh, Mikachu, I need your help mate14:39
emmanuel_ Also I have some lame ascii art to support my q.... would you be much annoyed if I paste here ? :-)14:39
macgerhard I've lost a big friggin commit14:39
no sign of it14:39
→ alley_cat joined14:39
macgerhard git fsck doesn't find it, nor does reflog14:39
Mikachu emmanuel_: use a pastebin14:39
macgerhard where did it go?14:39
emmanuel_ ok14:39
here is my q: http://pastie.org/26254114:40
macgerhard emmanuel_: git cherry-pick14:41
→ krh joined14:41
macgerhard when you're in the master branch, just cherry-pick the commits that you want14:41
it's just git cherry-pick SHA14:41
emmanuel_ hmmmm, that sounds easy.... will take a look at the man pages. Thanks!14:42
macgerhard Mikachu? Pieter? Any ideas?14:42
→ rdavila joined14:42
← voodootikigod left14:42
Mikachu nope14:42
bogolisk hi, what's the "best" way of fast-forward branch without check it out? Or maybe can I push a branch to another local-branch14:43
does "git push . branch1:branch2"14:43
does "git push . branch1:branch2" work?14:44
Mikachu yes14:44
or fetch14:44
Pieter bogolisk: or 'git branch -f branch2 branch1'14:44
Mikachu or that14:44
but it doesn't check for fast forward14:44
so you could accidentally throw away commits14:44
bogolisk Mikachu: thanx, I think I'd prefer the push! does the push check for ff?14:45
Mikachu bogolisk: if you don't use -f or +14:46
and in either case, it will tell you if it was a fast forward14:46
bogolisk Mikachu: ok, ty14:46
Mikachu and print the old sha114:46
bogolisk Mikachu: ah14:46
Mikachu the reflog is nice, but not using it is even nicer :P14:46
bogolisk Mikachu: true14:46
Mikachu: what's the "prefered" way , fetch for push?14:46
Mikachu i think neither is really preferred :)14:47
push is a letter less14:47
bogolisk Mikachu: so what't the preferred way?14:47
Mikachu i don't know what to say14:47
bogolisk Mikachu: I'm usually working ont wip branch and from time to time i want to update master.14:47
macgerhard Mikachu: am I screwed re my lost commit?14:50
Mikachu probably not14:50
macgerhard ok, that's good news14:50
what do I need to do to get it back?14:50
← alley_cat left14:52
parasti find a commit that's likely a descendant of the missing commit, then use git log14:52
...git log <descendant>, that is14:53
→ paltman joined14:53
cbreak what format do objects in .git/object have?14:53
ronny re14:53
→ alley_cat joined14:54
ronny paltman: looks fine, thanks14:54
→ gberdyshev joined14:55
emmanuel_ mmmm, I really like cherry-pick... :) That worked good! Another question about http://pastie.org/262541: Cherry-pick worked becaus I had a commit whit the files I needed (B, C). What happens if I had a commit with another file that I don't want to retrieve? (in this case A). Can I "partially" cherry-pick a commit?14:56
macgerhard parasti: ok, going for that14:56
akahn Anyone know of a guide to using gitweb with OS X and MacPorts and the built-in webserver?14:56
ronny ops i meant Pieter14:57
seanius anyone know if git-svn can be used to bootstrap a repo from a local copy of the svn repo?14:57
→ ben_h joined14:57
emmanuel_ akahn: I don't know anything about the subject but I do know that github has a section for git stuff on mac: http://github.com/guides/home14:58
seanius i.e. get a tarball of the svn tree, unpack and bootstrap from it, and then switch to the remote repo for everything else14:58
→ capisce_ joined14:58
akahn emmanuel_: neat, thanks for the link14:59
← afiestas left14:59
parasti seanius: all I know is that it can be done, couldn't tell you how though14:59
seanius: hang on, what do you mean by "svn tree"? the entire repo or just a working copy?14:59
Pieter akahn: you can also do git instaweb --httpd=webrick which will start a gitweb for the repo you're in15:00
akahn: but, you can't use gitweb to pull from15:00
seanius parasti: i want to set up a git-svn repo of someone else's project, but i don't want to hammer their server15:00
akahn Pieter: you can only clone from gitweb?15:00
seanius ideally i could just transfer the repo and do all the hard work locally15:01
Pieter akahn: no, it's just a gui15:01
akahn: you can only browse your repo, not do anything fancy with it15:01
akahn Pieter: ah, okay. i guess in this workflow we'll just stick with the command line then15:01
macgerhard parasti: cheers mate, your suggestion worked ;)15:02
I have the commit back15:02
seanius hrm... could it be as simple as just changing the svn-remote url? that'd be nice15:03
← irotas left15:03
→ irotas joined15:03
Mikachu seanius: i think there's some option for that15:03
--rewrite-root maybe15:03
seanius Mikachu: perfect, thanks!15:04
← CrnkyGeek left15:04
seanius or... hrm, maybe not entirely15:05
well, it will just require some experimentation, that's all :)15:05
Mikachu my guess is you can use that during the initial clone, then remove the option and change the svn-remote url, and the metadata should still match15:05
parasti there are probably some guides on how it works with SVK or svnsync... Google to the rescue, I suppose15:06
→ mako joined15:06
← Yuuhi left15:06
← standel left15:07
seanius Mikachu: i'll report back after i try it later tonight15:07
→ BBHoss joined15:09
→ zigzig joined15:09
← sciz left15:09
← capisce left15:10
→ rmh3093 joined15:10
→ litage joined15:11
← shoe left15:11
→ sf__ joined15:12
zigzig if I will use a git repo imported (from CVS) http://github.com/damien/pathauto/tree/master , how can I be sure the source wasn't altered?15:13
→ kergoth`work joined15:13
Pieter do a diff if you only want to check the latest release15:13
→ kreuter joined15:13
Pieter otherwise, do an import yourself and check the trees of each commit15:14
ronny hmm, im still doing someting wrong15:14
kreuter if I have branch A with commits C1 and C2, can I apply just commit C2 to the head of branch B? (C1 and C2 shouldn't really have been in the same branch.)15:15
ronny its alarming if the unittests only work for svn15:15
← rmh3093 left15:15
zigzig thanks Pieter!15:15
→ rmh3093 joined15:15
→ mheld joined15:16
mheld anybody have any experience with git in windows?15:16
cbreak only very limited15:17
→ gberdyshev_ joined15:17
macgerhard kreuter: git cherry-pick15:18
→ chombee joined15:18
macgerhard see the man on useage15:18
mheld cbreak: do you use a normal text editor then just use gitgui to push?15:18
macgerhard how appropriate is git for large files?15:18
kreuter macgerhard: thanks.15:18
cbreak mheld: I used the terminal to do git related tasks15:18
mheld cygwin?15:19
cbreak there was one in the start menu somewhere15:19
macgerhard I like to store my PSD files & artwork alongside my projects, but they run into 100s of MB.15:19
is it an appropriate approach?15:19
