IRCloggy #git 2007-04-25

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.

2007-04-25

→ sgrimm_ joined00:01
aroben mugwump: did my email go through to the list?00:06
mugwump: I have a feeling it didn't00:07
→ DrNick joined00:13
jrockway mugwump: eventually i'd like to be able to graft people's svn histories into these repositories00:21
so many people abandon modules and are never heard from again; i would like to be able to fully take over their project in that case00:21
(i being more people than just me of course ;)00:21
→ sgrimm joined00:32
→ aroben_ joined00:40
→ orospakr joined00:41
→ sgrimm joined00:42
orospakr Hey, what's the story on the Windows Shell extension for git?00:42
because that would be highly awesome.00:43
→ tango_ joined00:44
mugwump aroben: doesn't seem to have made it... hmm00:46
→ aroben_ joined00:46
tango_ hello all. I've strated using git somewhat more seriously these days and I've just hit the first ... nuisance :) basically, my smtp server doesn't like git sending all those mails one after the other. Would it be possible to add an option to git-send-email to sleep a configurable amount of time between emails?00:50
aeruder_ tango_: decent idea00:51
tango_ thanks :)00:51
tcoppi tango_: it looks like a perl script, so it should be easy to change00:51
jrockway shouldn't SMTP handle that for you?00:51
jrockway goes to look at the code anyway00:51
tcoppi hey look, theres a subrutine called send_message :)00:52
mugwump for x in 000*; do git-send-email --to ... $x; sleep 40; done00:52
tango_ tcoppi, possibly. it's just that at 2am I don't feel like learning how new options are added in git (sendemail.sendsleep ?)00:53
mugwump, which is good unless you use --compose00:53
aeruder_ i think a command line option would be good enough00:53
tango_ aeruder_, I think an option would be better: you want to set it once and forget about it, and it goes with the smtp settings00:54
aeruder_ fair enough00:54
aroben_ mugwump: any clue why my email didn't go through to the list?00:55
aroben_ → aroben00:55
jrockway bad net connection? ;)00:56
mugwump aroben: did it get accepted by the vger mail server? you'll need to check your MTA logs00:58
aroben mugwump: ah, yes, that's the problem00:58
mugwump: the firewall here doesn't like people doing things like that00:59
mugwump usually it's something like the return address isn't deliverable00:59
just be careful using your normal mailer, might do nasty things to whitespace in the diff00:59
aroben mugwump: I resent it, hopefully without munging it01:01
aeruder_ tango_: if you stick around for a sec, i'll send you a patch to test if possible01:02
tango_ sure01:02
→ aroben_ joined01:05
tango_ aeruder_, I won't be able to test it now though, I don't have big patchests to send01:05
aeruder_ tango_: eh, you can send some to yourself or something, just want another tester ;)01:06
tango_ oh, I can do that :D01:07
aeruder_ ok, let me test real quick, and then you can test and then i'll send to list01:10
→ yashi joined01:14
aeruder_ ah, crap01:18
i sincerely hope that those emails i just tested with really didn't get sent01:19
heh01:19
tcoppi :)01:19
aeruder_ i thought i'd just test send the last 6 git commits01:20
forgetting that it'd add all the CC's automatically01:20
tango_ lol01:20
anyway01:20
aeruder_ tango_: ok, i got a patch for you if you want to try it01:20
tango_ oh01:21
good01:21
because I have to go to bed :)01:21
aeruder_ at least in the process of potentially spamming the CRAP out of the git list and random other people01:21
tango_ link?01:22
aeruder_ http://dump.aeruder.net/sendsleep.patch01:22
that's against latest git, but i doubt you'll have a problem01:23
try the config option if you would (sendemail.sendsleep)01:23
i didn't test that part01:23
i did test the command line option (--sendsleep #)01:24
aeruder_ looks to see if he can find out where those emails disappeared to01:25
aeruder_ i don't want them queued up and sent out later, hehe01:25
tango_ aeruder_, it's not enough, it seems01:31
