IRCloggy #git 2007-01-06

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.

2007-01-06

→ xjjk joined00:19
tko I just saw a fairly long thread about umask.. any reason the user's current umask setting isn't used?00:38
→ xjjk-T42p joined00:47
clee if I have a given commit ID, and I want to reset the working copy to whatever the contents of the tree were as of that commit ID, how do I do that?01:05
git checkout doesn't seem to be my friend01:05
→ GeertB joined01:19
→ DrNick joined01:38
anholt clee: git checkout -b temp-branch <sha1>01:40
(lack of a git-seek <sha1> is a serious ui failing, I agree)01:40
→ Romster joined01:41
merlyn what would git-seek do?01:41
anholt like git-checkout -b temp-branch <sha1>, but with the temp-branch invisible to you. probably not let you commit until you branched properly.01:43
clee davej suggested I could do:01:48
rm -rf *01:48
git-read-tree $id01:48
git checkout01:48
aha01:56
looks like it's actually:01:56
git-read-tree $id01:56
git checkout-index -a01:56
ah, davej is a dirty liar02:00
(that doesn't actually work, for those of you following along at home)02:03
clee tries anholt's method02:03
anholt clee: I use it a lot when starting a bisect: git branch -d archaeology; git checkout -b archaeology <commit-of-interest>02:03
clee uh.02:05
that doesn't appear to work here, either.02:05
oh. wait.02:05
because I did git read-tree.02:05
right.02:05
anholt: hm. that doesn't really seem to work either, unless the entire repo is more screwed than I thought.02:07
anholt git diff HEAD and git status -a -v show nothing, right?02:08
if not, git-reset --hard HEAD to get back to sanity02:08
any series of commands beginning with rm -rf * just doesn't sound like a good idea to me.02:08
clee hahahaha02:11
hm. what about just doing git reset --soft commitish?02:12
dwmw2_gone → dwmw2_PVG02:12
clee or just 'git reset --hard commitish'?02:14
that seems to be doing something similar to what I had expected02:14
no, never mind. it still hates me02:15
clee sighs.02:15
Romster fighting with git can get frustrating :/02:15
clee yeah.02:16
clee shrugs02:16
clee it' sokay02:16
er, it's okay02:16
→ kmacleod joined02:17
clee so how does git-bisect do it?02:20
anholt clee: git reset --hard commitish would reset your branch, which is probably not what you want -- that's why I make a new branch to do stuff.02:23
clee hrm.02:23
so when I did git-read-tree e0bea0885e37aef8d2f9f1e3f769311ab080d86a && git-checkout-index -a02:24
I got the same results as when I did 'git-checkout -b nonexistent-branch-name e0bea0885e37aef8d2f9f1e3f769311ab080d86a'02:24
namely, a very small selection of the contents of the full repo02:24
(KDE has a dozen different folders, such as kdelibs, kdebase, kdeartwork, etc - the only one that showed up, in both cases, was kde-i18n; the only file that was changed in that commit was a file deep in the kde-i18n folder.)02:25
→ xjjk-T42p joined02:55
→ tchan joined03:51
→ mndrix joined04:40
→ rkaway3 joined05:15
xjjk-T42p → xjjk05:42
→ beu joined06:11
→ jeffpc joined06:16
dwmw2_PVG hm06:26
git-pull from inside china is not fun06:26
would be good if when a TCP connection stalled or timed out, it would pick up where it left off06:26
rather than having to start again from scratch06:27
→ xjjk_ joined06:42
→ xjjk joined06:44
→ xjjk joined07:02
→ devogon joined08:05
dwmw2_PVG hm. It is a little suboptimal, to say the least, that when 'git-pull origin' fails, after a couple of hours downloading, due to a single dirty file in the working tree, it doesn't actually _save_ anything so a second 'git-pull origin' is going to take another couple of hours08:32
Romster sounds bad, slow net speed or huge repo08:42
dwmw2_PVG both08:43
a kernel tree which hasn't been updated for a couple of months, and I'm in China08:43
Romster give a approxx08:43
dwmw2_PVG slow and unreliable network08:43
Romster ouch08:43
dwmw2_PVG Unpacking 36625 objects08:43
23% (8469/36625) done08:43
Romster is atm with so many cables broken and needing repairs08:43
dwmw2_PVG seems to be going faster this time08:43
this one is pulling by ssh from master.kernel.org instead of git://git2.kernel.org/08:44
Romster any reason to use git head?08:44
dwmw2_PVG what else would I hack on? :)08:45
Romster true but what part do you hack :)08:46
just cerious08:46
i use a few patches on the kernel atm, but they arn't upwards compatable wonder if i should get into the source more.08:46
dwmw2_PVG I hack any of it that offends me08:46
currently OLPC stuff08:47
that being what I'm here for08:47
Romster ah ok08:47
offends, meaning parts that don't function how you like it too08:47
dwmw2_PVG indeed08:48
Romster hope they finish fixing them cables its congesting up australia :/08:51
→ orospakr joined08:59
dwmw2_PVG Unpacking 36625 objects09:06
Disconnecting: Timeout, server not responding.09:06