cbreak or in the shell right click menu, "start git here" or something similar15:19
macgerhard lib dir is always the most hammered one15:19
chombee Hey folks. Lets say I create a new local repo, then I want to publish it to a central host. So I do git clone --bare, then copy the bare repo up to the remote host. Now I want to configure the original local repo to see the remote bare repo as its origin, just as if it was created from the bare repo with git clone. Is it git remote add origin ssh://me@mydomain:/path/to/bare/repo ?15:19
→ jonshea joined15:21
→ Lash| joined15:22
ronny Pieter: any idea how to figure if a file got just added or is unchanged with ls-tree ? both cases show up as cached (H)15:22
Tv ronny: ls-tree only sees *committed* things15:23
ronny ops, i meant ls-files15:24
Tv ronny: i think what you're asking is just "how to tell if index differs from working dir (for this file)"15:24
ronny hmm, so far git and darcs look like the whorst possible things to get information about the workdir state15:24
← EiNZTEiN left15:24
Tv ronny: no way, git gives you all you could possibly want15:25
ronny Tv: but in horrific formats that are hard to parse15:25
Tv ronny: huh15:25
Tv hugs ls-tree output format15:25
ronny well, i need changes from the workdir, ie not yet commited things15:26
Tv so ls-files15:26
ronny so far it seems ls-files doesnt know the difference betwen unchanged and new file just added15:26
Tv --stage -z ftw15:26
ronny so i will need multiple calls to get all infos?15:27
← sf left15:27
Tv ronny: in both cases, index and working dir are equal -- that is what ls-files is all about15:27
ronny: do you want to ask if index and HEAD differ?15:27
← Weiss left15:27
ronny Tv: im trying to wire up a ide integration, showing changes in the file-manager15:28
→ schmir joined15:29
ronny so i guess i will need to know whats different betwen head, workdir and index15:29
Tv ronny: not in that order ;)15:29
schmir hi all. how can I create a remote branch from a local one?15:29
spearce ronny: git diff-files and git diff-index are your friends. so is -z and --raw flags.15:29
Tv head<->index: diff-files15:29
index<->workdir: ls-files15:29
err15:30
head<->index: diff-index15:30
sorry ;)15:30
spearce and index<->workdir: diff-files and ls-files --others15:30
→ jlemieux joined15:30
Tv well you can do git-diff-files instead of ls-files15:30
to get in one go15:30
ronny oh god, how many times do i actually have to call to git to get all infos15:30
Tv ?15:30
→ james_w joined15:31
james_w Hi all, could someone explain to me what's going on in http://paste.debian.net/15938/ ?15:31
→ JoeSniff joined15:31
ronny so far it seems like i have to execute at least 3 different subprocesses to collect all infos i need15:31
james_w why can't I "add" a change in a versioned file?15:31
Tv james_w: it's trying to warn you that you have that file in ignores15:32
→ Ilari joined15:32
→ krawek joined15:32
Tv james_w: -f to force it, or fix the ignore, or git rm the file15:32
ronny: what's the third?15:32
james_w Tv: yes, but why? It's a versioned file, and it will happily add the change with "commit -a"15:32
← gberdyshev left15:32
ronny ops, confused those15:32
so its 2 times15:32
Tv james_w: it's just trying to be helpful *shrug*15:33
people complain if you don't have safeguards, people complain if you have them ;)15:33
ronny still suboptimal, and im not sure if these contain all infos i need :(15:33
→ patmaddox joined15:33
JoeSniff Hello, i'm having trouble using git-svn fetch. I've trying to convert a svn repo to git. It happly processes all my svn revisions until it hits one and stops. It throws the error: "rspec/.git/refs/heads/master was not found in commit." Does anyone have any ideas or pointers?15:34
ronny hmm, now the git backend will need a complete reimplementation :(15:34
tango_ ronny, you could parse git status instead 8-D15:34
that collects all the info15:34
staged, unstaged, changed and not tracked15:34
Tv that's just wrong15:35
tango_ Tv, why?15:35
→ sbeyer joined15:35
ronny messy, error phrone, ugly, fuzzy, no guarantee of output format15:35
Tv tango_: your washbin doesn't put it's water in your toilet bowl, does it?15:35
→ voodootikigod joined15:35
tango_ lol15:36
ronny git is giving me a hard time there15:36
tango_ Tv, in fact, I consider THAT a waste of water15:36
Tv ronny: why?15:36
→ Weiss joined15:36
← voodootikigod left15:36
Tv ronny: the outputted formats are nice and contain all the info15:36
ronny: you are asking about two different things, so you get two different chunks of data15:37
ronny shattered across multiple commands tho15:37
Tv ronny: i've programmed on top of them and enjoyed it15:37
← mheld left15:37
← james_w left15:37
Tv ronny: there's no "git dwim"15:37
← zigzig left15:37
ronny Tv: well, other vcs's just give me all intresting infos about workdir in one command/function call15:37
Tv ronny: most vcses don't have anything like index15:38
ronny: you can use git in that mode if you really want (as long as the ide has full control of the index)15:38
that's like saying that handling multiple parents for a commit is hard and with svn you didn't need to do that15:39
ronny the ide doesn't and shouldnt have full controll over the index15:39
→ koke_ joined15:39
Tv yes, because it had strictly less power15:39
← koke_ left15:39
← redondos left15:40
→ irotas_ joined15:41
← irotas left15:41
← sbeyer left15:43
→ voodootikigod joined15:43
→ standel joined15:44
Ilari ronny: See how Git-gui figures out file status?15:44
→ chrisb joined15:45
Tv set fd_di [git_read diff-index --cached -z [PARENT]]15:45
set fd_df [git_read diff-files -z]15:45
well duh15:45
← cbreak left15:46
ronny hmm15:47
well, needs a complete reimplementation to get working here ( the generic code only handles one subprocess)15:47
← alley_cat left15:47
Tv sounds like a funky place to put the subprocess call15:48
like, the thing doesn't even control what it does, something else feeds it the output of that one subprocess15:48
← sgrimm left15:48
← rmh3093 left15:49
→ digitalhobbit joined15:50
→ rmh3093 joined15:50
→ cbreak joined15:51
→ IdahoEv joined15:51
← chombee left15:52
→ cannonball joined15:52
spearce you could always do the subcommand as: /bin/sh -c "git diff-index ...; echo; git diff-files" and split the output on the blank line. :-)15:53
→ CodeOfficer joined15:53
← hurikhan|Work left15:54
→ lea_w joined15:54
→ rostedt joined15:55
→ bts- joined15:55
← koke left15:56
← lea_w left15:56
→ peeja_ joined15:56
rostedt hi, can git bisect work on a branch that is created by git remote add?15:56
Mikachu "yes"15:57
← mlins left15:57
rostedt I'm debugging something on linux-tip, that I have in a repo that clones linus's tree, and I get the linux tip with git remote add15:57
