IRCloggy #git 2007-08-11

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.

2007-08-11

bemo docbook-style-xsl-1.68.1-1.100:01
gitster Ah, Ok, thanks for another datapoint.00:01
bemo any time... glad to help (it's about the only level I can help at, so far) ;-)00:02
← meandtheshell left00:08
← z3ro left00:12
→ DougBTX joined00:19
→ z3ro joined00:19
← dorileo left00:24
← jasam left00:26
→ diacritical joined00:32
← davi left00:49
← ShadeHawk left00:50
← ignotus left00:52
← diacritical left00:56
← KirinDave left01:00
→ evilchelu_ joined01:03
← tongueroo left01:06
← Tv left01:09
← DougBTX left01:12
→ branstrom_ joined01:17
← Yuuhi left01:17
← branstrom left01:18
← evilchelu left01:18
← eventualbuddha left01:19
← GyrosGeier left01:23
→ branstrom joined01:24
← branstrom_ left01:25
← janm left01:26
→ tongueroo joined01:26
→ zed_ joined01:29
← aroben left01:29
← tongueroo left01:30
→ GyrosGeier joined01:30
← glezos left01:33
zed_ I'm noob01:33
what is it?01:33
git?01:33
:|01:34
Insount → Insount_away01:34
← kanru left01:37
← zed_ left01:39
← tchan left01:40
dwmw2_gone → dwmw2_PVG01:50
→ kanru joined01:52
← evilchelu_ left02:02
→ janm joined02:02
→ tchan joined02:06
→ stouset joined02:08
→ aroben joined02:09
dwmw2_PVG → dwmw2_gone02:11
dwmw2_gone → dwmw2_PVG02:20
→ KirinDave joined02:21
← KirinDave left02:21
→ KirinDave joined02:22
← KirinDave left02:23
→ orospakr joined02:35
← Insount_away left02:48
→ Insount joined02:48
← vmiklos left02:56
← Sho_ left03:03
→ nzkoz joined03:17
→ diacritic joined03:20
diacritic I'm having a tough time specifying the blob name to git-cat-file03:21
whee If I get a patch mailed to me and I want to take the patch plus make some modifications of my own, can I do that (and will it/should it retain the author as the original author (with me adding a signed-off-by for me))03:21
diacritic is it like: git-cat-file -t :objects/4b/82dc6... ?03:22
I want to list the file in ./.git/objects/4b/82dc6...03:22
diacritic → diacritical03:24
← aroben left03:25
← thiago left03:25
diacritical whee i guess you should specify GIT_AUTHOR_NAME and GIT_AUTHOR_EMAIL environment when committing03:25
whee looks like I could let git-am commit as-is, git-reset --soft HEAD^, change it up, and then commit with --author03:26
diacritical retaining GIT_COMMITER_{NAME,ENVIRONMENT} as yourself03:26
yeah03:26
whee but that seems like a lot of work03:26
nzkoz Hey guys, I'm using git-svn to do some reasonably drastic changes to a subversion project. I'd like to publish my local branch for others to investigate too (using git, not svn branches)03:26
any gotchas I need to be aware of?03:26
diacritical doesnt touch svn with 10ft pole03:27
whee heh :)03:27
nzkoz: what do you mean by reasonably drastic03:27
nzkoz whee: I'm moving crap around, changing functionality03:27
if I wanted to suffer, I'd be doing all this in an svn branch03:27
whee you shouldn't have any problems if they are just going to clone your git-svn repository and look at the changes03:28
they will have to do some things if they want to use git-svn though03:28
nzkoz but I don't, so I've been working locally with a whole bunch of git branches, and I just want to push those out for review03:28
yeah, I only intend them to clone the git repo,03:28
I'll do the svn dcommit etc03:28
whee it should be fine then03:29
nzkoz the git repository doesn't store my svn creds somewhere?03:29
→ thiago joined03:29
whee I don't believe it does, I grepped around before I let people clone mine at work :)03:29
→ aroben joined03:29
nzkoz heh, sounds good enough for me03:29
→ fhobia joined03:34
context nzkoz: it will use creds in ~/.svn i believe03:35
← bronson left03:36
diacritical any one point me to an example of using git-cat-file ?03:40
I cant figure out the syntax from the manpage03:40
nzkoz is still amazed how fast and small the git storage / operations are03:41
nzkoz honestly03:41
wow03:41
;)03:41
context diacritical: man git-cat-file03:42
nzkoz: ;)03:42
nzkoz: you using git-svn ?03:42
diacritical that wasnt helpful context03:42
I just said i couldnt figure out the syntax to specify a blob object after reading that man page a few times03:43
context nzkoz: it amazed me at first that half the time git-svn (with ENTIRE history locally of course) is smaller that an svn wc03:43