aeruder_ i.e. the sleep works but it doesn't fix your problems?01:31
tango_ yep01:31
aeruder_ boo01:31
tango_ MAIL FROM:<> [84.221.15.247] blocked - Too much traffic from this address.01:31
aeruder_ wow, you're just hosed01:31
probably think you're a spammer :)01:32
tango_ yeah :D01:32
would it be possible to do something different? like, catching this message and sleeping progressively more?01:32
I wonder if it depends on me using a gmail from address even though I'm using my isp smtp ...01:32
aeruder_, does git-send-email support tls connections?01:33
aeruder_ hm, not that i know of unfortunately01:33
segher_ your best bet would be to get a sane isp :-)01:34
tango_ lol01:36
not easy. tiscali is the best adsl one can get in italy, even though it has this annoying problem. other isps don't have this problem, but their services suck. and I don't want to go cable (fastweb)01:36
segher_ i've never had mail problems like this with any other isp in europe. and yes i've had ridiculous bounces from tiscali before, and unexplainable blackholes01:39
they just don't understand email i suppose01:39
tango_ which is quite a pitn01:41
ok guys, thanks for your help, I'll go to bed now and sleep it over01:41
'nite all01:41
jonesy n00b question: does it matter whether I base a branch on 'origin' or 'master'? Is one of them preferred?01:41
what's the difference?01:42
mugwump if they point to the same commit, no difference01:43
they're just labels01:43
jonesy ok.01:43
thanks.01:43
aroben_ → aroben01:45
aroben mugwump: am I supposed to do anything now that Eric acked my patch?01:49
mugwump well, tell us how you get on with it :)01:49
Junio or Eric will pull it into their branches and it will probably hit the next maint01:50
aroben mugwump: ok, great01:50
mugwump: that's what I assumed01:50
mugwump: thanks again for all your help!01:53
aeruder_ ok, i'm done playing around with git-send-email01:56
that time a completely unintended patch email DID get sent :)01:56
← xenon left02:27
→ CIA-13 joined02:38
→ matled joined03:00
orospakr damnit, why does git 1.5.1 keep trying to install itself in ~/bin?! I set PREFIX to /usr/local, just as INSTALL said.03:02
segher_ set it while running "make install" too03:03
orospakr I did.03:04
segher_ hrm, no idea then03:04
orospakr PREFIX=/usr/local/ sudo make install03:04
segher_ sudo PREFIX=/usr/local make install03:04
aeruder_ sudo make prefix=/usr/local install03:04
segher_ sudo starts a new shell...03:05
aeruder_ the line i put is what you want03:05
segher_ yah03:05
→ yashi joined03:20
→ CIA-14 joined03:38
→ rkaway joined04:28
→ mithro joined04:48
→ praka joined05:17
→ sgrimm joined05:20
→ praka joined05:25
→ robfitz joined05:33
→ tbf joined05:53
tbf has a big wish for git-svn05:53
tbf i'd like it to output percentage information when printing such lines during fetch: "r1464 = fd280e8edd7a3afa54664857ad7e647143253874 (trunk)"05:54
would do the patch myself - it i'd easily understand the perl code of git-svn...06:01
mithro tbf: what is git-svn?06:20
aeruder_ mithro: allows you to clone svn repositories and fetch/commit against them06:24
mithro ahh06:24
→ etnt joined06:24
tbf mithro: as aeruder_ said. cool stuff. degrades svn repos to data source, allowing dist scm everywhere06:27
s/data source/data sources/; s/data sources/data graves/06:27
mithro i was hoping it was like git-cvs06:28
aeruder_: currently in the process of joining the mailing list so that I can post the bug report about mmap06:29
aeruder_ mithro: ah, i noticed i hadn't seen that posted up yet :)06:31
mithro aeruder_: majordomo is giving me greif :/06:31
→ meyering_ joined06:32
mithro mailman as spoilt us all06:32
← meyering left06:33
meyering_ → meyering06:33
→ devogon joined06:37
→ GyrosGeier joined06:49
mithro aeruder_: email sent06:52
→ aroben joined06:53
mithro aeruder_: hopefully somebody has a solution :P06:53
→ aroben joined06:56
pigeon hmm does git tracks permission changes?07:40
file permissions/modes that is07:50
aroben pigeon: I believe it at least knows which files are executable07:52
→ matled joined07:56
mithro pigeon: it tracks file modes07:57
