IRCloggy #git 2006-06-11

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.

2006-06-11

coywolf joined00:10
Oeje1 left00:26
boto joined00:27
njs` joined00:43
ShadeHawk are the *_HEAD documented anywhere?00:58
boto_ joined01:07
boto__ joined01:20
tchan joined01:30
ShadeHawk what does 'fast forward' means (in git)?01:34
merlyn it means that your merge is a descendent of your current state01:35
so it doesn't have to do any work01:35
njs` you ask git to merge A and B, it notices that B is already a descendent of A, so it just returns B as the merge instead of creating any new objects.01:36
^^ can anyone confirm this description?01:36
merlyn Yeah - that makes sense01:38
you're tracking a - b - c - d - e, and you pulled at c, and now you pull at e01:38
there's nothing to "merge"01:38
it just makes all the files like e01:38
njs` so your head just jumps from c to e directly, yeah.01:38
merlyn however if you edited and committed something, it has to figure out the merge between your changes and e01:39
and you'll never be fast-forwarding again. :)01:39
ShadeHawk i.e. merge(A = branch1, A <- B <- C = branch2, into => branch1) results in A <- C' = branch1 (where C' has different parents from C)?01:39
and C' only parent is A, or C' has parents C and A?01:40
njs` merlyn: unless you push01:41
merlyn oh - right, you can eventually end up with the same files.01:41
then you have a common thread again01:42
ShadeHawk of 'fast forward' refers only to pull, and pull(A = origin, A <- B <- C = remote#master) results in A <- B <- C = origin?01:42
njs` I think your description of "pull" there is correct, but that it's not specific to pull -- pull is just communicate+merge?01:42
(in git)01:42
merlyn I think of it during pulling though :)01:43
njs` yeah, pulling would be the most common case to run into this :-)01:43
but another case might be, e.g., you have a feature branch that's ready to merge to trunk, so first you merge from trunk to the feature branch, to make sure nothing new has happened to break it, and then you merge from the feature branch to trunk...01:44
if nothing has been committed to trunk in between the two merges, the second merge will be a fast-forward01:44
pasky njs`: yes, your description is accurate01:47
ShadeHawk Hmmm... from what I seen in git-pull.sh it looks like fast-forward is/can be specific to pull/push. The situation you describe above is trivial merge, not fast-forward.01:47
I guess.01:47
correct me if I'm wrong.01:47
(IIRC push *requires* fast-forward)01:47
pasky njs`: and yes, the second merge in your latest message will be fast-forward01:47
njs`: it's a common counter-case for anything that relies on merge parents ordering ;)01:47
at least in cogito, fast-forward is just a type of merge (note that that's somethign else than a merge strategy)01:48
so if you cg-merge, it can do either a fast-forward merge or a tree merge01:49
I'd expect git-merge to be similar01:49
merlyn wow. sittin' on #git, answering questions01:50
just like the rest of my life. :)01:50
pasky yeah, git-merge can do fast-forward too01:50
ShadeHawk i.e fast-forward merge in the branch1 == a, branch2 == a-b-c case means just do branch1 := branch2?02:00
gitster joined02:05
mike_m joined02:12
DrNick joined02:35
boto joined05:51
meyering gitster, thanks06:46
Oeje1 joined07:28
_JusSx_ joined08:49
ferdy joined08:52
Eludias joined09:41
benlau joined09:44
chris2 joined10:04
Oeje1 Phew, programming makes me hungry.10:13
_JusSx_ which language?10:31
what are you programming?10:32
Oeje1 Git. I've just sent in a patch.10:32
ShadeHawk The safe_strncpy = strlcpy patch?10:35
Oeje1 Yes.10:35
_JusSx_: I've done a sweep replacing strncpy() with safe_strncpy().10:36
ferdy joined10:50
mike_m left11:10
GeertB joined11:14
eugene joined11:33
pschulz01_w_cuppschulz0111:46
alley_cat joined13:29
Beber` joined13:32
_JusSx_ does git support keywords expansion?13:43
agile joined14:23
Temaris joined15:21
Temaris hi. documentation for cg-admin-rewritehist mentions --msg-filter option but the script itself doesn't accept it.15:34
alley_cat arwed@Otherland:~/Work/git/grimoire$ time git commit -F .msg15:50
real 1m17.930s15:50
any idea why that takes so long?15:50
it's after a failed cherry-pick that needed manual merging15:50
GeertB joined15:51
Oeje1 alley_cat: What does git-status tell you?15:54
Got to go out for a bit.15:55
DrNick joined15:58
z3ro joined16:54
GeertB joined17:42
krh joined18:33
Oeje1 left18:47
Oeje1 joined18:49
Eludias joined18:53
Beber`ABasLeFoot19:49
Frut joined19:59
Oeje1 joined20:54
robfitz joined20:55
dwmw2_gonedwmw221:09