ought09:06
er09:06
ouch09:06
another hour wasted09:06
cehteh dwmw2_PVG: cant you rsync the source first?09:19
dwmw2_PVG don't packs screw that up?09:20
cehteh in what way?09:20
i dunno what you exactly try to do .. but if rsync on the server is available then it should do the trick at least getting you a copy of the server's repo09:21
later you can branch/merge/clone locally from it09:22
dwmw2_PVG I have a kernel tree from November 15th09:22
I was trying to update it to HEAD09:22
→ lyakh joined09:22
cehteh does kernel.org pack incrementally? i think yes ... and then rsync should work perfectly09:23
dont forget the -P option09:23
dwmw2_PVG my local repository is packed.09:24
I could clone this revision on a machine with decent bandwith. Then make another clone, using that repo as alternate object directory, and pull HEAD from it.09:25
that way I'll end up with a single pack which contains all the objects I want09:25
cehteh thats just local stuff .. as i understand your primary problem is to get the remote repo or?09:28
i admit it would be nice if git can resume aborted downloads and stores partially loaded things somewhere .. dunno how well thats possible09:29
but when you rsync the remote repo, then all other operations become local operations and wont fail09:29
→ Oejet joined09:31
→ Guyou joined09:32
Guyou hi09:32
I need expert help09:32
on my repo, I have 2 branches09:33
the master and second09:33
but, using gitk, second is not tied with master09:33
dwmw2_PVG cehteh: yes, that's local stuff which I can do on kernel.org itself. Then I just have to download the pack file09:33
Guyou I think I made something wrong when creating it09:33
how can I tied second to master?09:34
Saying differently: how can I add a parent to a commit?09:35
dwmw2_PVG → dwmw2_gone09:35
cehteh create a new branch from master and merge second into it09:36
Guyou: i am rather interested, how did you setup a branch without a parent? :)09:36
cehteh would like that sometimes and misses a nice ui for that09:37
Guyou cehteh, I don't know09:37
cehteh hehe09:37
ok09:37
Guyou perhaps a wrong interpretation of what gitk is showing09:37
cehteh maybe09:37
Guyou humh... merge creates a tie at the "end" of the branch09:43
I wish to create a tie at begining of second branch09:43
→ z3ro joined09:44
Guyou I think I found10:00
git-rebase duplicates the branch10:01
dwmw2_gone → dwmw2_PVG10:10
→ robinr joined10:48
Oejet Guyou: If you want to put the earliest commit of branch second onto the latest commit of branch master, then yes that is what git-rebase does.10:51
Well, it does a bit more. See 'man git-rebase10:53
tko I'm trying to turn an evolving branch (git-svn mirror) into a series of patches with stgit (combining and reordering commits etc.) and I'm wondering if there's a way to see which commits I haven't yet processed with stgit?11:43
→ nud joined11:44
→ z3ro joined11:49
→ chris2 joined11:51
→ lyakh joined12:33
→ jwatt joined12:35
→ ferdy joined13:16
→ benlau joined13:26
→ Oejet joined14:28
→ dwmw2_gone joined14:29
xl0_msk → xl014:36
→ kanru joined14:53
→ z3ro joined16:55
→ z3ro joined17:17
→ jeffpc joined17:35
tko can I undo stg float somehow? there are conflicts because I didn't mean to move the patch into the position it's going :-/17:49
→ GeertB joined17:49
tko umm, and same with push.. if I push a conflicting patch, I can't pop it again without destroying (i.e. refresing) the patch?18:01
robinr stg push --undo18:01
→ DrNick2 joined18:01
tko :-]18:02
robinr not sure the whole float operation can be undone18:02
tko thanks, it helped18:02
robinr I didn't write code to do it18:02
(though I wrote the float command)18:03
tko push --undo undoes the push, now I can twiddle with the series to get the patch where I meant it to be18:03
DrNick2 → DrNick18:04
→ jwatt joined18:32
jeffpc $ git-pull18:59
Cannot get the repository state from git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git18:59
hrm18:59
seems like next is broken-ish18:59
:)18:59
hpa Like "next"?19:20
→ jwatt joined19:35
→ Eludias joined20:15
→ Alex joined20:17
→ devogon joined20:19
anders_ Is there a way to get the upstream sha of a changeset printed by "git cherry"?20:34
→ jwatt joined21:43
→ robfitz joined21:52
→ dwmw2_gone joined22:02
→ rofro joined22:09
→ rofro joined22:16
→ rofrol joined22:16
→ rofro joined22:17
→ GeertB joined22:39
→ hub joined22:40
hub how do I cancel a git-am?22:40
I haven't done the git-commit yet22:40
s/cancel/undo/22:40
→ GeertB joined22:50
hub how to I restore a mistakenly rm'ed file from the current repository22:54
I mean in my checkout I did "rm" and it is marked as deleted in git-status22:54
nud git-checkout file ?22:56
hub *sigh*22:56
I feel dumb22:57
hub need a git-for-dummies22:57
→ Romster joined22:57
→ ludd joined23:52

Logs Search ←Prev date Next date→ Channels Documentation