IRCloggy #git 2006-06-27

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.

2006-06-27

→ agile joined01:35
→ gambler joined01:37
→ anholt joined02:31
→ agile joined02:37
→ PugMajere joined02:43
→ DrNick joined02:44
→ matled joined03:04
z3ro anyone know if there is a way to get the initial commit it for use with git-format-patch (to generate a patchset from the begining of the git history to the end)03:39
s/commit it/commit id/03:39
→ tonyj joined04:12
→ tonyj joined04:25
→ Pug|Lagged joined04:29
→ DrNick joined05:24
→ Tv joined06:25
→ kanru joined07:15
→ robfitz joined07:21
→ dst_ joined07:27
dwmw2_gone → dwmw207:56
→ ferdy joined08:19
→ anders_ joined08:25
→ cworth joined09:09
→ anholt joined09:32
dwmw2 → dwmw2_gone10:00
→ chris2 joined10:58
→ coywolf joined11:05
→ GyrosGeier joined11:36
→ cworth joined11:37
→ Oejet joined11:44
→ robin joined12:19
dwmw2_gone → dwmw212:29
→ boto joined12:55
→ timlarson_ joined13:00
→ anholt joined13:04
→ robin joined13:19
jdl blinks13:30
Oejet eyeballs jdl.13:32
jdl Who you lookin' at?13:32
jdl got setn 4 patches this morning. The first one doesn't apply.13:33
jdl Feh.13:33
s/setn/sent13:33
Two bloody hunks and it doesn't apply.13:33
How do I get pick-axe to work again?13:39
I have one line I want in some file I want to figure out what happened to it.13:39
→ GeertB joined13:43
kampasky cg-log -S 'line' file13:43
:^)13:43
jdl Well then.13:44
davenport.home.jdl.com 1078 % git log -S '#define CFG_PCI_MEMORY_SIZE' -- include/configs/MPC8641HPCN.h13:46
fatal: bad revision '#define CFG_PCI_MEMORY_SIZE'13:46
Feh.13:46
Oejet jdl: Drop the space after -S.13:47
jdl Yeah, just discovered that. Thanks.13:48
→ benlau joined13:50
→ Tv joined13:51
→ agile joined13:52
pjz ls/whois jdl13:57
Oejet pjz: Good question.13:57
pjz oh, I know who he is13:57
just suprised to see him here13:57
jdl is Jon Loeliger; he's got a couple of git articles in the Linux Magazine13:58
jdl Hi!13:58
pjz woo, he's still here even :)13:59
jdl Indeed.13:59
There are other surprising folks present here occasionally too.14:00
pjz I'd imagine14:00
I'm trying to champion git at work, but... they're still using cvs14:01
jdl Smack 'em.14:01
Use the "All the Cool People use git." argument too.14:01
pjz bunch of oldschool ex-Kernel Group guys14:01
jdl In Austin?14:01
pjz yah14:01
jdl They should know better. :-)14:02
We could all meet for a brew and discuss it too.14:02
pjz everyone's always busy; difficult to do.14:03
jdl understands...14:03
pjz svn would be easier to push because the commands are essentially identical14:04
jdl You could do the underhanded git/cvs-server approach, too, perhaps?14:04
Set up a git tree, and then front it as CVS with git-server, perhaps?14:05
Make 'em send you patches.14:05
→ GeertB joined14:05
jdl I'm doing that in at least one instance.14:05
pjz I don't know how that would work14:07
jdl Technically, or politically?14:08
pjz technically14:08
b/c to work politically they'd have to be able to use identical (or very similar) commands from the cmdline14:09
jdl Are you the maintainer of the repo? Do they hav commit ability on it too?14:10
pjz everyone has commit ability n the main repo14:10
jdl Perhaps you should argue for Cogito.14:10
It is pretty similar to CVS in its presentation.14:10
Granted, cmds are spelled differently...14:11
pjz hrm14:11
cg looks pretty thin.. I mean, it doesn't seem to add much on top of git14:13
jdl Conceptually, it is different in that it totally hides the Index.14:13
pjz hrm, I must not grok git well enough yet14:14
jdl Perhaps my second article might help some. :-)14:16
pjz wow, if only I knew where I could get a copy14:19
though I did hand the secretary the Linux Magazine subscription card the other day and say 'order this'14:20
jdl Heh. Cool.14:20
So, the online version of Linux-Mag articles appear online 60-days after print.14:20
My 2nd article appeared in the May 2006 issue (Apache cover). It might be about time for it to appear online. Not sure yet.14:21
pjz it's not 60 days after May yet14:22
sadly14:24
jdl Yeah, I just checked too. Not yet.14:25
Ooo. Not even the April edition is up yet. :-(14:26
kampasky pjz: it doesn't really add much on top of git (well it does automate couple of things for you), but what it does add (and is its main purpose) is an intreface that should be trivial to grok for a cvs user14:29
→ int joined14:38
pjz hrm, okay, so it may be what I need to convert the hordes here14:38
jdl What is the nature of their reluctance?14:39