pigeon: see http://git.thousandparsec.net/gitweb/gitweb.cgi?p=tpclient-pywx.git;a=commit;h=1a1e14a6ae54f4235affa983a3456ff37d763c3b for an example :)07:57
pigeon hmm, it only seems to track whether it's executable then.07:59
doesn't seem to care, say, from 644 to 60008:00
→ matled- joined08:00
aroben pigeon: the link mithro gave shows two files going from 0644 to 075508:00
pigeon yeah, i'm trying it here, it ignores 644 -> 60008:01
→ Simon- joined08:02
matled- → matled08:03
→ Simon-_ joined08:03
→ robtaylo1 joined08:04
pigeon looks like it ignores group and world permissions08:05
Simon-_ → Simon-08:05
mithro pigeon: [changed mode: 0644->0755]08:05
→ winter-mute joined08:06
etnt hi, stupid question: for those large open-src projects using git, what is the common practice for deciding what goes in a release, and how is the actual release repo handled ?08:06
pigeon well, at least for my version of git here, which is 1.5.0.6, it only cares owner's permission it seems08:06
etnt I mean for the linux-kernel I guess everything goes via Linus08:07
pigeon i tried 644 -> 600, 644 -> 666, 644- > 677, etc, git doesn't care.08:07
etnt I guess I'm still stuck with my svn damaged brain...08:07
anders_ etnt: Not sure if we count as a large open-src project... But we have http://git.xmms.se/merge/ that gives an overview of what the different trees contain and helps showing what to merge.08:08
mithro wouldn't that be what you want?08:09
pigeon: ahh08:09
etnt anders_: ok, and how (who) picks what goes into a new release?08:09
I guess xmms2-stable is the 'release' branch , or ?08:11
anders_ etnt: Well, everything our main development tree (xmms2-devel which that page shows differences to) goes into the release. And what commits that are imported into -devel is discussed on irc mostly.08:12
mugwump is über-lagged08:12
→ mithro_ joined08:12
anders_ etnt: xmms2-devel works much like linus tree. Everything just in a much smaller scale.08:13
mithro_ → mithro08:13
mugwump etnt: look at the way the git project works. there are several branches08:13
first, there are the multitude of branches where people are working privately.08:13
second, once a change is considered worth evaluating by a wider developer audience, it goes into "pu" (proposed updates)08:13
patches may be removed from proposed updates08:13
once it is considered worth going ahead with, it goes into "next"08:14
→ dkagedal joined08:14
mugwump finally when "next" passes integration testing it is merged to "maint"08:14
"maint" changes are bugfixes, and increment the last digit of the version, eg 1.5.0.108:15
(when they are tagged, of course)08:15
mithro git has a SHA1 for each individual patch as well as one for the repository? is that correct?08:15
mugwump when "next" hits "maint" this is usually a third-level or second level change08:15
mithro: the repository itself has no SHA108:15
etnt ok, so is it one person who 'owns' for example the 'pu' branch ?08:16
mithro mugwump: so how do I know if an identical patch is committed to two different repositories?08:16
mugwump so, things to note about the process:08:18
- patches are never removed from "next", so you can safely merge it. they might be reverted, though08:18
- only bugfixes and well tested releases go into "main", so you can safely merge that without worrying about introducing features08:18
"maint" should be08:18
the release manager can then decide what goes in the release based on what is stable08:18
sure, one person owns each branch08:19
mithro: well, if the patch is exactly the same, in the same point in history, the SHA1s match08:19
etnt ok, I think I understand :-)08:20
mugwump it the patch is exactly the same, at a different point in history, then git-merge-recursive will notice and not try to apply twice08:20
mithro mugwump: but if there where different revisions before, there is no way to check they where the same?08:20
mugwump git-cherry goes some of the way08:20
etnt anders_: that web page, is it using the gitweb package ?08:20
anders_ etnt: No, that is a homebrew script running from cron.08:21