Tv bisect most likely works without any branches at all..15:58
Mikachu you get it with git fetch, not with git remote add, that's what the quotes were for :)15:58
rostedt I started with the tip/master, and then did "git bisect start; git bad; git good <SHA5>"15:58
Mikachu 5?15:58
rostedt what ever that funky number is ;-)15:58
Mikachu 1 :)15:58
rostedt ah15:58
anyway, in the bisect branch that is made to run next, it does not have the "good" in the logs15:59
in fact, it looks like it went to an older version of good, and not something between good and bad15:59
Mikachu it can sometimes go a bit behind due to merges..15:59
and what git version do you have? bisect detaches head since some time16:00
rostedt I have 1.5.2.216:00
it puts the bisect into its own branch called "bisect"16:00
Ilari rostedt: That's bit old...16:00
rostedt yeah, it's what I got on RHEL 5.116:01
→ bentob0x joined16:01
Mikachu it probably works fine, even if it appears confusing ;)16:02
rostedt ok, I'll continue then16:02
Mikachu try git bisect visualize16:02
rostedt I just didn't want to waste a lot of time, and not find the bug16:02
Mikachu well, i can't guarantee anything16:02
← emmanuel_ left16:03
rostedt hmm, it may be correct. just jumping around branches16:03
ok thanks16:03
Mikachu bisect is probably one of the more exercised code paths in git16:04
offby1`` used bisect successfully yesterday16:04
rostedt Mikachu: yeah, but I was a bit worried my older version of git may have a bug16:04
offby1`` → offby116:04
← sf__ left16:05
← univate left16:07
→ schacon joined16:09
→ koke joined16:10
← aroben left16:11
← aiur left16:11
→ bdiego joined16:14
→ univate joined16:15
← patmaddox left16:16
← abenea left16:16
← kbingham left16:16
→ sgrimm joined16:20
← rdavila left16:20
→ zerok joined16:21
← schmir left16:22
← BarryCarlyon left16:29
→ supercali_ joined16:34
bremner see faq16:35
bremner blushes16:35
chuck bremner talked in the wrong window, point and laugh!16:36
Pieter hahaha16:36
bremner oh man, grade 4 all over agin16:36
again even16:36
← irotas_ left16:36
bremner akahn: http://git.or.cz/gitwiki/Gitweb16:37
← schacon left16:37
drizzd_ I think git stash apply used to default to what's now git stash apply --index, right?16:37
← voodootikigod left16:38
→ jackdempsey joined16:38
bremner akahn: you need a web server, and to enable cgi16:38
akahn: but now that I think about it, to enable cloning via http, I don't think you need it16:38
→ peritus_ joined16:39
drizzd_ I think it's a little dangerous to have apply without --index as the default, because that way you can potentially forget about the staged changes stashed away and lose them16:39
→ aroben joined16:39
→ supercali__ joined16:39
→ polpak joined16:40
drizzd_ The man page talks about potential problems with conflicts, but I think apply should simply fail in that case.16:40
bremner akahn: just make a bare repo inside the web tree, and enable git-update-server-info in ./hooks/post-update (from memory)16:40
→ mlins joined16:40
→ supercali___ joined16:41
← ciaran29d left16:42
→ aroben_ joined16:43
← aCiD2 left16:43
→ supercali____ joined16:44
← macgerhard left16:46
→ supercali_______ joined16:46
← rmh3093 left16:47
→ rmh3093 joined16:48
→ bryanl_ joined16:48
→ supercali joined16:50
→ aquabu joined16:50
← ebel left16:50
Pieter ciaranm: is it possible to create separators in drop down lists with bindings?16:50
← Aikawa left16:51
← supercali_ left16:52
→ supercali_ joined16:52
← standel left16:52
sunblush → sunblush|away16:53
← chrisb left16:54
→ chrisb joined16:55
← supercali__ left16:56
← stouset left16:57
← rmh3093 left16:57
← supercali___ left16:57
← aroben left16:58
→ rmh3093 joined16:58
← supercali____ left17:00
← mlins left17:03
← supercali_______ left17:04
→ Aikawa joined17:04
→ joydivider joined17:05
ciaranm Pieter: huh?17:05
→ guille_ joined17:05
guille_ hi17:05
← pygi left17:05
→ iml joined17:05
→ Fullmoon joined17:05
← warlock_ left17:06
Ilari Separators with keybindings? Aren't separators supposed to be unselectable? Or what kind of bindings you mean?17:06
guille_ how can I avoid git following symlinks?17:06
← supercali left17:06
← bentob0x left17:07
Ilari guille_: You can add symlinks to repo... How is chasing symlinks problem?17:08
guille_ yes, but i don't know why it's getting circular so it has two duplicate files, so it fails17:10
the thing is that i don't want to index them too17:10
so It could be helpful to know how17:10
← RubyHunt left17:12
Ilari guille_: Add .gitignore or exclude rules for the symlinks?17:12
→ pygi joined17:12
guille_ Ilari:17:13
but could be nice to have a rule for symlinks and not for files (~=)17:13
← kumbayo left17:14
Ilari guille_: Git should store symlinks in working tree as-is... What exactly is going wrong?17:14
← ben_h left17:14
guille_ two symlinks that points to the same tree (at different leaves anyway) fails17:15
Ilari guille_: Fails how? Error message? Symlinks turn into directories?17:16
guille_ "You have both: and points to the two paths that point to the same place"17:17
well after ':' it's not a quotation17:17
Ilari guille_: What Git version?17:18
guille_ 1.5.4.317:18
← rmh3093 left17:19
→ rmh3093_ joined17:19
← _slacker_ left17:20
→ GyrosGei1r joined17:20
→ grahal joined17:21
← Kbyte left17:23
→ pivotal joined17:23
pivotal → nkallen17:23
nkallen question, I would like to cherry-pick a commit range: 1a3b3378c1441c8af2d2c53933117feb06538f47..91419e79b92c38704dc03dc7d4bf2970d3f4656b ... cherry-pick doesn't seem to support ranges. What's the best way to do this?17:24
← qrush left17:24
→ ehc joined17:25
krh nkallen: I think rebase can do it17:25
with enough arguments17:25
doener nkallen: git checkout -b tmp 91419e79b92c; git rebase --onto $target_branch 1a3b3378c1^; git checkout $target_branch; git merge tmp17:25
→ peeja__ joined17:25
→ foca joined17:25
← ehc left17:26
doener this might do too: git checkout $target; git reset --hard 91419e79b92c3; git rebase --onto $target{1} 1a3b3378c14^17:26
→ LostInTheWorld joined17:27
nkallen crap, got conflicts doing git rebase --onto $target_branch 1a3b3378c1^17:27
← GyrosGeier left17:28
→ ehc joined17:28
← Weaselweb left17:28
doener resolve them, git add, git rebase --continue17:29
nkallen (resolving conflicts, then will continue)17:29
→ irotas joined17:29
→ qrush joined17:29
guille_ when i remove files from the cache they get removed from the whole repository or for the current commit?17:29
→ pmf joined17:30
→ rmh3093 joined17:30
thiago_home guille_: do you mean git rm --cached?17:30
guille_ thiago_home: yes17:30
Ilari guille_: They presumably get removed from next commit...17:30
thiago_home the file is removed from the index17:30
← rmh3093_ left17:30