pjz I haven't lobbied too hard yet, just mentioned that "something better than CVS would be a good idea"14:39
but I can predict pushback if it's too difficult to learn or grok or use.14:41
→ anholt joined14:44
pjz so: "Why migrate at all?" is probably the first question. Having a set of best practices and advantages they bring would be nice, but I'm not holding my breath.14:45
jdl Have you ever been annoyed by CVS' branch behavior?14:46
Do you do any form of distributed development?14:46
Is CVS painfully slow for you?14:47
pjz no distributed development really15:01
cvs isn't painful yet15:01
otoh, atomic commits are a win15:02
er, changeset commits instead of individual file commits I mean15:02
int i wonder if easy way exist to get changes from http://www.kernel.org/git/?p=linux/kernel/git/airlied/intelfb-2.6.git;a=summary ?15:08
jdl Just clone the tree or pull it?15:08
int ok, i have cloned tree, how can i get intelfb specific changes?15:10
jdl What do you mean?15:11
You have them in that tree, right?15:11
int jdl: yeah15:11
jdl: i want a patch to apply it to plain 2.6.1715:11
jdl Is 2.6.17 tagged in that tree? Is that tree cloned from linus-2.6.git?15:12
int jdl: yeah, i see v2.6.17 tag15:12
jdl Are your changes on a branch, tagged or where?15:13
The intelfb changes you want.15:13
How are they added to that tree?15:13
You might be able to do "git diff v2.6.17 HEAD" straight up?15:14
git diff v2.6.17 intelfb15:14
Pug|Lagged → PugMajere15:14
int cg-branch-ls show only one branch "origin"15:14
jdl You are working on the "master" branch, though?15:15
int git diff v2.6.17 HEAD shows a lot of non-intelfc specific changes15:16
jdl: i din't make any changes in that repo15:17
jdl Understood.15:17
→ ruskie joined15:17
jdl So maybe do this:15:18
make a new branch with vanilla 2.6.17 (from linus-2.6.git) pulled into it.15:18
Then compare your head in this tree now, to that 2.6.17 pull.15:19
int jdl: into the same repo?15:19
jdl But I feel you should have 2.6.17 here, in that repo already.15:19
You can pull any branch from any repo you want into your repo.15:19
Wait. You can _fetch_ any branch....15:20
Sometimes you can pull... :-)15:20
OK. That repo has the patches you likely want directly on a branch.15:23
That branch appears to be called intelfb-patches.15:23
Invoke the name "pasky" to figure out how to add that branch to your repo when using Cogito...15:24
Oh mighty pasky-ky-ky!15:24
int cg-clone copy only one branch from repo?15:27
jdl I believe so.15:27
but cg add-branch I think is your key here.15:27
(I don't use Cogito, so I'm not sure.)15:28
int jdl: ok, i've just added intelfb-patches branch, how can i get diff?15:31
jdl "git diff v2.6.17 intelfb-patches" might do it.15:32
"git format-patch v2.6.17 intelfb-patches" might also.15:33
cg-mkpatch in some form too...?15:33
int jdl: cg-status says that intelfb-patches has same id as origin and master branch15:36
it seems that git diff v2.6.17 intelfb-patches gives me diff between v2.6.17 and head15:37
how can i see which branches exist in remote repo?15:38
jdl git-ls-remote15:39
do this: git format-patch --stdout v2.6.17.. | less15:40
It will generate a series of patches...15:41
int i'm now doing: cg-branch-add linux-2.6 git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git, and then i will try to see diff betwen intelfb-patxhes and linux-2.615:41
jdl: git-ls-remote says that master and intelfb-patches has different ids, but in my repo they have same id (same as master in remote repo)15:49
can i get diff between two branches in remote repo?15:49
jdl Did you "pull" or "fetch"15:50
?15:50
Pull will have merged them.15:50
no, the branches have to be local to diff them.15:50
You might refer to HEAD^ or intelfb-patches^ to get the commit prior to the merge commit if that is what happened. Or so.15:51
How come "git apply --check my.patch" doesn't show whitespace warning issues?15:52
int jdl: it seems i added branch in a wrong way15:54
how can i add remote branch with git?15:54
jdl various ways: git fetch URL remotename:localname15:55
You can add a "Pull: remotename:localname" line to a .git/remotes/... file as well.15:55
→ cworth joined15:56
jdl Didn't cg do this for you already?15:56
int ok, i found my mistake. right way: cg-branch-add intelfb 'git://git.kernel.org/pub/scm/linux/kernel/git/airlied/intelfb-2.6.git#intelfb-patches'15:58
now cg-diff -c -r intelfb show me right changes :)15:58
jdl: thanks for help!16:01
jdl Excellent. Most welcome.16:06
→ GeertB joined16:13
dwmw2 → dwmw2_gone16:26
→ z3ro joined16:42
→ chris2 joined16:51