nzkoz context: yeah, same here. those .svn folders sure do add up...03:43
context nzkoz: ;)03:44
gitster diacritical: still here?03:49
or whee, who wanted to apply a patch with fix-up?03:49
diacritical yep still looking :) the object is in .git/objects/4u/etc. cant i get look at it with -t ?03:49
gitster First of all, just out of curiosity, what made you interested in ".git/objects/4b/82dc6..."?03:50
diacritical I'm investigating a "missing blob" from git-fsck on ntfs03:51
whee gitster: I did03:51
diacritical trying to replay the actions03:51
whee :D03:51
gitster I am curious because most of our tools do not say 4b/82dc6..., but 4b82dc6...03:51
Ah, I see, fsck output.03:51
thiago diacritical: what are you trying to do?03:51
gitster If it is "missing" then by definition it does not exist so cat-file would not help you .03:51
diacritical no no03:51
i have the object via find03:51
gitster If it is listed as "dangling", then you have it.03:51
diacritical i want to see what object it is03:52
this is unrelated to the fsck output03:52
i have made a mess of explaining what i'm trying to do03:52
gitster Ok, anyway, ".git/objects/4b/82dc6..." is an implementation detail of one representation of 4b82dc6... object.03:52
diacritical ah!03:52
ok03:52
now i get it03:52
gitster So, remove .git/objects, and slash after the 4b, then "git cat-file -t 4b82dc6"03:53
diacritical thanks03:53
gitster You can omit the trailing digits as long as it is still unique.03:53
Now, whee's turn. Have you managed to apply with a fix? If you are on the mailing list and/or follow git.git repository, you probably have noticed that I do that all the time.03:54
There are two primary ways I do it, depending on the scale of changes.03:54
(1) for very small change, I just edit the patch before applying. You need to know what you are doing, but this is quicker than the second option.03:55
(2) otherwise, I just "git am" the original, tweak in the working tree and "git commit --amend" (if I started from a clean working tree, I would say "-a", otherwise I'd update the index to include only the necessary changes).03:56
whee hmm, I like the --amend way03:57
diacritical [I've a followup on the colon syntax]03:58
gitster Its much less taxing on the brain, I would agree.03:58
whee awesome, this is nicer :P. thanks03:58
diacritical [git-cat-file -t tells me it is a tree. can I use the colon syntax to list it? i.e. git cat-file tree <SUFFIX:PATH-SYNTAX> --- probably works better with `blob' instead of `tree']03:58
gitster "git cat-file -t $tree_or_commit:$path"03:59
What's <SUFFIX:PATH-SYNTAX>???03:59
← janm left03:59
gitster Examining objects human-readable way is probably easiest with "git show" these days. E.g. "git show 4b82dc6"03:59
diacritical thats the part i'm trying to fill in :) I keep getting "Not a valid object name"03:59
thiago git-ls-tree <id>04:00
gitster Yeah, "git show $tree" is a thin wrapper around "git ls-tree $tree".04:00
diacritical i assume a missing blob is an object that is marked as reachable but has nothing pointing to it04:00
I checked the ids of all objects in ls-tree and verified that the missing blob was not listed among those04:01
gitster That describes a dangling or unreachable blob.04:01
diacritical gitster, i understood that from looking at the code diagnostic where it prints out "missing"04:01
gitster missing is something that ought to be there (e.g. a tree points at it) but does not exist.04:01
diacritical in builting-fsck04:01
gitster Since you have the object in .git/objects/ it cannot be missing ;-)04:02
diacritical i may be wrong04:02
no gitster, i confused you04:02
these are two unrelated objects04:02
← tcoppi left04:02
diacritical one issue is the fsck output - which lists a missing blob04:03
but that missing blob is not listed in any of the ls-tree outputs, and i looked at where that message is being printed04:04
it seemed to me (i may be wrong) that it is printed when the reachable flag is set by mistake04:04
now i am trying to replay the actions and understand whats happening in the repository at each step04:05
gitster "not listed in any of the ls-tree"? you examined _all_ the reachable trees?04:05
diacritical i did log --all and looked at all the commits04:05