thiago_home meaning it will not be present in the next commit17:30
that's it17:30
→ chris2_ joined17:31
← sverrej left17:32
← peeja_ left17:32
← mikearr left17:32
guille_ thiago_home: but could I restore them checking out, the problem is that i've tested some services that handles videos with porn, and it's not indicate for the production enviroment17:33
→ standel joined17:34
sunblush|away → sunblush17:34
guille_ they're good films, but........ you know17:34
bremner guille: git filter-branch17:34
→ timprouty joined17:34
bremner guille_: ^^17:34
← rmh3093 left17:35
→ BarryCarlyon joined17:35
→ Oejet joined17:35
→ rmh3093 joined17:36
bremner thinks "how do I remove the incriminating files from my repo" is probably in the faq17:36
→ tongueroo joined17:36
thiago_home you have to rewrite history17:38
← rmh3093 left17:39
→ rmh3093 joined17:39
→ voodootikigod joined17:40
→ parolkar joined17:40
parolkar Hi I am getting this while adding remote , Whats going wrong? Can't locate Git.pm in @INC (@INC contains: /home/abhishek/share/perl/5.8.8 /etc/perl /usr/local/lib/perl/5.8.8 /usr/local/share/perl/5.8.8 /usr/lib/perl5 /usr/share/perl5 /usr/lib/perl/5.8 /usr/share/perl/5.8 /usr/local/lib/site_perl .) at /home/abhishek/bin/git-remote line 4.17:41
← Oompa left17:41
← JoeSniff left17:41
guille_ ok, so=> git filter-branch --tree-filter 'rm ./static/users*' should work? take care that it's a relative path within the root of the git repos (they're symlinks)17:42
thiago_home where's the Git.pm file?17:42
parolkar: ^^^17:42
bremner guille_: yes, 'rm -rf static/users* ' should work. You want -f17:43
parolkar thiago_home: where should it be?17:44
← foca left17:44
← rmh3093 left17:44
→ foca joined17:44
→ ReinH joined17:45
→ thannoy joined17:45
ReinH what's the least brittle way to get the current branch name?17:45
thiago_home parolkar: in one of the paths it told you17:45
parolkar: but where is it?17:45
ReinH git name-rev --name-only --always HEAD17:45
← chris2 left17:45
← pmf left17:45
← koke left17:46
guille_ wow worked like a charm :)17:46
thanks17:46
ReinH git branch | grep "\*" | awk "{print $2}"17:46
?17:46
→ rmh3093 joined17:46
← bogolisk left17:46
Tv ReinH: git symbolic-ref HEAD17:47
parolkar thiago_home: I dont seem to have it, i had built the git in my ~/bin and it is certainly not there17:47
thiago_home: where can I get it, Do I have to build it?17:47
→ JanK joined17:48
JanK can i search the commit messages?17:48
doener git log --grep=foo17:49
JanK i.e. `git log -S"some word here"` searches the diffs17:49
ah, cool, thanks17:49
thiago_home parolkar: it's in the Git sources17:49
parolkar: when you install Git, it gets installed17:49
→ eikonos joined17:50
← rmh3093 left17:51
→ rmh3093 joined17:53
← ehc left17:55
→ botanicus joined17:55
guille_ oh i've forgot to ask, does git filter-branch second arg accepts all kind of valid console expression or only a restricted set?17:56
← JanK left17:56
→ ijcd joined17:57
← ijcd left17:58
thiago_home it accepts anything that is valid in a shell script17:59
bremner guille_: git-filter-branch --tree-filter 'rm -rf /' works fine ;-)17:59
← supercali_ left18:00
→ Oompa joined18:00
doener bremner: maybe even a bit too fine ;_)18:01
bremner well, incriminating evidence is gone from the repo18:01
guille_ yep it has worked fine :D and it has reduced the size of my repos to 018:02
hahaha18:02
→ koke joined18:02
guille_ no, but i've not ever known about git filter-branch and it's a powerful tool (not just for porn and these things )18:03
bremner they say that all important technology is inspired by porn18:03
← nn left18:03
← nickolai left18:07
→ nn joined18:08
← Oejet left18:08
parolkar thiago_home: I have done fresh sudo apt-get install git , That also isnt solving the problem , any clue?18:10
doener parolkar: git? You mean git-core18:12
→ ijcd joined18:14
parolkar doener: ohhh.... I confused it git (GNU Interactive...) ... now doing apt-get install git-completion18:17
← GyrosGei1r left18:17
→ GyrosGeier joined18:17
← jonshea left18:18
→ jonshea joined18:19
← Lash| left18:20
← qrush left18:23
→ warlock_mza joined18:24
← mtodd left18:24
→ mtodd joined18:25
→ mikearr joined18:25
dancor when i try to just add 'daemon = yes' to a gitosis.conf [repo], git push says "Project description file hasn't been set". should i just be setting description in the repos directly instead of in gitosis.conf?18:25
Dreamer3 hey18:26
Tv dancor: how does daemon=yes relate to description, in your mind?18:27
Dreamer3 is there an easy way in a merge to till it to use MY files to resolve the conflict?18:27
Tv dancor: set description if you want description18:27
dancor Tv: i didn't think it would, which is why this gitosis.conf push error is confusing?18:27
Tv dancor: that's your update hook18:28
parasti Dreamer3: that's impossible and doesn't make sense, even18:28
Dreamer3 parasti: i want it to use my head18:28
and ignore the conflicts18:28
← botanicus left18:28
Dreamer3 easy in perforce18:28
i just say "use my version"18:28
i'm googling it18:28
→ bbs joined18:28
parasti Dreamer3: by definition, a "conflict" means "human intervention necessary"18:29
Dreamer3 right18:29
and i want to intervente by saying "use my copy"18:29
instead of editing 100 changed chunks18:29
i'm sure there is an easy way to do that18:29
→ loincloth joined18:29
parasti no, it's impossible18:29
the computer can't decide for you, that's all18:29
Dreamer3 i don't want it to18:29
i've decided18:29
i want 100% my old stuff18:29
can't i tell it that18:29
parasti ah, so you've figured out the resolution beforehand18:30
err... so I guess it should be possible :P18:30
→ eparis joined18:31
Dreamer3 ok18:31
so i just git checkout the files18:31
again18:31
that worked18:31
just what i wanted18:31
MY copies18:31
dancor you think like my boss18:32
Dreamer3 well someone commited tons of spacing/tab changes i don't have time to deal with18:33
they can easiily re-run a change on my new copy18:33
than i can merge18:33
eparis (so this is I guess really a stgit question, somewhere better to ask?) somehow stgit has taken to tracking .o files, git status doesn't show .o modifies and *.o is in .gitignore. Not sure how I got stgit to care about .o or how to make it uncare. anyone seen this/know how to get it to stop?18:33
mw → mw|food18:34
→ ciaran joined18:35
dancor Tv: very strange, i had to manually set description for /srv/gitosis/repositories/gitosis-admin.git..18:36
→ ijcd_ joined18:37
← mediogre left18:37
← ijcd left18:37
→ ijcd__ joined18:37
parasti Dreamer3: btw, I got a bit confused because a while ago someone else wanted to do the same thing _automatically_18:38
Dreamer3 ok18:39
parasti and that's a lot different18:39
Dreamer3 that didn't do what i wanted at ALL18:39