mugwump there is no patch UUID like in bzr that allows you to track patches that have been applied in different orders, instead they are spotted based on the actual changes they make08:21
which has its benefits - if someone picks a patch, merges it badly but commits it, you didn't just lose the original change08:23
caveat emptor08:23
if someone picks a patch, merges it well, you probably have to re-merge it08:24
(without git-rerere)08:24
→ Dodji joined08:26
→ yashi joined08:30
→ yorgen15 joined08:44
→ yorgen15 joined08:47
→ winter-mute joined08:51
→ nud joined08:52
tbf wonders if git-svn does a full checkout when finding a new branch....09:02
tbf ...or is it smart and just clones the branch's origin?09:02
....which should save quite some bandwidth09:02
→ winter-mute joined09:03
→ tchan joined09:16
eMish Is there 'git unlock' command that removes the lock ?09:22
like in svn ?09:22
like when a process crashed or was killed in the middle leaving the lock behind ?09:22
→ GyrosGei1r joined09:24
eMish I have bare repo X, and two clones A and B, same file xxx was modified in A and in B. In A trying 'git-pull origin', I get 'xxx: needs update'. Whcih commands do I use not to merge xxx on A ?09:29
Tali eMish: just commit this file.09:31
→ pdmef joined09:32
→ sgrimm joined09:32
→ tcoppi joined09:32
→ mspang joined09:32
→ puzzles joined09:32
→ kblin joined09:32
→ nickh joined09:32
→ winter-mute joined09:32
Tali eMish: if you don't want to commit it just now, then commit it to some throw-away branch, merge your updates to the normal working branch, then merge your temporary branch to get the file merge and remove the last merge without touching the working directory (with git-reset)09:32
or just revert the changes if you are not interested in them any more09:33
→ tchan joined09:35
tbf shouldn't "git clone" include branches?09:50
Tali tbf: you will get branches in the remote repository as "remotes" in the local one09:52
tbf: try git branch -a09:52
tbf Tali: ah! ok.09:52
Tali tbf: you can create new working branches based on these if you need to09:53
→ eMish joined09:56
→ Sho_ joined10:03
→ jeffpc joined10:27
→ eMish joined10:49
→ llnz joined10:54
→ yorgen15 joined10:54
llnz i commited a patch with the wrong author information and pushed it to another repo, and I now need to change the author details (in both)10:56
how do i go about it?10:56
Randal that's part of the SHA110:56
so you'll have to commit a new commit, and push that10:56
and then remove it from the other repo too10:56
llnz ok, so cg-admin-uncommit to remove the old patch?10:57
Randal I don't use the cg tools10:57
eMish how do i print name of the branch in which i am working ?10:57
Randal git-status shows the current branch10:57
amongst other things :)10:57
git-branch does too10:58
eMish ok10:58
Randal git-branch -v shows even more cool stuff10:58
eMish i removed accidentally one or working files. how fo i get it from the git10:58
Randal was it part of the current commit?10:58
as in part of HEAD?10:59
git-reset --hard makes all tracked files be the same as HEAD10:59
eMish what is difference between HEAD and the branch ? /me feels himself idiot10:59
Randal the "current branch" is the branch that will be updated when you make commits. HEAD is the head of that branch11:00
llnz hummm...11:00
eMish git HEAD is the refs, right ?11:00
Randal I can't parse that11:01
eMish i think refs is used a s singulat throughut git docs, but i might be mistaken11:01
*singular11:01
or maybe it's not11:02
Randal HEAD points at one of the branches11:02
that's your "current branch"11:02
that's what "git-branch" shows with teh asterisk11:02
eMish you don't use term refs ?11:02
Randal I do, but I didn't understand your use of it11:02
eMish ok Randal. thanks11:02
llnz so if not using cg for uncommiting, what is the command?11:03
Randal I'd do something like:11:04
git-checkout -d SAVE master11:04
Hmm11:05
No. that's not going to do it11:05
basically, park the current HEAD where you can get at it, then backup one step, then recommit with the right username11:05
