IRCloggy #git 2006-09-21

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.

2006-09-21

gitster Yup. Looks much more obvious.00:00
pasky shall I resend the patch? the tweak is trivial00:00
gitster No, I did not realize you added the missing link to root. Sorry for the noise.00:00
pasky I've almost set up full-fledged git project hosting using DAV at repo.or.cz today, and did another really cool thing I wanna keep as a surprise, but it'll have to wait until tomorrow or so00:01
gitster Something I can pull from repo.or.cz would be nicer (I am feeling lazy ;-).00:01
pasky but with current gitweb feature set there, I'm almost pondering mirroring Linus' tree at repo.or.cz as well ;)00:01
gitster I am fine with the new "link to root" link. Some people with weaker eyesight and/or not-so-good hand-eye coordination might find a single slash hard to click, though.00:02
pasky gitster: I'll just send you some funny produced-by-not-properly-set-up-stgit patch from repo.or.cz... the repo has some local modifications and such mixed in and I don't have time to set up anything proper now00:02
gitster "[root]/ src/ util/ tree.c" might be a way to alleviate the problem.00:03
pasky Current directory: [glibc] / libc00:03
is how cvsweb shows it00:03
gitster nice.00:03
pasky yes00:03
it should've occured to me to just look at how others do it :)00:03
gitster so "[hed.git]/ util/ tree.c"?00:03
pasky I think " / " looks nicer than "/ "00:05
sent another iteration of the patch00:05
gitster I am easy either way (that's pretty much bikeshed); you decide.00:06
pasky the patch I sent has " / "00:06
gitster that's good.00:06
will take a look.00:07
pasky I should setup my stgit on repo.or.cz properly once it's not another night where I have 6 (night) hours before an exam and I didn't even look at it yet00:07
→ GeertB joined00:18
→ DrNick joined01:13
→ benlau joined01:52
→ Gitzilla joined03:14
→ yashi joined03:47
→ yashi joined04:08
→ yashi joined04:10
mfischer anyone tried git diff --color somefile | less04:47
and watched the ugliness?04:48
mugwump try less -R ;)05:12
→ devogon joined05:29
→ spuk- joined05:55
→ robinr joined06:54
kampasky or cg diff -c somefile ;)07:11
→ polyonymous joined09:18
→ benlau joined10:03
dwmw2_gone → dwmw210:18
→ GeertB joined10:40
→ thi joined10:49
→ chris2 joined11:09
→ cyrillos joined11:28
← cyrillos left11:34
→ GyrosGeier joined11:40
fonseca or tig diff somefile ;))11:44
→ cyrillos joined11:45
← cyrillos left11:58
→ GeertB joined12:08
→ dwmw2 joined12:28
→ timlarson_ joined12:38
→ cyrillos joined13:02
→ GeertB joined13:26
← cyrillos left13:57
→ cyrillos joined13:57
← cyrillos left13:57
→ cyrillos joined13:57
← cyrillos left13:57
→ cyrill joined13:57
cyrill → cyrillos13:58
→ cyrillos joined13:59
→ aggieben joined14:15
aggieben what is the '100' that gets inserted before the mode part of the header of a tree object?14:16
GyrosGeier aggieben, it's a full mode spec14:16
aggieben, that includes file type (reg, dir, link, ...)14:17
aggieben, and suid bits14:17
aggieben, although they are never used in git AFAIK14:17
aggieben GyrosGeier: ok, setuid bits are set in the 4th most significant byte. what's in the other two? dir and link?14:18
GyrosGeier aggieben, these are not bytes14:25
aggieben, these are octal digits14:25
→ robinr joined14:26
GyrosGeier aggieben, I think the first two digits are file type, because there can be more than 8 types14:26
aggieben yes, yes, not bytes. Digits. Anyway, ok. Thanks.14:30
→ coywolf joined14:35
→ agile joined14:53
→ mfischer_ joined14:57
dwmw2 → dwmw2_gone15:14
→ spearce joined15:25
→ apw joined15:35
→ benlau joined15:58
→ coywolf joined16:00
→ agorf_ joined16:46
agorf_ hello. is there svn-git comparison? im happy with svn and was wondering if there are any advantages for switching/using git16:46
there a*16:46
spearce I've sort of written one in an email to someone. I can go dig it up if you are interested.16:47
agorf_ i'd appreciate it :)16:47
spearce Acutally I should do that no matter what and put it on the GitWiki.16:47
gimme 15 minutes.16:47
agorf_ sure, thanks16:47