→ mgrimes joined04:05
gitster There are a few other places reachability come other than the tips of branches and tags (the starting points of "log --all").04:06
One of them is the index.04:06
diacritical i have to learn to read the index now :)04:07
gitster for example, I just did (don't try this without reading to the end):04:08
diacritical tries to get cat-file under belt first04:08
nzkoz I've run git-instaweb and now have: http://git.koziarski.com:1234/04:08
gitster $ echo some random garbage >foooo04:08
$ git add foooo04:08
→ tcoppi joined04:08
gitster $ git rev-parse :foooo04:08
nzkoz I was expecting to see entries for my stuff in there? but I'm obviously missing some silly step, any docs on what git-instaweb does and where to go from there?04:09
gitster $ rm -f .git/objects/xx/xxxxx.. (the answer to the rev-parse)04:10
$ git fsck04:10
Then that object becomes missing.04:10
diacritical ok04:11
← madewokherd left04:11
diacritical [FTR I didnt mess inside .git/ without git commands]04:12
→ aroben_ joined04:13
← aroben left04:14
aroben_ → aroben04:14
diacritical i think it can happen with `git mv' but it looks like an hour or two before i can figure it out04:14
after introducing some mismatch in the index04:14
diacritical waves hand04:14
dwmw2_PVG → dwmw2_gone04:14
diacritical git cat-file tree a10e5:$PATH <-- what can $PATH be heer?04:15
gitster whatever path that tree a10e5 contains. If it corresponds to the top level of git.git project, for example, then you would say Makefile, README, Documentation,...04:18
But Makefile and README are not trees, so "cat-file tree" would barf ;-)04:18
diacritical ok i get it04:18
gitster "git cat-file tree HEAD:Documentation" would show the tree.04:18
diacritical smacks forehead04:19
gitster If you do not have anything on the LHS of colon, it will pick the path from the index, but because the index holds only blobs and no trees, you cannot say "git cat-file -t :Documentation"04:19
diacritical on the other note: the fsck lists missing blob even when i check out another branch -- this means index is gone. so the reference to the missing object cannot be in the index, right?04:20
if i can find who has a reference to the `missing blob' i'll be happy04:21
gitster it could be in one of the commits recorded in the reflog.04:24
diacritical ah04:26
← rkaway1 left04:27
→ rkaway1 joined04:29
diacritical [grep doesnt find it.]04:30
dwmw2_gone → dwmw2_PVG04:30
diacritical how can i list the objects in a pack ?04:31
← whee left04:35
diacritical the link to technical/pack-format.txt is broken in the user-manual.html document04:38
jrockway diacritical: incidentally, that document doesn't answer your question04:38
jrockway was just looking there ;)04:38
jrockway perusing the code didn't find anything promising, but obviously i'm not looking hard enough04:42
diacritical i'll have to try again later.04:43
jrockway my advice is to read builtin-rev-list.c04:44
it knows about revs in pack files, and i assume it's not guessing :)04:44
gitster "list the objects in a pack?" if you have .pack file and only that then you are essentially doing what index-pack does, so that would be a good source of information. Documentation/technical/pack-format.txt is more or less correct, but it might not be quite up-to-date.04:46
If you want to list the pack-*.idx file, then git-show-index, perhaps?04:46
diacritical i'll check that when i go offline in a bit, thanks gitster04:48
oh one more question! if i push by mistake to a branch which is checked out, which reset incantation to invoke in the working directory (of the repo which was pushed into, and where the branch is checked out)04:50
i dont want to undo the push ...04:50
jrockway hard?04:50
gitster depends on what the working tree looked like. "git diff" (without any option) would show what you were in the middle of doing in the checked out working tree, so save that away first.04:51
diacritical save it, nuke it, and then checkout. ok!04:52
gitster "git diff HEAD" would be useless as that would show revert of what the pushed branch did, plus your local changes.04:52
So "git diff >P.diff" then "git reset --hard HEAD" and perhaps "git apply P.diff" and hope for the best.04:52
diacritical thanks04:52
← diacritical left04:53