if you see a cg tool that does that nicely, use it. :)11:05
ahh - it's the first example under 'git-reset'11:06
"undo a commit and redo"11:06
I knew I'd seen that somewher. :)11:06
eMish cool11:06
Randal that's for llnz11:07
eMish looks like i finally switched my personal files from svn to git11:07
i had a bug with order of pull, push, commit in my scrripts. now i fixed it and i learned to resolve conflict11:07
wow what alleviation11:07
even though i totally don't know git11:07
← dduncan left11:08
eMish it is really nice when you can use something with very little knowledge of it11:08
saves time11:08
Randal like Perl :)11:08
eMish well when i started with perl, i used to know very little of it11:09
:-)11:09
i was totally shell and awk utilities11:09
and C11:09
perl andf git are not things that i can swallow in one year or in one gulp11:09
but the read of 100 lee begins with a single step11:10
and master of kung0fu used to say11:10
that was 100011:10
*road11:10
Randal I heard that as the walk of 1000 miles11:10
eMish thinks that her never walked more than 20km in one walk11:11
eMish *he11:11
this porcelain-vs-plumbing division, does it exist ? or higher-level commands migrate into core git ?11:13
Randal many things first existing only in add-ons have now been migrated to the core11:13
so the core is now both porcelein and plumbing11:13
last year, you *needed* cg11:14
this year, I'm not using it at all.11:14
the last thing of cg I was using was "cg-status"11:14
now that I know git-branch -v" does that, I've stopped.11:14
eMish i basically made my own scripts on top of git, now i can do all i need w/o touching git by hands11:15
i think it's goot11:15
because i 'm using same script of mine that i used earlier when it was svn below them11:15
"without touching git by hands" is a bit of exaggeration, but not much11:16
Randal except that you aren't really "thinking git" if you're using it only the way you used svn11:16
eMish true11:16
Randal for example, the very cheap branching in git encourages branches for every new edit11:16
eMish i'm not sure it's my goal to "think git" at this stage of life11:16
my goal was to switch from svn to git11:17
quickly11:17
because git is basically much much faster11:17
Randal well - you aren't switched until you are actually using it as intended. :)11:17
eMish i switched11:18
it's my goal to use at small subset as possible11:18
Randal I'm not sure that's a sane goal11:18
eMish btw Randal how come you switched from perl to git ? How did you get in touch with git ?11:18
Randal Uh - I didn't switch from *perl* to *git*!11:18
c'mon!11:18
eMish ok11:18
how did you come in touch with git ?11:19
Randal That'd be like saying "do you prefer chocolate or kentucky?"11:19
eMish which versioning system did you use before ?11:19
Randal that's a false dichotomy11:19
eMish ok ok11:19
Randal you mean *still*?11:19
I have used pretty much every system over the years,11:19
except for the new ones like hg and bzr11:20
I *must* use CVS and SVN for some projects11:20
eMish used for .. like for the purpose of trying ?11:20
Randal but I prefer git now for new projects where I choose.11:20
eMish - I'm a consultant... I use what the client says to use.11:20
eMish ah11:20
i see11:20
how much are you per hour11:20
Randal I don't bill by the hour11:21
eMish ok ok11:21
Randal I bill by the project.11:21
eMish how much time did it take you to learn git ?11:21
Randal I've been learning it slowly as it was being developed11:21
so I can't really tell.11:21
it's like asking me how I learned Perl 5. :)11:21
slowly, as it was being developed.11:22
I give a two-hour short-course on git though.11:22
Now twice.... once in Portland (my home city) and once last week in brazil.11:22
eMish you are sort of genius, right ?11:22
Randal I'm still in brazil, in fact.11:22
eMish - that's for others to say11:23
I'm just a guy. :)11:23
GyrosGei1r → GyrosGeier11:24
eMish thinks of something to ask Randal11:25
→ ferdy joined11:30
→ devogon joined11:43
→ lcapitulino joined11:44