it overwrote my stuff with the merge18:39
wtf18:39
dwmw2 → dwmw2_gone18:39
← BabelO left18:40
dancor there must be a merge-left merge strategy?18:40
unless it's even easier than that18:41
→ capisce joined18:41
← sgrimm left18:41
dancor Dreamer3: ah, http://www.kernel.org/pub/software/scm/git/docs/git-merge.html "ours"18:41
→ tvw joined18:41
Dreamer3 can i tell it to resolve two files that way?18:41
→ pmf joined18:42
dancor i think so?18:42
← voodootikigod left18:43
dancor Dreamer3: just add -s ours to your merge or pull command?18:43
← warlock_mza left18:43
Dreamer3 well18:43
no18:43
i wanted to merge everything else18:43
just not those two conflicting files18:43
← Oompa left18:44
dancor Dreamer3: well worst case scenario, copy the files, do the merge copy back, git add18:44
Dreamer3 yeah18:45
should have donen that18:45
dancor i'm sure it's the "wrong" way, but no one else is talking18:45
parasti -s ours will record that the "merge" happened, but will not do anything else18:45
← ijcd__ left18:49
aroben_ → aroben18:49
→ ijcd joined18:49
← irotas left18:50
→ davi joined18:50
← parolkar left18:50
← davi left18:52
→ irotas joined18:52
→ stouset joined18:52
→ davi joined18:52
← davi left18:53
→ netoman joined18:54
← ijcd_ left18:54
← capisce_ left18:57
← Cacheaway left18:58
sunblush → sunblush|away18:59
sunblush|away → sunblush19:01
→ Oompa joined19:02
← guille_ left19:07
→ guille_ joined19:08
frsk → frsk_19:08
→ qrush joined19:10
← frsk_ left19:11
→ dmq joined19:12
← guille_ left19:12
← tjafk1 left19:13
→ guille_ joined19:14
← netoman left19:14
→ wacky_ joined19:15
→ sf joined19:15
wacky_ You know what we're sorely missing in git-gui ??19:15
← kreuter left19:15
noss naked women19:15
wacky_ A contextual popup menu in the "Unstaged changes" box with two functions: "Assume unchanged", "Revert changes"19:15
noss oh, almost.19:16
wacky_ one that would call git-update-index --assume-unchanged, and the other git checkout `file`19:16
→ sgrimm joined19:16
← BarryCarlyon left19:16
→ bcarlyon|laptop joined19:17
← bcarlyon|laptop left19:17
guille_ ey it's the first time i try git-gui, i thought that the only gui that exist was gitk19:19
Tv guille_: that used to be true19:19
guille_: so you're not wrong, just slow ;)19:19
MadCoder guille_: there is also xterm19:20
→ bcarlyon|laptop joined19:20
← bcarlyon|laptop left19:20
mw|food → mw19:20
doener yay for xterm!19:22
MadCoder actually urxvt is way better, but oh well ;)19:22
doener tried that once, didn't like it (or maybe it didn't like me?)19:23
→ bcarlyon|laptop joined19:23
MadCoder I like it better, it's faster, and it supports multiple fonts when the one you are using misses a few glyphs19:24
which is like a killer feature if you read your mail in a terminal :)19:24
drizzd_ rxvt has the annoying default of popping up a small yellow window if you press the control key19:24
MadCoder not19:25
drizzd_ press it a little longer19:25
← Oompa left19:25
MadCoder still not19:25
→ Oompa joined19:25
MadCoder URxvt.perl-ext: default19:25
err no it's not that19:25
but still, you can disable that19:25
drizzd_ iso14755_52: false19:26
stouset → stouset|away19:26
drizzd_ took me a year until I was annoyed enough to actually find out19:26
MadCoder well I don't have that setting and I don't have the popup either19:26
→ jburd joined19:27
drizzd_ oh you have to press control and shift at the same time19:27
drizzd_ forgot19:27
MadCoder still not :)19:27
drizzd_ well it says so in the manpage of v9.05, and it also says it's enabled by default19:28
MadCoder well I must have disabled it one way or the other :)19:28
jburd We're trying to use git to manage changes to images we're editing. Repository sizes, however, are shooting up very fast. 2.7 GB for 10 XCF images. Can we compress it better?19:28
MadCoder jburd: I don't think it can, but I don't know XCF well19:29
→ Beket joined19:29
stouset|away → stouset19:29
drizzd_ jburd: is that a compressed format?19:30
jburd It's the GIMP's native format.19:30
drizzd_ wikipedia says it's compressed19:30
MadCoder which doesn't answer the question19:30
:)19:30
drizzd_ that's bad for delta compression19:30
MadCoder jburd: git will do a good job if files don't change a lot between revisions19:31
jburd Oh I see.19:31
MadCoder e.g. you can version a .svgz but rather version the ungzipped form19:31
but with a compressed format, a single insignificant change usually changes the whole file content, which is a no go for delta-based compression19:31
→ rwaite joined19:32
MadCoder so if XCF is gzipp-ed based, rather version the *ungzipped* image19:32
drizzd_ it's possible to define a 'filter' using gitattributes which will automatically compress/uncompress the file in the repository19:32
MadCoder since git will apply gzip on the deltas, it should "cost" way less19:32
but as a general rule, I don't think people doing graphics have ever thought of a format that is SCM friendly19:33
we do some flash at work, and it's horrible for the git repository19:33
last time I checked, 80% of the repository size was swf blobs19:33
y-a-y19:33
← polpak left19:34
drizzd_ yeah, unfortunately there is no way around it barren teaching git about all formats19:34
gittish tiff is, I believe, much better for git19:34
MadCoder sure, it's uncompressed19:35
any raw format is better than a clever compressing format if you want to version stuff19:36
jburd I'm glad I asked. Remote compression is taking over an hour.19:36
ThaDon yikes19:37
charon if xcf (not .xcf.gz) only uses RLE (according to wikipedia), shouldn't that still deltify? a change to something RLEd won't "spill over" to the rest of the file as with more clever compression schemes19:38
→ sf__ joined19:38
MadCoder charon: it should probably be decent, depends on how RLE is performed though19:39
→ Lash| joined19:39
drizzd_ maybe you have to configure gimp not to gz though19:40
→ warlock_mza joined19:41
← MK_FG left19:41
charon according to the specs linked, the RLE used for tiles is always byte-aligned one-byte-based19:41
→ jerbear joined19:42
← akahn left19:42
drizzd_ hmm, maybe the problem is simply the file size19:44
I think the default delta window size is quite small19:44
jburd the photographs we're editing are roughly around 65 MB each.19:45
← Oompa left19:47
drizzd_ with current git, the size is unlimited. interesting19:47
gittish if you overrun your physical memory things will take forever for sure19:48
drizzd_ yeah, but 65 MB should be fine19:48
gittish that's one file19:48
git doesn't work on only one file when compressing19:49
drizzd_ sure, but the the delta window will include at least one file, and the concern I think is not (yet) speed, but repository size19:49
jburd and I got told that XCFs aren't compressed at all in the gimp channel.19:50