← nzkoz left05:01
→ jeffpc joined05:31
← dduncan left05:38
→ Tv joined05:49
← Tv left05:51
→ Tv joined05:52
← Tv left06:09
→ Tv joined06:10
→ spuk joined06:16
← spuk- left06:18
→ polyonymous_ joined06:24
← polyonymous left06:24
← Tv left06:26
→ Tv joined06:27
→ jerbear joined06:31
jerbear is there a way to set git to use a different ssh key when connecting, either in .ssh/config or in git's config?06:32
DrNick IdentityFile directive in ~/.ssh/config06:34
→ bdiego joined06:35
jerbear DrNick: is there a way to have ssh use that identity file if i connect with a specific user?06:35
DrNick I read the man page last time. your turn.06:36
jerbear man ssh?06:36
DrNick ssh_config06:36
jerbear i still don't see how this can be done06:40
DrNick I don't know that it can. That's why you're reading the manual. :)06:41
you could probably just create a Host entry and supply a User and IdentityFile06:42
← Tv left06:43
jerbear DrNick: can a host entry be user@host?06:44
→ Tv joined06:44
← mgrimes left06:46
jrockway i think ssh will try all available keys until one works, so why bother configuring one key per host?06:58
← Tv left07:00
→ Tv joined07:02
→ mgrimes joined07:17
← Tv left07:18
← jrockway left07:18
→ Tv joined07:19
← DrNick left07:23
← Tv left07:35
context why would you use one key per host ?07:35
→ Tv joined07:36
otavio I'm able to use git fetch without a working tree07:38
but it do not update HEAD and refs/head/master07:39
how I can update them without a working tree?07:39
jerbear i have a password protected key that i use for normal ssh, but i want a passwordless key to use on my lan just for git07:39
← jerbear left07:39
← bdiego left07:40
← aroben left07:41
← jwb_gone left07:42
→ jrockway joined07:47
← Tv left07:52
→ Tv joined07:53
→ lyakh joined08:08
← Tv left08:09
→ Tv joined08:11
→ davi joined08:13
← mgrimes left08:19
→ ofri joined08:22
← Tv left08:27
→ Tv joined08:28
→ lyakh_ joined08:35
→ nikodemus joined08:37
← Tv left08:44
→ Tv joined08:45
← lyakh left08:46
← lyakh_ left08:50
→ lyakh joined08:50
→ nashdj joined08:55
→ d0k joined09:01
← Tv left09:01
→ Tv joined09:03
robin otavio: git fetch only updates remote refs09:04
by design09:04
kblin is there a way to track branches on gitweb via RSS/ATOM?09:05
the RSS link only gives me the master branch09:06
← davi left09:11
← jrockway left09:17
→ alley_cat joined09:18
← baggito left09:18
← Tv left09:19
→ Tv joined09:20
← lu_zero left09:21
→ davi joined09:27
← capisce left09:29
gitster http://git.pastebin.com/m2a472ebc09:34
← Tv left09:36
gitster In the meantime, you could add ";h=next" yourself, like:09:36
http://repo.or.cz/w/alt-git.git?a=rss;h=html09:36
→ evilchelu joined09:36
← evilchelu left09:37
→ evilchelu joined09:37
→ Tv joined09:37
→ janm joined09:37
→ ferdy joined09:40
→ meandtheshell joined09:46
→ puzzles joined09:47
← ofri left09:48
← nashdj left09:52
← Tv left09:53
← fhobia left09:53
→ puzzels joined09:54
→ Tv joined09:54
dwmw2_PVG → dwmw2_gone09:57
← puzzles left09:59
dwmw2_gone → dwmw2_one09:59
dwmw2_one → dwmw2_gone09:59
puzzels → puzzles09:59
→ doublec joined10:05
→ sgrimm joined10:07
← Tv left10:10
→ Tv joined10:12
← davi left10:13
← doublec left10:15
→ elmex joined10:16
→ DougBTX joined10:19
→ davi joined10:27
← Tv left10:27
→ Tv joined10:28
→ kumbayo joined10:35
← davi left10:35
→ davi joined10:35
← Tv left10:44
→ Tv joined10:45
→ lu_zero joined10:56
← Tv left11:01
→ Tv joined11:03
→ Sho_ joined11:06
dwmw2_gone → dwmw2_PVG11:10
corecode uhm11:11
how do i set up that my master will be updated from remote/origin/master?11:12
→ jrockway joined11:13
context use pull11:14
corecode: its set in the .git/config11:15
it should be done by default11:15
corecode well, i didn't cloe11:15
clone*11:15
i thought this fancy branch --track is supposed to do that11:15
context think you need to use git remote11:16
and maybe git branch11:16
→ nud joined11:16
corecode hm.11:17
branch.master.merge = refs/heads/master11:17
isn't that11:17
redundant?11:17
siprbaum you need branch.<name>.remote and branch.<name>.merge11:18
← Tv left11:18
siprbaum read the man page of git-config and search for merge, there is an example there11:19
corecode i know11:19
but why?11:19
→ Tv joined11:20
siprbaum why to read the manpage? it normally teaches you something ;-)11:20