spearce So I've posted the bulk of the message here: http://git.or.cz/gitwiki/GitSvnComparsion16:53
I need to clean it up however. There's additional detail that I should go ahead and spend time on.16:54
agorf_ cool; thanks :)16:54
→ anholt_ joined17:05
robinr spearce: I don't understand one thing about SVN in the page. I made a branch i SVN and browsing the history showed the whole history a file in the branch17:14
spearce did that history include the history before the branch?17:14
robinr yes17:14
spearce hmm; last time I used svn (ok, over a year ago) the tools I was using didn't show that unless you added some obscure option to the command.17:15
robinr just svn copy17:15
using the BDB backend I think17:16
not sure that's relevent though17:16
spearce i don't think the backend would matter. i've made a note about your comment in the page so i don't forget about it. i want the page to be accurate not my fuzzy memory. :-)17:16
robinr not that SVN is in anyway better than GIT ofcourse17:16
I've seen other make the claim too17:18
spearce what claim, that svn is better than git?17:18
robinr the copy without history.17:25
spearce I think it was an old bug and people who remember that bug have it burned into the mind. :)17:25
robinr btw, I just checked it. the history before copy apears in the copy too17:25
spearce likely than that the SVN folks fixed the bug between when I last used SVN and now. Not surprising as I know they are still actively developing the tool. :)17:26
robinr still that's strange because that feature was claimed long before svn became 1.017:27
spearce iirc somewhere around the 1.0 timeframe you needed an option to view the history before the copy. having the history available was a feature from day 1, it just wasn't immediately available in the UI level of the tool.17:28
robinr creating a repo in SVN is simple too. svnadmin create somepath17:28
spearce i think people found the option awkward and the default behavior changed.17:29
robinr ok17:29
spearce robinr: I know about svnadmin create and agree its easy, but you then need to do a checkout to a working directory. in git its the same. :) 17:29
(sorry, I meant to say "its the same directory")17:29
robinr agreed, that's one step less17:30
spearce but its a minor detail.17:30
agorf_ yeah i hate the additional checkout17:30
agorf_ → agorf17:30
robinr the BIG thing with git is the marvelous support for branches17:30
the hard part, i.e. merging17:31
spearce agreed.17:31
robinr and speed17:31
spearce if you read the outline on http://git.or.cz/gitwiki/GitSvnComparsion i've got those in the top 4 git features. :)17:32
robinr a bad thing is, assumes everybody uses the same charset17:33
spearce yes.17:33
robinr only subversion and (maybe cvnt nowadays) handle that.17:33
spearce that's ok, there's only US-ASCII (says the dumb American)17:34
agorf it doesn't support utf8?17:34
robinr no, eight bit works17:34
spearce Right, most Git users work in UTF-8 for the files they store.17:34
robinr but most people here use ISO-8859-1/1517:34
spearce Commit messages and filenames should be in UTF-8.17:35
robinr windows here (cygwin), use Windows-1252 (close to 8859-1)17:35
→ alley_cat joined17:38
→ monkeydiver joined17:52
← thi left17:53
→ spearce joined18:03
→ krh joined18:21
→ ynezz joined18:29
← ynezz left18:53
→ GeertB joined19:14
→ spearce joined19:32
→ lyakh joined19:45
→ xjjk joined19:58
→ two-face joined19:59
aggieben does anyone know of a way to get pmake to do target-specific variable assignments? From what I can tell, it can't do them directly.20:03
spearce no clue20:04
two-face Hi20:04
Is there anyway to clone a given branch only of a given repository?20:05
spearce you can clone it without doing a checkout, then checkout that specific branch. but by default clone takes all branches into the new repository and there's no way to stop that.20:06
two-face ah ok20:06
thanks20:06
kampasky two-face: add '#branchname' to the url20:13
two-face huh? Reallly?20:13
kampasky (when giving it to clone)20:13
two-face is it documented somewhere?20:14
kampasky hmm20:15