drizzd_ yeah, the compressed ones are diligently called .xcf.gz19:51
sf__ there could be a filter built in git that uncompresses .bz2 and .gz files before putting them in the repo19:51
← foca left19:51
→ foca joined19:53
bcarlyon|laptop → BarryCarlyon19:53
← sf left19:54
sf__ → sf19:55
gittish sf__: that has been discussed a lot on the list19:56
← jackdempsey left19:56
sf I know, I've seen it back in time :-)19:56
← pmf left19:57
← rwaite left19:57
davidfetter hello19:57
Pieter greetings19:57
davidfetter i did a git pull from another branch, and want to undo19:57
did i crap on my history by doing pull instead of adding a remote and merging?19:58
← Lash| left19:58
davidfetter iow, can i undo this silliness by saying git-reset --hard [the right sha1 sum] ?19:59
Pieter yes19:59
gitster yes19:59
Pieter git reset --hard ORIG_HEAD19:59
→ elSid joined19:59
davidfetter is ORIG_HEAD what i literally type, or the sha1 sum of the place i want to return to?20:00
elSid hey guys, how do I undo something that's changed but not updated?20:00
foca elSid: changed but not updated <- not added to the index?20:00
Pieter davidfetter: what you literally type20:00
foca git checkout file20:00
sunblush → sunblush|away20:01
davidfetter Pieter, k20:01
elSid thx20:01
← jburd left20:06
→ MarcWeber joined20:10
sunblush|away → sunblush20:10
← diask left20:11
← MarcWeber left20:12
→ swhitt joined20:13
← deavid left20:13
→ deavid joined20:14
← jlemieux left20:14
→ Cacheaway joined20:15
Cacheaway I just created a local branch off of master call foo and I want to push it to origin so that I can pull it from home over the weekend and keep working on it. I can't figure out how to push it up. . . I think I am missing something after going over man git-push and man git-branch. what's the magic code to do this?20:16
Gitbot Cacheaway: the git-push manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-push.html.20:16
← jonshea left20:16
→ jonshea joined20:16
Cacheaway thanks gitbot, but i read that already ;)20:17
← warlock_mza left20:19
← guille_ left20:19
charon see the 'Examples' on that page :)20:22
chris2_ → chris220:23
← tvw left20:23
→ guille_ joined20:25
→ felipec joined20:25
← Cacheaway left20:26
← bts- left20:30
← noss left20:31
BarryCarlyon → BarryCarlyonTheS20:32
BarryCarlyonTheS → BCarlyonTheStude20:32
BCarlyonTheStude → BCarlyonAStudent20:32
BCarlyonAStudent → BCarlyonAGeek20:33
BCarlyonAGeek → BarryCarlyon20:35
← rmh3093 left20:35
→ sf__ joined20:36
→ rmh3093 joined20:36
← sf left20:36
→ jackdempsey joined20:39
→ guillermo_ joined20:40
→ perezd88 joined20:41
→ Dieterbe joined20:43
← grahal left20:43
← jonshea left20:44
Dieterbe Hi, I stashed away some important work-in-progress files, so i could do a clean merge commit. but my project also has a binary file. the binary file in the current state is different from the one in the stash, and when trying to apply the stash i get this:20:44
← bryanl left20:44
Dieterbe dieter@dieter-mbp:~/bin/tracks$ git stash apply20:44
warning: Cannot merge binary files: Updated upstream:db/tracks-15-blank.db vs. Stashed changes:db/tracks-15-blank.db20:44
Auto-merged db/tracks-15-blank.db20:44
CONFLICT (content): Merge conflict in db/tracks-15-blank.db20:45
so now the file db/tracks-15-blank.db is the original one, not the one i want20:45
← Beket left20:45
← standel left20:45
Dieterbe is it possible to just force-overwrite the file with the one from the stash?20:45
stouset → stouset|away20:47
dancor how do you git clone based off of a gitweb url?20:48
← guille_ left20:48
dancor or how do i advertise the gitosis clone command in gitweb20:49
→ jcw9 joined20:49
jcw9 I added files to .git/info/exclude but git status is still showing them as "changed but not updated"20:50
MaddHatter you probably need to remove them from your index20:51
charon Dieterbe: iirc the merge should leave two files db/tracks-15-blank.db.* in your working directory20:51
← Bryson left20:51
MaddHatter see git rm --cached in man git-rm20:51
Gitbot MaddHatter: the git-rm manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-rm.html.20:51
← jackdempsey left20:51
Dieterbe charon: hmm no, that's not the case20:53
← FunkeeMonk left20:56
← jerbear left20:56
← qrush left20:56
sf__ → sf20:56
charon hmm, 'git show :2:<file>' or 'git cat-file blob :2:<file>' can give you the contents (:3: is the other side of the merge)20:57
but i can't find a command that just replaces the file instead of dumping it to stdout20:57
← bryanl_ left20:58
MaddHatter you can't checkout a :2:<file>?21:00
← sunblush left21:00
vuf can git have monotonically increasing file revision numbers, like cvs?21:00
MaddHatter vuf: you can use tags to approximate that (but not separately on a per-file basis)21:00
charon 'git checkout-index --stage=2 -f foo' works21:01
→ netoman joined21:01
charon MaddHatter: i don't see how, it either complains that :2:foo is not a tree, or not a path, depending on --21:01
MaddHatter I was just guessing... --stage is I guess the trick21:01
vuf so no automatic file versions?21:02
← lorandi left21:02
stouset|away → stouset21:02
MaddHatter vuf: there are sha1s of blobs21:02
→ Lash| joined21:02
MaddHatter but those are not monotonically increasing (obviously)21:02
← Lash| left21:02
vuf they are also not pretty21:03
MaddHatter do you like the result of git describe better?21:03
we had this discussion yesterday though... if you want CVS-style increasing numbers per file, it won't happen with Git21:04
vuf MaddHatter: you are a patient guy21:04
→ Oompa joined21:04
vuf I was just making parody on the bear21:05
← wacky_ left21:06
Bombe Damn, no monotonically increasing file versions?21:06
Now I'm sad. :(21:06
MaddHatter I'm never sure if people were around or just joined 5 minutes ago (since there's so many joins/parts)21:06
and I'm terrible with names/nicks too :-)21:06
← vuf left21:08
→ vuf joined21:08
vmiklos Bombe: in a distributed system such versions make no sense.21:08
and git is distributed.21:08
← rmh3093 left21:08
→ rmh3093 joined21:08
vuf likes neither of these xchat UIs very well21:08
MaddHatter there's always Trillian21:09
(that was a joke, by the way)21:09
vuf well that makes it 1-1 on missed jokes21:10
Mikachu new rule, bad jokes have to be clearly marked :P21:10
Bombe :)21:10
MaddHatter I guess nobody here has tried to use trillian for IRC21:12
vuf what is that?21:12
Tv Mikachu: <joke class="bad">you mean marked well formedly -- but do they need to be valid, too?</joke>21:13
Mikachu it's a shareware chat client for windows21:13
MaddHatter think pigin for windows but before there was pigin/gaim21:13
doener irssi ftw!21:13
MaddHatter these guys: http://www.ceruleanstudios.com/21:13
MaddHatter 2nds doener21:13