corecode ah!11:20
merge is the remote name11:20
i got confused because the local ref name of course isn't refs/heads/master, but refs/remotes/origin/master11:21
siprbaum yepp.11:22
this is needed if you pull from many repos (like Linus does) without storing their heads in refs/remotes/*11:22
corecode oh?11:28
why wouldn't you do that11:28
robin the refs/remotes/... is to confuse users less, i.e. make remotes and local branches have different names11:31
it is relatively new11:31
→ evilchelu_ joined11:31
robin you can pull without setting pre-configuring11:33
you can pull without pre-configuring branches11:33
← nud left11:33
→ DougBTX_ joined11:35
← Tv left11:36
→ Tv joined11:37
dwmw2_PVG → dwmw2_gone11:37
→ goloo joined11:44
← evilchelu left11:46
corecode WOW11:48
git rebase11:48
heh11:48
that's nice11:48
← DougBTX left11:51
← fultilt left11:53
→ chris2 joined11:57
← Tv left11:57
→ whee joined11:57
→ Tv joined11:58
← DougBTX_ left12:02
← lyakh left12:03
→ jwb joined12:05
→ DougBTX joined12:06
jmspeex git bisect saved my ass once again!12:08
dwmw2_gone → dwmw2_PVG12:11
dwmw2_PVG → dwmw2_gone12:13
← Tv left12:14
→ Tv joined12:15
dwmw2_gone → dwmw2_PVG12:20
→ ofri joined12:22
← kanru left12:28
← Tv left12:31
→ Tv joined12:33
→ evilchelu joined12:39
→ devogon joined12:41
kblin gitster: thanks :)12:48
← Tv left12:49
→ Tv joined12:50
← evilchelu_ left12:51
→ kanru joined12:58
← kanru left13:02
→ Dodji joined13:04
← Tv left13:06
→ Tv joined13:07
→ kanru joined13:13
→ ponch_ joined13:22
← Tv left13:23
→ Tv joined13:24
ponch_ → ponch13:27
← sgrimm left13:27
→ pd joined13:30
→ evilchelu_ joined13:31
← evilchelu left13:33
evilchelu_ → evilchelu13:33
→ branstrom_ joined13:33
← branstrom left13:34
← ponch left13:36
→ branstrom joined13:38
← branstrom_ left13:40
← Tv left13:40
→ Tv joined13:42
← branstrom left13:43
→ branstrom joined13:44
→ branstrom_ joined13:49
→ bdiego joined13:49
→ capisce joined13:49
← branstrom left13:49
→ madewokherd joined13:50
→ branstrom joined13:54
← branstrom_ left13:54
← Tv left13:58
→ Tv joined13:59
→ branstrom_ joined13:59
← branstrom left13:59
→ stillLotR joined13:59
→ branstrom joined14:04
← branstrom_ left14:05
← LotR left14:05
→ branstrom_ joined14:09
← branstrom left14:10
← DougBTX left14:14
→ branstrom joined14:14
← Tv left14:15
← branstrom_ left14:15
→ Tv joined14:16
← branstrom left14:19
→ branstrom joined14:20
← branstrom left14:21
→ mithro joined14:25
mithro hi! is there a git-describe command which works on a remote repository?14:25
loops mithro, no, you'll need to fetch the branch you're interested in into your local repo for such operations. Depending on what you're looking for "git ls-remote", might give you a bit of info though.14:30
← MrPloddy29 left14:31
← Tv left14:32
→ MrPloddy29 joined14:32
→ Tv joined14:33
← MrPloddy29 left14:36
→ bentob0x joined14:37
→ MrPloddy29 joined14:37
chris2 → chris2_14:43
→ lyakh joined14:45
chris2_ → chris214:49
← Tv left14:49
→ Tv joined14:51
← polyonymous_ left15:00
← Tv left15:06
→ Tv joined15:08
→ blergit joined15:22
← blergit left15:22
← Tv left15:24
→ Tv joined15:26
bemo gitster: another datapoint... FC4 doesn't have docbook2x-texi (and their docbook2X rpm has db2x_docbook2texi and db2x_texixml as the only binaries with "texi" in them)15:33
→ evilchelu_ joined15:34
→ nud joined15:35
← Tv left15:44
→ Tv joined15:45
← evilchelu left15:46
← Tv left16:01
→ Tv joined16:02
dwmw2_PVG → dwmw2_gone16:06
Insount → Insount_away16:08
→ evilchelu joined16:10
← Tv left16:18
→ Tv joined16:19
→ jasam joined16:22
← evilchelu_ left16:23
← dwmw2_gone left16:26
mnemoc hi, how can i avoid to get "Merged from...." commits when pulling?16:26
→ dwmw2_gone joined16:26
jasam mnemoc, I don't understand exactly what you want, if you "fetch" instead "pull", no merge is done16:28
mnemoc jasam: ah, thanks! so i need fetch + rebase :D ... i got it :)16:29
jasam :)16:35
← Tv left16:35
→ Tv joined16:36
→ hFonteBoa joined16:39
→ rtmfd_ joined16:44
→ whee_ joined16:45
← whee left16:49
whee_ → whee16:49
thiago mnemoc: make sure you allow only fast-forward merges16:52
← Tv left16:52
→ Tv joined16:54
→ evilchelu_ joined16:57
← evilchelu left17:09
← Tv left17:10
→ Tv joined17:11
mnemoc thiago: uh, is that a config option?17:13