→ GyrosGeier joined17:06
→ boto joined17:24
→ PugMajere joined17:34
→ flobber joined17:55
→ gitster joined18:07
→ krh joined18:17
→ juvenis joined18:18
jdl is back from lunch!18:34
→ mchehab joined18:36
← mchehab left18:37
→ Eludias joined18:49
→ moh joined18:54
→ mort_ joined19:12
mort_ → moh19:14
matled gitster: did you see the documentation patch for git grep also mentionioning support for multiple patterns? and what do you think about GIT_TRACE and git grep --and (this should probably get a short option too)?19:22
→ Beber` joined19:24
gitster (1) yes. (2) fun toy with slightly dubious practical use but not strongly opposed to it -- if it does not deal with some commands as Johannes say it might be more confusing than helpful. (3) grep --and may sometimes useful but I am wondering if we would rather want to do something like expression `find` command let's you build. IOW,...19:25
git grep --extended-expression '(' 'foo' -o 'bar' ')' -a 'frotz'19:25
About -e documentation, (1-1) real grep lets you have multiple -e, so it might not be worth mentioning, (1-2) if we document it, the way to do so may be "[ -e <pattern> ]..." with textual commentary that says "if you have only one pattern that cannot be mistaken as an option, you can omit -e".19:27
matled that looks nice too, one common thing i just do is just adding another |grep foo to narrow the search down until I have what I want, and then I've got no context anymore, allowing to specify how the patterns are combined would be fine too, but I don't know how complicated that would be to implement and how much use it still gets if it is too complicated to use19:28
gitster that's true. -e 'foo.*bar' -e 'bar.*foo' does not scale, so I do understand wanting to say -e 'foo' --and -e 'bar', but I am just wondering if that is too restricted enhancement.19:31
matled err19:32
gitster once a "too restricted enhancement" is in place, more generic enhancement if we ever want to do so (the 'expression' thing, which I am not sure either) needs to be made backward compatible, so...19:32
matled so with the extended-expression way you'd have to do something like -e foo -a -e bar -a -e baz to get the same result as --and?19:36
dkowis → evilDiktatz0r19:38
matled so I'm not sure if this is really a problem, with the extended-expression thing one could specify exactly how the patterns are combined (using braces, and, or, not) and the other options just changes the default behaviour if there is more than one pattern from matching if one pattern matches to matching when any pattern matches, so this does not affect the more explicit first version19:44
→ moh joined19:44
→ anholt joined19:45
→ ShadeHawk joined19:55
→ yacc joined19:56
ShadeHawk How strongly hardcoded in git is "one line summary + empty line + longer description" format of commit message?19:56
I am about to import from CVS, where this format was _not_ preserved.19:57
gitster Only porcelainish. The core does not care.19:57
ShadeHawk which commands would break, and which would use first line for summary, but not e.g. add empty line where isn't one, or remove second line of message?19:58
pasky porcelainish shouldn't care per se either, it's only that in some tools text will be cut off in the middle of sentence19:58
except that iirc there was some bug in git-rebase19:58
but gitster will know that better :)19:58
ShadeHawk was?19:59
I won't rebase CVS imported sources, I think...19:59
I'm just fed up with CVS. No file copying support, branching I cannot understand...20:02
→ juvenis- joined20:05
Oejet ShadeHawk - CVS: 0 - 120:05
→ argyro joined20:05
argyro How do I get this source?20:05
http://bcm43xx.berlios.de/?go=Development20:06
What do I type...I can't figure it out. I've used subversion and cvs before..but this doesn't work/make sense for me...here is what I get.20:06
jdl git clone <URL>20:06
cg-clone <URL>20:06
argyro http://www.rafb.net/paste/results/sJPnlR74.html20:07
I did that jdl20:07
I get those results.20:07
jdl You don't have git installed.20:08
You need _git_, not the git file manager.20:08
Not the GNU Interactive Tools package.20:08
Type this to verify you are using git, the SCM tool:20:09
git --version20:09
git version 1.4.0-dirty20:09
argyro jdl, ok, I have cogito installed; what do I type now?(I'm on Ubuntu Dapper)20:11
jdl Do you have git installed?20:11
robin argyro: try git-clone instead of git clone20:11
jdl What does "git --version" say?20:12
argyro http://www.rafb.net/paste/results/zIpxhP72.html20:13
That is what I get with git-clone robin20:13