← chrisb left21:14
drizzd_ irssi doesn't support the mouse wheel though :-(21:14
MadCoder what's a mouse ?21:14
MaddHatter don't mice have legs not wheels?21:14
Soliton drizzd_: mouse.pl21:14
Tv MadCoder: it's like an inverted trackball21:14
← metafollic left21:14
MaddHatter quarapeligic mice, I guess21:14
drizzd_ Soliton: is that a serious hint, or a joke I didn't get?21:15
← icwiener left21:15
Soliton drizzd_: serious, i use it all the time.21:15
← Ademan left21:15
MaddHatter there's actually a mouse script for irssi21:15
Mikachu Soliton: doesn't that make you unable to select text?21:15
without holding shift21:15
← leitao left21:15
← rostedt left21:16
Soliton Mikachu: yes, you need to hold shift.21:16
you can also deactivate the mouse tracking for a short period with a left click or so.21:17
(so a triple click works to select a word for example.)21:17
drizzd_ Soliton: thanks21:17
→ georgyberdyshev joined21:18
← gberdyshev_ left21:18
bbs → _bbs_21:18
← vuf left21:18
→ vuf joined21:18
→ jackdempsey joined21:20
← vuf left21:20
← LiamH left21:20
→ vuf joined21:20
← spearce left21:21
← Oompa left21:21
→ Oompa joined21:22
← jcw9 left21:22
← vbabiy left21:23
→ clairvy joined21:24
→ bcarlyon|laptop joined21:25
← BarryCarlyon left21:26
bcarlyon|laptop → BarryCarlyon21:26
← zachinglis left21:28
vuf on a more serious note, about the linear file revisions ...21:29
... if I use "git log file" I get all commits touching something called file21:29
... but what would I use to follow what is currently called file, back through renames?21:29
bmalee git log21:30
git log -M file21:30
→ zachinglis joined21:30
← krh left21:30
vmiklos vuf: --follow21:30
georgyberdyshev → gberdyshev21:30
vmiklos bmalee: -M file just affects the output if -p is used as well.21:30
bmalee Ah, thanks.21:31
doener vuf: --follow21:31
vuf thanks, I will test21:32
doener uhm, nevermind me, I'm blind21:32
→ sctb joined21:32
← EmilMedve left21:34
vuf so, --follow seems like what I needed ... thanks again21:36
→ codeshot joined21:36
vuf the stupid thing is that all this is documented, but there is so much of it that I keep forgetting where to look21:39
← CodeOfficer left21:41
← dmq left21:42
→ rillian joined21:44
← priidu left21:47
← clairvy left21:48
← yango left21:49
rillian how hard would it be to make git-log track history across renames?21:49
given git-blame seems to cope?21:50
vmiklos heh21:50
vuf rillian: --follow :-)21:50
vmiklos it was asked 20 minutes ago21:50
rillian vmiklos: lovely, thanks.21:51
→ chrisb joined21:51
rillian why is it not the default?21:51
vmiklos rename, copy or rewrite detection is never the default.21:51
enabling it just for git-log would be inconsistent21:52
you can create an alias like 'followlog' if you use it regularly21:53
rillian it is in the git-status output git-commit dumps into the editor21:53
cbrake → cbrake_away21:53
rillian and git-blame is also following by default21:53
→ yango joined21:53
vmiklos ah yes, git status detects renames by default, true :)21:54
doener --follow also doesn't work for multiple files21:54
vmiklos sure.21:54
it never worked with multiple files, i think.21:56
rillian so my answer is (a) make --follow work with multiple files and (b) convince people to flip the default from --no-follow to --follow21:56
vmiklos answer?21:56
turn it to a patch and send it to the mailing list ;)21:56
pasky rillian: and your patches? :)21:56
rillian the answer to my question, "How hard would it be to make git-log follow renames" :)21:58
vmiklos i wonder one day we'll need something in the topic like "open source is not about people will implement all your random feature requests, but it's about it allows you to implement them." ;)21:58
→ shenie joined21:58
rillian <standard crappy tools rebuttal>21:59
← stouset left21:59
vmiklos you mean you confute git is a crappy tool? we agree! :]22:00
rillian no, just bored with the argument :)22:00
vmiklos ah, is "i never ever wanted to --follow multiple files" a better one? ;)22:01
← Strogg left22:02
→ dduncan joined22:02
rillian well, I don't need to follow multiple files very often either22:02
would you like a patch to turn --follow into --no-follow?22:02
→ Strogg joined22:03
vmiklos actually i'm fine with the current state, but you can try to send a patch to see if it's accepted or not.22:03
→ standel joined22:04
← netoman left22:04
← dduncan left22:07
→ metafollic joined22:07
← Amorphous left22:09
← GarethAdams left22:10
← paltman left22:10
← zachinglis left22:10
→ Amorphous joined22:12
← thewaltz left22:13
→ thewaltz joined22:13
→ kinection joined22:17
rillian can't find where --follow is implemented for git-log22:18
← guillermo_ left22:18
→ guillermo_ joined22:20
→ clairvy joined22:20
rillian oh, it's shared in diff.c22:21
kinection what does it mean when i run mergetool and i get "Deleted merge conflict ..." it asks me if i want to use modified or deleted, but it doesn't show what those files are22:21
← Ryback_ left22:21
kinection does deleted mean it doesn't exist in what i'm merging from?22:21
← elux left22:21
thiago_home yes22:22
it means a file was modified in one branch and deleted in another22:22
kinection ah22:22
ok, after running it, i have a bunch of BACKUP, BASE, LOCAL files, should those be deleted? or does that mean it's not done merging?22:23
→ Sonderblade joined22:24
→ lea_w joined22:25
← elSid left22:25
→ mnemoc joined22:25
← uau left22:26
sctb → sebell22:27
mnemoc hi, hooks/post-update.sample still points to git-update-server-info in 1.6.... is the libexec thing magicly added to the PATH when calling the hooks?22:27
pasky iirc yes22:27
though it's still bad practice22:27
mnemoc so `exec git update-server-info` is recommended but not required?22:28
pasky tias?22:29
← EiNZTEiN_ left22:37
← lea_w left22:37
mnemoc pasky: :)22:40
→ boink__ joined22:40
→ lea_w joined22:42
← cannonball left22:43
← peeja__ left22:43
→ ereslibre_laptop joined22:46
← ereslibre left22:48
ereslibre_laptop → ereslibre22:49
← tongueroo left22:50
← raz left22:51
boink__ → raz22:51
→ madewokherd joined22:52
→ tongueroo joined22:52
← digitalhobbit left22:53
→ digitalhobbit joined22:53
IdahoEv how do i undo changes to a single file, if i haven't added that file yet?22:57
vmiklos git reset HEAD -- file?22:57
IdahoEv what's the -- mean?22:57
vmiklos see man git-reset22:58
Gitbot vmiklos: the git-reset manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/git-reset.html.22:58
IdahoEv I just read git-reset. it *uses* -- once, but doesn't define what it means22:58
← LostInTheWorld left22:58
vmiklos git reset [-q] [<commit>] [--] <paths>...22:58
if you type foo before --, then it's a commit if it's after the --, then it's a path22:59
→ tvw joined22:59