etnt is it worthwile upgrading from 1.4.4 to 1.5.1 ? (Debian 4.0 has 1.5 in unstable)11:45
git-config doesn't seem to exist in 1.4.4 for example... ?11:46
llnz thank Randal11:48
← llnz left11:48
→ jonesy joined11:52
jwb_gone → jwb11:53
jonesy what's the difference between "CONFLICT (content)" and "CONFLICT (add/add)" ?12:07
Randal I can guess from the titles12:11
but I wouldn't know for sure. :)12:11
content conflict sounds like the same file has been edited two different ways12:11
jonesy that's what I'm figuring too.12:12
Randal and add-add probably means that in both previous versions, the file wasn't there, and now in both new versions it is.12:12
jonesy but the markers in the 'add/add' files just look like 'content' conflicts to me, so I thought I might do better than to guess. :-)12:12
Randal I suspect it's whether or not the file was present in the common ancestor12:12
jonesy that would appear to not be the case. The file is present in both, but it's different.12:13
Note that the question is partially academic12:13
Randal what about in the common ancestor?12:13
jonesy I saw 'content' in the docs, but 'add/add' in reality :-)12:13
lemme see.12:13
→ yorgen15 joined12:15
jonesy lemme do this in gitk :-/ this is harder on the command line.12:16
btw - is there a way to tell what a branch was created from? ie, what it's start point is?12:17
Randal heh12:17
git-find-common-ancestor or something12:17
jonesy thanks.12:17
Randal git-merge-base12:18
gitte jonesy: the conflict markers in add/add look like they do, because we try harder than RCS merge.12:20
jonesy gitte: I'm not complaining about the conflict markers - I'd just like to know what that means, exactly, as opposed to just seeing 'content' in the CONFLICT line in 'git merge' output.12:22
Randal - thanks for the tip. I'm new here and hadn't come across git-merge-base in the docs yet. :-)12:23
gitte As both of you suspected, "content" means that the path existed in the common ancestor.12:24
jonesy I *did* find 'git reset --hard HEAD' last night though. I did 'git merge <commit ID>' followed closely by 'git reset --hard HEAD' about a thousand times last night trying to figure out which exact version of Moodle I had made my local changes to.12:24
mountie etnt: Absolutely... Git 1.5.1 is night-and-day different from git 1.4.412:25
gitte jonesy: Linus posted a script to determine the most similar commit, given a current state.12:25
jonesy :-o12:25
gitte goes looking12:25
etnt mountie: thanx!12:26
jonesy hey - I had another question that I couldn't find an answer to. Since branches are cheap and easy in git, is there the equivalent of 'pop' and 'push' to move between them?12:26
;-)12:26
aeruder_ jonesy: not that i know of, but you could probably write yourself a tiny script to do so :)12:28
jonesy sure. I'm slow to write any scripts this early in my githood. I'd probably duplicate an effort.12:28
gitte http://news.gmane.org/find-root.php?message_id=%3cPine.LNX.4.64.0704171800290.5473%40woody.linux%2dfoundation.org%3e12:30
That contains a script finding a commit, given a specific file (with name).12:31
jonesy ok - thanks. I'll muck with that :-D12:31
gitte To switch between branches, use `git checkout <branchname>`.12:31
That is fast and easy.12:31
jonesy well, yeah, I knew that one!12:31
gitte It's no stack, though.12:31
Randal I'm not sure I'd want a "stack" of branches12:36
changing a branch isn't exactly cheap12:36
if you need to view many branches at once, make a git-clone so you can have them all checked out at the same time12:37
→ nud joined12:42
gitte Or use `git show <branch>:<dir>/` and `git show <branch>:<dir>/<file>`12:46
← eMish left12:53
→ nud_ joined13:00
→ gitte joined13:02
→ nud_ joined13:06
→ spuk- joined13:34
→ segher__ joined13:44
→ ferdy joined13:45
→ orospakr joined13:50
→ dirker joined14:03
→ dirker joined14:03
→ praka joined15:29
→ segher_ joined15:40
→ cods joined15:42
→ robinr joined15:48
→ spuk- joined15:52
→ MadCoder joined15:52
→ tonyj joined15:52
→ tokkee joined15:52
→ meyering_ joined15:52
→ davi joined15:52
meyering_ → meyering15:52