jdl Don't do this as root.20:13
robin argyro: you have and extra clone there, remove it20:13
jdl No ..20:13
loses.20:13
eleusis ShadeHawk: hello20:15
argyro http://www.rafb.net/paste/results/JyiESk23.html20:15
I got rid on one "clone" and I typed the command as non-root20:16
jdl Looks like it worked.20:16
do you have a wireless-2.6 directory now?20:16
argyro jdl, It's just hanging..should I control c?20:16
robin argyro: uninstall the git-package if you don't need it20:16
jdl No!20:17
do a "top" in a different window.20:17
argyro jdl, yes, I have that directory...but it's empty20:17
jdl Waaaait for it....20:17
argyro It's been 3 minutes20:17
jdl How fast is your link?20:17
argyro 10mbit20:17
jdl You are dowloading a lot.20:17
argyro It's just a small wireless module20:18
jdl 150M20:18
argyro damn20:18
Why are there no files in the directory yet?20:18
jdl Ah, not full linuux tree then?20:18
robin usually git clone starts talking numbers right away20:18
argyro jdl, no20:18
jdl Do you have firewall problems?20:18
argyro no20:18
iptables is accepting all20:19
robin jdl: I tried the command and not much happens here either20:19
jdl Cool.20:19
argyro robin, Are you getting the same trunk as I am?20:20
jdl This is a full linux tree20:20
evilDiktatz0r → dkowis20:20
ShadeHawk When git-rebase was corrected c.f. commits not using "summary\n\nlonger description" commit format? What was the bug?20:21
robin the data starts flowing after a while20:22
jdl Waaaiit for it...20:22
It is possible that the repository isn't packed on kernel.org too.20:23
robin maybe high load20:23
jdl Maybe one of k.o's machines is slower than the other too.20:23
→ _yacc_ joined20:23
_yacc_ → yacc20:24
robin ShadeHawk: I think rebase was using git-am (the mail import stuff)20:26
→ Paavo_ joined20:30
jdl Getting any data yet, argyro?20:30
argyro jdl, no20:33
→ dwmw2_gone joined20:34
robin strace -p <pid-of-git-clone> and se if it does anything, else abort20:37
dwmw2_gone → dwmw220:40
→ cworth joined20:47
jdl What version of git are you using?20:49
→ Paavo joined20:57
→ lilo joined21:04
→ DrNick joined21:07
ShadeHawk Hrmmm... cvsimport did error out when I tried to import CVS sources halfway between modules (i.e. there is basename module, basename/src/name-a, and basename/src/name-b; basename, name-a and name-b are CVS modules, I tried to import from basename/src)...21:12
with rather strange error message...21:12
→ cworth joined21:21
→ FrankB joined21:30
FrankB Hello! I have just tried to merge another branch into master, while master had uncommited local changes. The merge failed and I can't resolve the errors at the moment. Is there a possibility to revert the changes that the uncompleted merge introduced while keeping the changes that were present before?21:35
→ anholt joined21:37
→ merlyn joined21:40
merlyn what were the instructions for restarting a stopped "git-svn fetch" again?21:40
git-fsck-objects21:41
and something about making sure the svn is consistent21:41
and has the right ref for the remote21:41
normalperson merlyn: are you using the svn libraries or command-line client?21:55
if you're using the libraries, nothing, really21:57
with the command-line client, svn revert -R .git/svn/git-svn/tree, svn up -r<last revision imported>21:57
err svn up -r <last revision imported> .git/svn/git-svn/tree, of course21:58
you can get the last revision imported by looking at the git-svn-id: line of the current remotes/git-svn commit21:58
that's probably the fastest, most bandwidth efficient way21:59
rm -rf .git/svn/git-svn/tree .git/svn/git-svn/.rev_db .git/svn/git-svn/index && git-svn rebuild also works :>22:00
→ DrNick2 joined22:01
DrNick2 → DrNick22:02
→ GeertB joined23:23
dwmw2 → dwmw2_gone23:25
→ Beber` joined23:45
dormando what're a few other distributed RCS things?23:51
I forget the names :(23:51
pasky there's git's "amicable arch-nemesis" mercurial, others worth a mention might include monotone, darcs, possibly svk23:57
DrNick has anyone talked about replacing SHA-1 in git with SHA-256 or something else?23:58
pasky not very seriously23:58
but it should not be very hard23:58
DrNick just introduce some new tree/commit/tag object types that use the new hashes?23:59
pasky if you toss backwards compatibility, it should be just a matter of replacing a few routines, actually; so updating documentation might be more work ;)23:59
→ gitster joined23:59
pasky no, that's quite a bad way to go about it, I'd say23:59

Logs Search ←Prev date Next date→ Channels Documentation