actually, I'm not sure it works with plain git20:15
give it a try20:15
spearce kampasky: I'm looking at the git-clone.sh code and unless its buried down in the low level fetch code I'm not seeing how it could work. 20:16
two-face git clone git://git.kernel.org/pub/scm/git/git.git#next20:16
fatal: unexpected EOF20:16
fetch-pack from 'git://git.kernel.org/pub/scm/git/git.git#next' failed.20:16
spearce must be a cogito extension...20:17
two-face kampasky: I suspect there's an ongoing implementation from you :-)20:17
kampasky ok, so git can't do it20:17
hmm, that's a bit unfortunate20:17
well... :)20:18
two-face porting to git?20:18
spearce at 419 lines of ugly shell code, git-clone could use some help.20:18
kampasky two-face: cogito can do it20:18
two-face: otoh cogito can't clone more than one branch at once currently20:18
two-face kampasky: ah20:19
spearce wonders if Cogito userbase is getting smaller.20:26
two-face why using cogito?20:27
robinr wonders too20:27
spearce I used to be a Cogito user. I stopped being one in Jan.20:27
two-face git is now a scm :)20:28
porcelain commands that is20:28
robinr stgit is cool though20:29
spearce True, but there are a number of cool things in Cogito that aren't in Git. And yes, StGit is also quite cool for change development.20:29
two-face not pg? :)20:30
spearce I've got a Perl script named `tb` sitting around here that I've been trying to work on to help manage things like `pu` but commit back to the topic branch.20:30
robinr haven't tried pg20:30
spearce I'm probably going to stop developing pg. In favor of core Git and `tb`.20:30
Actually tb is probably ready aside from it saying "cowardly refusing to perform a 3 way commit" if more than one topic branch has modified the file you are trying to commit to from your temporary branch.20:31
I actually haven't used pg myself in about 3-4 months now. :)20:31
two-face stgit is said not to be well integrated with git20:32
spearce it is and it isn't; there's things that stgit does that play alongside of git but that git doesn't understand or handle gracefully.20:32
two-face too bad20:32
robinr there are some things you shouldn't do with git commands when using stgit20:33
e.g. git pull is bad20:33
spearce tb is my attempt to replace both pg and stgit with something that plays better with git, has cleaner history of a "patch" and uses more native git concepts for that history.20:33
two-face good20:33
GeertB What files have the canonical definitions of various git file formats (tree, blob, commit)?20:33
spearce but that 3-way commit problem has me stuck at the moment, which is why i haven't announced it.20:34
robinr 3-way commit?20:34
spearce GeertB: Did you read Documentation/core-tutorial.txt ? 20:34
robinr spearce: time to ask for help then :)20:34
spearce robinr: If you attempt to commit changes to a topic branch from a file which has been altered by a different branch that was merged on top of the branch you are about to commit to things are a little "weird". 20:35
GeertB spearce: yes, but it indeed really is more of a tutorial with some details thrown in.20:37
spearce: maybe I should gather the info from various places and add some docs in technical/ ?20:37
spearce Yes. :-) That's AFAIK the only documentation on it. Everything else is folklore and source code that ain't so clear.20:38
GeertB OK, before trying to write something up, I wanted to be sure I didn't miss something obvious...20:38
spearce Although ``A Large Angry SCM`` tried to write something up and posted it to the mailing list a while back (around July?) but it never got accepted into the tree.20:39
His document was the most detailed anyone had written thus far and also covered the pack file format.20:39
robinr spearce: export the change to a file, drop it and apply at the head of your topic branch (if I understand correctly what you want to do...)20:40
spearce robinr: Yes, its just code. ;-) I just haven't written it yet. Where it gets ugly is when it doesn't apply cleanly. Then you probably need to switch the user to the topic branch by itself and have them fix up the change. 20:40