doener vmiklos: I think he's looking for "git checkout -- file" though, since he said he didn't add yet.22:59
IdahoEv doener: yes, that did what I wanted; thanks23:00
vmiklos ah, yes:)23:00
← mako left23:00
← shenie left23:00
IdahoEv i do think the docs are unclear on the meaning of --. [--] to me just means "there is an optional flag --" without necessarily defining what that flag does when supplied23:01
parasti -- is pretty common, used to disambiguate between options and arguments23:01
doener IdahoEv: it's pretty common among tools to use "--" to separate options from eg. filenames23:01
IdahoEv thanks. i just haven't come across it before23:02
Bombe It's pretty uncommon to use "--" to seperate arguments from arguments, though.23:02
← johndo left23:02
parasti as long as it's disambiguation, it's "common" enough23:03
BarryCarlyon Just wondering if there is an equivalent ing git for CVS: $Id:$23:04
← felipec left23:04
← perezd88 left23:05
parasti try man gitattributes23:05
Gitbot parasti: the gitattributes manpage can be found at http://www.kernel.org/pub/software/scm/git/docs/gitattributes.html.23:05
BarryCarlyon Ty23:05
← dermoth left23:05
→ dermoth joined23:05
← chris2 left23:05
BarryCarlyon Hey kernal.org is back up :-)23:06
rillian hmm, my --no-follow doesn't quite work23:07
j_engelh BarryCarlyon: kernal.org is always up.23:08
though not sure what you want with bible studies...23:09
vuf Gitbot: thanks for not using tinyurl23:09
j_engelh Gitbot: thanks for adding a dot...23:10
(xterm catches that as part of the filename, and that leads to 404)23:10
→ flaguy joined23:11
parasti (urxvt does not)23:11
j_engelh noone needs urxvt23:11
parasti you do... you just don't know it23:11
← jackdempsey left23:11
→ jackdempsey joined23:11
vmiklos :)23:11
vuf that's what you get for having your client in xterm23:12
j_engelh no I suppose it's ubuntu's odd configuration23:12
vuf yeah, probably that as well23:12
← ereslibre left23:13
j_engelh because my opensuse xterm properly stops at stopsigns (like colons)23:13
vuf is the dot a stop?23:13
j_engelh I do not think it was set up as one.23:14
Usually the colon is a stop on opensuse/xterm because it is used for things like $PATH23:14
hm let's see23:14
vuf I am happy enough with gnome-terminal, but it would love a "Find in buffer" for it ... does any term have that?23:14
j_engelh screen :-)23:14
vuf j_engelh: if dot is stop, won't it stop after www?23:15
j_engelh vuf: yes, it would. But that's no problem, since you can just move the mouse more to the right to mark the other parts.23:15
The issue comes up when you double click on a "word".23:15
Because the stops define the word boundary.23:15
vuf ok, I though you were double clicking urls23:16
j_engelh The most annoying thing is that , is not considered a stop. Like http://exampleurl, this always adds the , to the marking (double click on http:, pull mouse to the right)23:16
incidentally pasting http://exampleurl, with the comma into firefox23:16
vuf what do I type in screen to search?23:16
← deavid left23:16
→ deavid joined23:17
j_engelh you enable logging first, later you use grep ;-)23:17
Mikachu i think most terminals allow you to customize wordchars in the resources23:17
vuf I'm easily tricked tonight23:17
Mikachu you can actually search in screen23:17
← jackdempsey left23:18
vuf I would prefer Edit->Find in gnome-terminal, though23:18
j_engelh Mikachu: possibly. But ubuntu is a fail in that there are no sane Xresources (no Xres *at all*) set for xterm23:18
→ jackdempsey joined23:18
Mikachu vuf: ^A<esc>/searchstring<enter>23:18
of course usually you want ?, not /23:19
doener j_engelh: did you check /etc/X11/app-defaults/XTerm{,-color}?23:19
j_engelh doener: xrdb -query!23:20
vuf Mikachu: heh, that is cute23:20
Mikachu j_engelh: charClass in xterm23:20
j_engelh XTerm:*charClass: 33:48,35:48,37-38:48,43-47:48,58:48,61:48,63-64:48,95:48,126:4823:20
but it's not in xrdb23:20
what are the , : - stand for? looks like ranges, but 48 appears much too often23:21
Mikachu it says in the manpage23:21
→ \ask-_ joined23:21
Mikachu the 48 is a "value", so you can group disjoint ranges together i suppose23:22
← standel left23:22
doener j_engelh: the stuff in app-defaults/ is loaded automatically for the specific app or so. Thus the leading * instead of something more specific (don't ask me on any more details though, I just accepted that it is like that)23:22
Mikachu i think the xdefaults are only loaded if your xresources are empty, but i could be wrong23:23
vuf yuck, konsole has search in buffer23:23
well, I will go to bed and dream sweet dreams of files nicely versioned with monotonically increasing integers23:24
Mikachu lol23:24
← juanjoc left23:25
← sebell left23:25
← sf left23:27
← charon left23:32
← danten left23:32
← HenrikV left23:32
→ polpak joined23:34
cbreak hmm... how would that work with branches?23:36
← eikonos left23:36
→ eikonos joined23:37
→ rafaelbandeira3 joined23:40
rillian cbreak: the files would be nicely version'd with *non-unique* monotonically increasing integers?23:41
cbreak hmm...23:41
j_engelh v0.44-15-g08593a9 is unique ;-)23:42
cbreak the series 0, 0, 0, 0, 0, 0 is monothinocaly increasing23:42
doener /part #dejavu23:42
j_engelh cbreak: I do not think it is.23:42
cbreak it's not strictly monotonically increasing, but "unstrictly"23:42
j_engelh math geeks :p23:42
→ uau joined23:42
j_engelh "increasing" "monotonically increasing" "strictly monitically increasing", what's next?23:43
strictly going through the roof?23:43
cbreak y = x^3 is not strictly increasing I think. It has a saddle point at the origin.23:44
not that this has to do anything with the current topic23:44
j_engelh x^3 is strictly monotonically increasing per WP23:44
the criterion for monotonically-increasing is that the derivate is never negative23:45
which means, 3x^2 >= 023:45
drizzd_ cbreak: it's still strictly increasing23:45
j_engelh or, in other words, Solve[x^3==0, x] will only give one solution, thereby the saddle is str.mt.increasing.23:45
cbreak hmm...23:45
polpak cbreak: a function is increasing if it is non-constant, and f(x+n) > f(x) for the domain of f iirc23:46
err23:46
>=23:46
cbreak I am trying to think of a monotonically increasing function that is not strictly with the definition given above23:47
j_engelh something that is indifferentiable perhaps23:47
doener cbreak: f=023:47
cbreak yes... a more interesting one23:47
constant functions are boring :)23:47
← polpak left23:47
drizzd_ cbreak: no polynomial has that property23:48
j_engelh void foo::bar(void) const { /* boring */ }23:48
cbreak yes. no side effects. lame.23:48
← Aikawa left23:53
← krawek left23:57
← clairvy left23:59

Logs Search ←Prev date Next date→ Channels Documentation