→ albertito joined15:52
→ robinr joined15:53
→ rphillips joined16:25
→ Oejet joined16:35
→ tomprince joined17:12
→ aroben joined17:31
→ aroben_ joined17:33
aroben_ → aroben17:49
→ GyrosGeier joined17:58
→ Eludias joined18:05
→ z3ro joined18:36
→ etnt joined18:38
→ aroben_ joined18:53
→ zeisberg joined18:56
→ devogon joined18:56
→ aroben_ joined18:57
lcapitulino gitster: hi, who's Hermes Trismegisto? :)19:22
gitster leaked his pseudonym by accident.19:28
lcapitulino hehe19:29
it seems that there isn't a wikipedia article in english about him19:29
but I found one in portuguese :)19:29
gitster If you reply to it the way you usually do, your message would begin "Hermes Trismegisto escreveu".19:30
tango_ it's just hermes the god as identified with a couple of other gods19:30
gitster Google for "tabua de esmeralda" ;-)19:30
lcapitulino haha, yes19:30
cool19:31
gitster: hey, where are you from? :)19:31
gitster Japan; I do not speak nor read Portuguese.19:32
lcapitulino okay19:32
ribas gitster: oguenki desuká? :)19:32
lcapitulino: i do read portuguese ;)19:32
lcapitulino ribas: nice :)19:33
→ winter-m1te joined19:35
→ sgrimm joined19:37
ribas I was thinking to implement some ACL for git. mainly for SSH [to allow push`s and pull`s].19:39
I was thinking to extend git-shell.19:39
→ mspang_ joined19:40
ribas gitster: you that is the master.. what do you think about extending git-shell to ACL?19:42
robinr ribas: you mean setting acl's on remote branches via git-shell?19:46
lcapitulino hmm19:48
git has a lot of linked lists19:48
what about using kernel's list.h ? :)19:48
zeisberg lcapitulino: which kernel?19:49
lcapitulino zeisberg: sorry, linux19:50
zeisberg lcapitulino: I wanted to point out that git run on Solaris, too. And there doesn't exist such a list.h19:51
lcapitulino zeisberg: yeah, I'm stupid sometimes calling everything 'linux'19:52
robinr and kernel headers are a no-no in user space anyway19:53
lcapitulino you can port the linked list stuff, no problem19:53
robinr true19:54
→ aroben joined20:00
→ zeisberg joined20:02
→ tomprince joined20:05
ribas robinr: yep...i could set a conffile that says who can/cannot work at some repo and when git-shell is executed it will verify if that user can get/upload packs20:06
← tango_ left20:07
ribas robinr: what I was thinking too.. is to the no need of having a user on a machine to work via ssh20:09
robinr so you want to roll your own authentication?20:11
hmm20:11
→ GyrosGeier joined20:19
ribas robinr: sort of... because sometimes i dont have root access to the machine, and i want my friends to get some work from repo... but they cannot get from all repos inside20:23
because sometimes some work we get different groups ;)20:24
or just to avoid some people commiting to a repo that they are not allowed to20:24
→ raalkml joined20:24
→ bronson joined20:27
→ alley_cat joined20:29
→ tomprince joined20:35
→ sept joined20:46
→ dkagedal joined20:56
→ Eludias joined20:58
→ dduncan joined21:03
→ Dodji joined21:06
→ tomprince joined21:16
→ aroben_ joined21:17
aroben_ → aroben21:22
jwb → jwb_gone21:23
← sept left21:26
→ jadhog joined21:37
← jadhog left21:38
→ aroben_ joined21:45
→ aroben joined22:09
→ aroben_ joined22:11
aroben_ → aroben22:13
← Simon- left22:15
raalkml gitster: "Fix handle leak in write_tree" - the patch was resent. Dunno how, but there was a close(newfd) almost directly after hold_locked_index!22:29
gitster: Sorry. I hope the patch didn't got too far22:30
→ jonesy joined22:33
→ jasam joined22:45
→ cworth joined22:48
→ fultilt joined23:14
→ chris___1 joined23:25
→ aroben joined23:29
chris___1 Q: does git-svn support dcommiting the addition of symlinks?23:30
→ aroben joined23:30
cworth Can anyone make sense of this error message for me?23:31
$ git-am < /tmp/0001-PDF-Add-support-for-transparent-gradients.txt23:31
fatal: cannot convert from utf-8 to utf-823:31
Hmm, deleting the line "Content-Type: text/plain; charset=UTF-8" actually seems to help somewhat...23:33
→ mithro joined23:34

Logs Search ←Prev date Next date→ Channels Documentation