Gitzilla GeertB: Check www.git.rsbx.net in a few minutes and the current state of my Git file formats document will be there.21:06
→ ShadeHawk joined21:07
ShadeHawk Is there any way for lost form of magic open "-|" to redirect standard output to /dev/null21:08
open my $fd, "-|", "git", "cat-file", '-t', $hash or return;21:08
spearce uh, stdout is tied to $fd; stderr is tied to your stderr. and no way to redirect stderr to /dev/null, which is i think what you mean...21:08
segher put the whole command line in just one string and you can use shell redirection21:09
GeertB Gitzilla: Great, thanks in advance!21:10
segher so, open my $fd, "| git cat-file -t $hash 2>/dev/null" or ....21:10
ShadeHawk rather21:11
open my $fd, "git cat-file -t $hash 2>/dev/null |" or return21:12
spearce of course that's only safe if $hash is safe. :)21:12
and your forking twice.21:12
ShadeHawk but then lose advantages oflist form ($hash safety, and shell forking)21:12
Gitzilla GeertB: It should be there now. Keep in mind that it's a work in progress and I'm _still_ finding inaccuracies.21:12
spearce Gitzilla: TAG_CONTENTS is missing 'tagger'. 21:14
[spearce@d13-70 git]$ git cat-file tag v1.4.2.1 | grep tagger21:14
tagger Junio C Hamano [email@hidden.address] 1158138501 -070021:14
but otherwise this document is quite complete and very clear.21:15
Gitzilla spearce: Thanks! I'll look into it later.21:16
spearce also Nico just added a new method of storing deltas in packs. Junio hasn't merged it yet but its probably coming in soon.21:16
Gitzilla spearce: I saw. I haven't looked at it in detail yet. I'm waiting for it to cook in next for a bit.21:18
spearce probably wise. just wanted to make sure you were aware of it. :)21:18
GeertB Gitzilla: that's really great info! Could you submit this to be included with git?21:19
→ ShadeHawk joined21:31
→ spearce joined21:31
→ GeertB joined21:31
→ krh joined21:31
→ alley_cat joined21:31
→ anholt_ joined21:31
→ agorf joined21:31
→ apw joined21:31
→ mfischer_ joined21:31
→ robinr joined21:31
→ aggieben joined21:31
→ dwmw2_gone joined21:31
→ GyrosGeier joined21:31
→ polyonymous joined21:31
→ spuk- joined21:31
→ devogon joined21:31
→ Gitzilla joined21:31
→ segher joined21:31
→ robfitz joined21:31
→ Tv joined21:31
→ flz joined21:31
→ moh joined21:31
→ CIA-21 joined21:31
→ dmlb2000 joined21:31
→ anders_ joined21:31
→ matled joined21:31
→ pasky joined21:31
→ shy joined21:31
→ beu joined21:31
→ njs joined21:31
→ auke joined21:31
→ MadCoder joined21:31
→ makx joined21:31
→ lu_zero joined21:31
→ neuralis joined21:31
→ ruskie joined21:31
→ cworth joined21:31
→ Alex joined21:31
→ emrys joined21:31
→ normalperson joined21:31
→ kampasky joined21:31
→ Newsome joined21:31
→ bartman joined21:31
→ Thumper_ joined21:31
→ philips joined21:31
→ vlajos joined21:31
→ afrayedknot joined21:31
→ PugMajere joined21:31
→ BearPerson joined21:31
→ kreaturr joined21:31
→ tonyj joined21:31
→ kblin joined21:31
→ nickh joined21:31
→ fonseca joined21:31
pasky what about just maintaining it in the wiki"?21:34
Gitzilla I find that Wikis to be quite painful for very structured documents.21:37
→ mfischer_ joined22:14
→ robfitz joined22:14
→ Beber` joined22:34
pasky gitster: resent the patch, turns like I didn't stg refresh before sending it tomorrow23:27
and now I'll finally setup the stgit there :)23:27
agorf mm question. in git, you checkout all of the repository? not just a directory?23:28
pasky yes23:33
ShadeHawk yes, by default you checkout whole directory tree. You can even now do partial checkout, but it is not supported by high-level UI (meaning you have to use low-level commands for that)23:33
agorf okie thanks23:34
→ Gitzilla_ joined23:38
→ DrNick joined23:39
Gitzilla_ /msg nickserv link Gitzilla rsbxgitirc23:41
spearce damn spacebar :)23:41
agorf lol23:47
Gitzilla_ /msg nickserv link Gitzilla rsbxgitirc23:47
agorf dude, drop the whitespace23:48
→ segher joined23:50
Gitzilla_ Sorry, bad CnP23:50
→ Gitzilla joined23:53
→ Gitzilla joined23:56

Logs Search ←Prev date Next date→ Channels Documentation