→ rjbs joined17:14
← alley_cat left17:16
← rtmfd_ left17:16
→ rtmfd_ joined17:17
thiago it's the src:dest refspec17:19
rjbs looks for a consultant to shave his git yak for him, so he can play Nintendo instead.17:20
→ geckosen1tor joined17:23
geckosen1tor I noticed git diff pipes through less or something.. is there a way to disable this?17:23
→ hackbert-hh joined17:23
← nud left17:25
bdiego geckosen1tor, it pipe to the PAGER environment variable (or pager in you .gitconfig)17:25
geckosen1tor what if I don't want a pager?17:26
← Tv left17:27
→ Tv joined17:28
loops geckosen1tor, export GIT_PAGER=cat17:30
geckosen1tor is the .gitconfig in my home directory? or at the root of each repository?17:35
loops there is one ~/.gitconfig and a .git/config in each repo17:35
you can use git config --global core.pager cat to turn off pager for all your repos17:35
geckosen1tor ok cool17:35
I tried the above and it didn't seem to do anything17:36
loops hmm.. what version of Git?17:36
geckosen1tor 1.5.217:37
loops geckosen1tor, doesn't work for 1.5.2 apparently17:38
geckosen1tor ok17:38
I need 1.5.3 or something?17:38
loops yeah.. which isn't quite out yet17:39
geckosen1tor can't I git it with git?17:39
loops oh yes..17:39
geckosen1tor I'll try that17:40
← Tv left17:44
→ alley_cat joined17:45
→ Tv joined17:45
geckosen1tor oh interesting, git has autotools support but it's optional17:45
loops yeah.. not sure where it's actually needed though.. never used it here17:46
geckosen1tor I'm trying it out17:46
it is nice because you only have to enter your prefix once17:47
when it is compiling it looks just like the linux kernel17:47
oh no.. make error: asciidoc: Command not found17:48
and the configure script didn't even know17:48
loops geckosen1tor, i don't think autotools is used nearly as much as the standard make file17:49
geckosen1tor ok17:49
Insount_away → Insount17:50
geckosen1tor well I installed asciidoc and it seems to be working.. but taking a while17:51
loops yes, the documentation takes longer than the source17:52
geckosen1tor yeah, now I don't have xsltproc17:52
git uses xml?17:53
thiago geckosen1tor: another way to make git stop using the pager: add | tee or | cat to the end of the line17:53
loops geckosen1tor, asciidoc uses xml17:54
geckosen1tor thiago: yeah I wanted to do it forever17:54
thiago then loops's suggestion17:54
though, once you get used to it, you start forgetting in other places17:54
more than once I've lost several minutes doing "svn annotate" without less, then having to wait for it to run again17:55
← Tv left17:56
loops geckosen1tor, not sure what distro you're on.. it's surprising that installing asciidoc didn't bring in all its dependencies17:56
geckosen1tor thiago: but I have infinte scrollback anyway17:57
loops: no I had xml, I didn't have libxslt17:57
apparently I need xsltproc too17:57
loops geckosen1tor, yeah asciidoc has a rather long dependency list18:00
geckosen1tor well I mean asciidoc didn't need that18:02
but the git make doc did18:02
→ nud joined18:03
loops geckosen1tor, ahh, indeed..18:03
← rtmfd_ left18:07
geckosen1tor oh no, now I don't have xmlto18:10
does it generate documentation from the source?18:10
→ evilchelu joined18:15
→ dduncan joined18:17
loops no.. it just converts the Documentation/*.txt files (which are in asciidoc) into the different formats18:18
← evilchelu_ left18:27
← hackbert-hh left18:28
← alley_cat left18:33
→ alley_cat joined18:35
← chris2 left18:38
← stouset left18:48
geckosen1tor ok finally finished building18:50
and now the git config worked18:51
it just modifies ~/.gitconfig if --global is specified?18:51
← rjbs left18:51
bdiego geckosen1tor, yes18:52
if you specified --global it read/write to ~/.gitconfig18:53
→ lcapitulino joined18:53
← alley_cat left19:00
← bentob0x left19:03
→ alley_cat joined19:06
→ evilchelu_ joined19:10
→ rtmfd_ joined19:21
← evilchelu left19:22
→ Tv joined19:26
→ Yuuhi joined19:27
← jmspeex left19:29
→ jmspeex joined19:29
→ DrNick joined19:37
← Tv left19:42
→ Tv joined19:43
← rtmfd_ left19:54
→ rtmfd_ joined19:56
← Tv left19:59
→ Tv joined20:01
→ evilchelu joined20:02
← rtmfd_ left20:07
→ Beket joined20:09
→ rtmfd_ joined20:13
← evilchelu_ left20:14
← Tv left20:17
→ Tv joined20:18
stillLotR → LotR20:20
← Tv left20:23