jdl So, can someone with some C language lawyerese?21:48
Like, 3.14 versus 31.4f for example.21:48
And, to make matters worse, in GNU C versus VC++ under MS.21:49
Oeje1 left22:00
jdl I have two versions of some source base. The first I used to create a git repo. No problem.22:02
I want to take the second, overlay it on the new git repo, and then commit it as the second git rev.22:03
So I un-tar'ed the second version into the git repo.22:03
Some files changed, and many did not.22:03
But, of course, the time stamps on everything changed.22:03
pasky git-update-index --refresh22:04
jdl So the index _thinks_ things are newer...22:04
pasky if you really are so lowlevel22:04
jdl Yep.22:04
And it won't touch the "really changed" files, right?22:04
pasky right22:04
jdl Exactly. Slick. Thanks!22:05
somegeek joined22:19
ruskie joined22:19
ruskie lo22:20
lilo joined22:20
Ramune joined22:30
Ramune hi all. when doing a git cvsimport, has anyone seen: "Branch FOO does not exist!"22:31
followed by "fatal: Not a valid object name bar" "read-tree failed: 32768"22:31
doing an import of work-related src, and it's blowing up on me. using cvsps 2.1 and git 1.4.022:32
with -v -p -x -p --no-cvs-direct added, i'm getting: invalid cache revision line 'file:path/to/file with spaces; another file with spaces; pre_rev:INITIAL; post_rev:1.1; dead:0; branch_point:0'22:40
'!' another file with spaces'22:41
laj_ Ramune: sorry no idea, but try the mail-list it usually gives quick answers (adr. in topic)22:49
ruskie I'd like to use virtual users for git access can it be done, was it done, is it documented someplace?22:50
Ori_B .22:51
er, oops, hit keyboard accidentally. ignore it.22:51
ruskie lol22:52
Ramune i guess i'll see if parsecvs manages to finish on the archive first and see if it pukes too, for additional info.22:52
jdl So, I fat fingered a "git branch -r" into a "git branch -d".23:00
Silently executed. But what did it do? :-)23:00
kylem rm .git/refs/heads/${branch} more or less.23:00
jdl Yeah, but _which_ branch, if any. I didn't specify one at all.23:02
Seems like it should have groused about a missing branch arg.23:03
Or, perhaps it is "silently correct". Dunno.23:03
krh jdl: no branches23:04
jdl: git-branch -d takes a list of branches to delete23:05
empty list == noop23:05
jdl So it is correct in the sense that it means "delete all the branches I specify", and I happen to have named none...23:05
krh hehe, up23:05
jdl OK, that's fine.23:05
krh erh, yup23:05
jdl cvs update: warning: unrecognized response `Use of uninitialized value in concatenation (.) or string at /usr/bin/git-cvsserver line 1817.' from cvs server23:06
Feh.23:06
jdl digs through git-cvsserver now...23:08
jdl $filename = $state->{prependdir} . $filename;23:08
kreaturr jdl: what cvs command were you performing?23:08
jdl "cvs update"23:08
kreaturr jdl: what's the cvs client version?23:09
jdl davenport.home.jdl.com 922 % cvs --version23:09
23:09
Concurrent Versions System (CVS) 1.12.9 (client/server)23:09
kreaturr hmmm.23:09
jdl It's the server side, likely.23:09
kreaturr I got similar errors using an older version.23:09
jdl Question is, is it due to my running an advanced git daemon?23:10
Or is it independent of git-daemon?23:10
kreaturr perhaps. I think it's independent.23:10
do me a favor, try 'cvs up -d'23:10
jdl Same stuff.23:11
kreaturr ok, thanks. probably unrelated to my errors then.23:12
jdl Line 1732 too.23:12
kreaturr yeah, cvsserver wasn't correctly setting the $state23:12
which sounds like your problem, too.23:12
jdl Code guards for empty filename, so this must be the "prependdir" being empty.23:12
Looks like it ultimately did update files, so cool there.23:13
kreaturr seriously, I'm looking at an old diff and that those lines are exactly what I tweaked.23:15
along with a couple other small lines.23:15
hmm, there was a thread on the ML about it.23:16
jdl needs to catch up on ML...23:17
coywolf joined23:21
kreaturr jdl: http://marc.theaimsgroup.com/?l=git&m=114861313921708&w=223:24
nothing much, all I did was quiet the errors and make sure it finished processing (which it may or maynot have in your case)23:24
jdl what's the subject on that ML entry?23:29
laj_ "Re: git-cvsserver wart?"23:30
kreaturr yeah. ;)23:31
jdl kreaturr == Cam?23:33
kreaturr nods23:33
kreaturr old nick, not really sure why I keep it. habit I guess.23:34
in any case, quieting the errors was enough for me - and I haven't got a chance to dig into it again to really be helpful.23:37
jdl Ooo. Dinner! BBIAB23:39
GeertB joined23:50

Logs Search ←Prev date Next date→ Channels Documentation