→ levander joined20:24
levander Is there a high level python interface to svk? I really just want to be able to do a "svk st" in python. And, probably in the future a "svk ci".20:24
Whoops, I meant that question for git, not svk.20:25
← rtmfd_ left20:31
→ vmiklos joined20:32
→ Dodji_ joined20:33
← geckosen1tor left20:39
hliusv561 levander: I'm not aware of any native interface. But some git subcommands are designed for scripting...20:40
levander: What would those commands do?20:41
lcapitulino levander: I've a python module for git, but currently it's ony capable of revision listinh20:43
* listing20:43
levander: http://repo.or.cz/w/git/libgit-gsoc.git20:45
levander: there's a directory called 'libgit-thin'20:45
→ rtmfd_ joined20:48
← Dodji left20:49
Dodji_ → Dodji20:52
← ofri left20:52
→ ofri joined20:59
mnemoc hi, if the update hook is done before loading the data pushed, how can the ciabot catch those commits?21:08
DrNick the example CIA notification script is a post-receive hook21:11
← rtmfd_ left21:16
→ rtmfd_ joined21:17
← rtmfd_ left21:21
→ rtmfd_ joined21:23
← d0k left21:25
mnemoc DrNick: oh, thanks :)21:32
→ KirinDave joined21:34
→ Tv joined21:45
← rtmfd_ left21:48
→ rtmfd_ joined21:48
→ geckosenator joined22:01
geckosenator is there a way to tell git not to track certain files22:02
vmiklos gitignore?22:02
← ofri left22:07
geckosenator thanks22:12
LotR are the colours gitk uses for the branches graph documented somewhere?22:12
← rtmfd_ left22:13
robin LotR: the colors do not have any meaning I think. Just different to make it less likely to have crossing lines of the same color22:14
thiago sometimes it changes for no reason I can understand :-)22:15
LotR oh, ok22:16
robin isn't random() reason enough?22:16
thiago when you have a merge (commit with two parents), note that the Parent: line in the log below matches the colour of the arrow22:16
LotR this is a git clone of an svn repo which was created with cvs2svn, so it might be weirder than usual :)22:17
← hFonteBoa left22:18
robin thiago: not here22:18
→ fultilt joined22:18
robin not that I have arrows, only lines22:19
jasam what get_pathspec function from setup.c is for?22:20
thiago oh, right22:20
mnemoc DrNick: i added some debugging on post-receive and it doesn't receive any argument... how can i know what was received?22:20
thiago I see one red and one blue, but the lines are brown and orange :-)22:20
← Beket left22:20
DrNick mnemoc: have you tried reading the hooks documentation?22:21
→ KirinDave_ joined22:22
→ Beket joined22:23
→ evilchelu_ joined22:25
→ rtmfd_ joined22:28
← Dodji left22:31
mnemoc DrNick: thanks, i forgot to read pre-receive info22:33
← evilchelu left22:37
→ aroben joined22:38
← alley_cat left22:41
← KirinDave left22:41
← Beket left22:43
← elmex left22:59
→ riddochc joined23:00
riddochc Hi, folks.23:00
evan is there a way to put gitignores in a repo?23:00
to have others pull them down23:00
oh, just in the root it looks like.23:01
thanks me!23:01
riddochc So, I finally got around to using rebase to remove a file from a project that shouldn't have been there. Nobody else has the repo (yet), and I've run 'git gc', but I'm wary that the old objects might still be around.23:02
Is there a way for me to be sure?23:02
robin clone the project23:02
loops good idea.. you can run git fsck --unreachable to see if there are any stray objects23:03
robin I think those use the same code23:04
clone won't look for unreachable objects and I see no reason for it to reference any others23:04
loops like i said.. cloning was a good idea :)23:05
riddochc Okay, thanks.23:05
robin you lose you reflogs and may have to copy the config from the old repo23:05
riddochc That's fine. There wasn't anything special in the config anyway...23:06
Looks good. When I did the clone, it counted fewer objects, so I think it's right now.23:08
← KirinDave_ left23:08
Insount → Insount_away23:08
riddochc Cool. Now the repo is suitable for release upon the unsuspecting public. Muhahaha. Thanks.23:14
← peff left23:14
← lcapitulino left23:18
evan has anyone worked on git-daemon being able to accept packs, ie push?23:19
robin it does23:19
evan and is stash just in 1.5.3?23:19
→ robfitz joined23:19
robin yes23:19
evan git-daemon does accept packs? huh... i don't think the docs say that...23:20
robin the git daemon has no authentication so it is not recommended for pushing23:20
evan does it do any passwords checks or anything?23:20
AH23:20
maybe i'll have to add that.23:20
robin generally read only setups should be done23:20
evan if git-daemon could use passwords in a safe way (using, say something like CRAM-MD5), would that be an acceptible change?23:21
robin MD5... oh23:21
evan or CRAM-SHA123:21
something that uses challenges over the wire23:22
so the password isn't transmitted23:22
robin ssh keys or X509 certs would be fine23:22
evan both of those are complicated.23:22
a SHA1 challenge auth is easy.23:22
← aroben left23:22
→ aroben joined23:24
← jmspeex left23:24
robin kerberos could be useful23:25
or TLS maybe23:26
evan yeah, krb5 would be nice.23:26
robin for now, however push via SSH is just fine23:26
for me23:26
evan being able to give out passwords to a public repo so people could push to it is a desirable feature for me23:27
hliusv561 robin: X509? Have you any idea what mess it is?23:28
If one needs to do password authentication right, then SRP is way to go. It even produces shared secret.23:30
robin hliusv561: fortunately I'm only a user of authenticated protocols and do not have implement them23:30
repo.or.cz uses ssh for pushing, no passwords23:31
evan robin: do you have to setup a key with them?23:32
them == repo.or.cz23:32
← robfitz_ left23:32
robin you have to create a key, yes23:32
evan thats what i'm doing now23:33
robin and give repo.or.cz the public part23:33
evan for my committers23:33
hliusv561 And then there is question of threat model. Protect only integerity of refs transfer or protect confidentiality of entiere transfer, or...23:33
evan hliusv561: i'd say that for most git projects, integrity is far more important23:34
most are public projects anyway23:34
robin encrypting is pretty affordable today even if you don't need it23:34
evan thats what ssh is for23:34
robin mr. bush may not be able to see everyting but he can pull instead23:35
pgp could be better than ssh keys since it is more widespread and the key server infrastructure is in place23:36
evan true23:38
thiago does PGP offer a stream encryption?23:38
evan no23:38
thiago it would be cool if it did23:38
evan you could exchange symetric keys in messages encrypted with pgp though23:38
hliusv561 Encryption is trivial. Key management is not...23:38
robin The commercial version includes tunneling and such23:39
hliusv561 evan: That's RSA key exchange, and I don't like it at all...23:39
evan hliusv561: use DH then.23:39
robin even SSH uses symmetric keys for encryption of the data stream23:39
evan or El Gamal23:39
thiago I think even PGP uses a symmetric key for encrypting the payload of the data23:40
hliusv561 robin: Right, but SSH uses DH, which (together with signatures) does not suffer from problems plaguing RSA key exchange.23:40
thiago the asymmetric key only encrypts the symmetric key and checksum, or something23:40
it's a LONG time since I read the format.23:40
hliusv561 thiago: It uses Diffie-Helman to agree the master key, which then is used to derive MACing keys and encryption keys.23:41
evan hliusv561: what are the problems with RSA exchange?23:41
the fixed nature of the endpoint keys that encrypt the symetric key?23:42
hliusv561 evan: Server key compromise allows decrypting captured past sessions.23:42
robin thiago: you're right http://www.pgpi.org/doc/pgpintro/#p1023:42
evan hliusv561: ah, so yep. fixed endpoint keys.23:43
hliusv561 evan: Worse yet, badly implemented RSA may allow oracle attacks.23:43
evan ahh23:43
robin hliusv561: badly chose keys?23:44
evan yeah, DH has the nice temporal property23:44
hliusv561 And DH always protects against passive attacks, no matter how much one screws up endpoint authentication.23:45
evan and, it's simplier to implement23:46
hliusv561 Diffie-Hellman (using group notation) is: A = aG, B = bG, S = bA = aB. A, B and G are public, a and b are private, and S is shared secret.23:47
→ Rednaxela joined23:47
Rednaxela How can I pull a tag instead of the head?23:48
robin doesn't it work?23:48
Rednaxela huh?23:51
← bdiego left23:54
← dduncan left23:59

Logs Search ←Prev date Next date→ Channels Documentation