IRCloggy #git 2009-06-21

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.

2009-06-21

fynn interesting, thought tbh I'm not sure what you mean :)00:00
robinr tomoj: yes, but most people assume we can prove things with math00:00
fynn I don't know enough about medicine.00:00
robinr but that is based on axioms that are true, but if they aren't we're lost00:00
tomoj robinr: prove things about what? the real world? the proof is still valid even if the axioms don't hold00:00
fynn btw, about the alpth_1 / beth_1 thing, it's not a problem: it's well known that certain theorems would be true under axiom set A, and false under axiom set B00:00
Ilari fynn: Also probably some stuff related to education.00:00
fynn "education" is a social science, so it's not really a science at all :)00:01
basically, there has to be a huge, outrageous conspiracy of very strong forces to keep any real science in category 4.00:01
robinr tomoj: the proof in mathematical sence, but not in relation to the world around us00:01
for which we use math00:01
fynn tomoj: the proof in mathematics stay in mathematics.00:02
err, for robinr00:02
robinr yea, :)00:02
fynn the fact that there is this curious correspondence between mathematical proofs and the real world is of no importance or relevance for math itself.00:02
Ilari fynn: If there is lots of money riding in the wrong interpretation, such conspiracies _CAN_ exist.00:02
fynn thus, it can never be a mathematical problem if a mathematical concept applies or not to the "real world" (whatever that is)00:03
Ilari: feel free to PM what exactly you mean :)00:03
tomoj my logic professor told us he worries that the real numbers don't exist. I rofled00:03
cpg|biab → cpg00:03
fynn tomoj: dude, you're obliterating my respect for UAT00:03
robinr tomoj: isn't he right? numbers are human constructions00:04
fynn Ilari: you seem to hint at something, but I have no idea what it is.00:04
robinr: so we constructed them, so how can they not exist?00:04
robinr but those aren't "real"00:05
tomoj fynn: you can be a competent logician and a realist at the same time00:05
fynn that's like saying "I'm thinking of 2, but now I am concerned that my thought may not exist"00:05
← __iron left00:06
Ilari fynn: That would imply that either 1 wouldn't exist or addition of integers wouldn't exist.00:06
:-)00:06
fynn tomoj: he's a logician, hence he's teaching in the Philosophy department, in a math-related field, yet holds such naive beliefs that any first year philosophy of math course would revise?00:06
→ __iron joined00:07
robinr Asimov wrote a short story on a professor that though mathematicians were mysticist, because they believe in imaginary numbers00:07
tomoj fynn: I'm sure his philosophical views are more nuanced, he was sort of just joking with us and going into detail would've been out of the scope of the class00:07
fynn robinr: funnily enough, the origins of math have relations with mysticism; cf Pythagoras.00:07
← sunoano left00:08
robinr yes, and gambling00:08
← Inthewings left00:08
fynn and destroying greek ships with huge cranes!00:08
in ancient Greece, mathematicians would be adored by women, second only to the adoration of git experts.00:09
hence the common phrase: "getting laid like a gitster".00:10
← bx2 left00:10
Ilari One way to express converse of continum hypothesis is: Is there set thats strictly bigger than integers but strictly smaller than reals?00:10
robinr how would that help me get some sleep?00:11
fynn Ilari: hm, how can that be, seeing that the set of reals is a superset of the set of integers?00:11
Ilari Continum hypothesis says no, there isn't. But adding axiom of symetry to ZFC yields that there is such set.00:12
→ sunoano joined00:12
→ melvinram joined00:12
fynn 's brain asplodes00:12
fynn my personal way to deal with math is to limit my interest to its computer-science aspects.00:12
the wonderful thing about computers, apropos our discussion, is that they are both math and real world.00:13
bremner fynn: my personal way to deal with computer science is to limit my interest to its math aspects :-)00:13
→ demio joined00:13
fynn they are an aspect of the real world which is a closed system, governed by axioms.00:14
bremner: then a Venn diagram would clearly show that we are both interested in the exact same things ;)00:14
bremner: p.s., what do you study / do?00:15
Ilari In ZFC, one can't prove that such set exists and one can't prove that such set doesn't exist.00:15
← plediii_ left00:15
robinr Zzzz00:16
tomoj I find this one more brain asploding: in ZFC, you can prove that there is an uncountable set, yet there are countable models of ZFC.00:16
→ sakyamuni joined00:16
Ilari Another way to express continium hypothesis: The size of smallest uncountable set is size of the set of reals.00:18
bremner fynn: theoretical CS00:18
Ilari Interestingly, size of reals is 2 to power of size of integers.00:19
fynn there's a proof of that?!00:19
← memiux left00:20
klbate For some definition of "power".00:20
← aspotashev|eeepc left00:23
→ cemerick joined00:24
Ilari That impiles that powerset of integers and set of reals are isomorphic (there are functions that can map arbitiary real to subset of integers and arbitiary subset of integers to real number).00:24
→ aspotashev|eeepc joined00:24
fynn tomoj: "you'd only need to do that if you wanted to push foo changes and your local master commits at the same time" - yeah, the emerging insight is that it's actually *better* if foo works directly against origin/master.00:24
tomoj fynn: foo pushes to origin?00:25
fynn tomoj: pulling it with a --rebase, pushing back to it so it FF's the new foo commits.00:25
tomoj wow that's bizarre00:25
Ilari: isomorphic? really?00:25
fynn yup, foo pulling --rebase and pushing back commits, just like master.00:25
tomoj fynn: and then pull with rebase in master to get the new foo commits?00:26
← JNRowe left00:26
fynn Ilari: stuff gets twisted like that once you start to discuss things like infinite sets... maybe we should treat that notion as simply ill-defined?00:26
tomoj: yup, the beautiful thing is that's how master works _anyway_.00:27
tomoj fynn: what do you mean?00:27
→ wshimmy joined00:28
fynn tomoj: so, foo pulls --rebase when it wants to update itself, and pushes the commits it wants to push. master does the same thing.00:28
this is a super simple workflow. no caveats, no possible problems, no edge cases.00:28
→ foobarfoo joined00:29
← Yuuhi left00:29
→ plediii_ joined00:29
fynn our mistake yesterday is that we thought we should find a way for foo to interact with master, when in fact there's no reason to do that and it only creates a problem with the rebases.00:30
Ilari fynn: You can have local repo with topic branches only. No master branch.00:30
← xyz left00:30
fynn Ilari: yeah, though for our needs, we want a master branch (effectively, the trunk)00:30
every developer has a master branch, and 0 or more topic branches.00:30
Ilari fynn: Setting push to tracking comes especially handy in such setup (requires quite recent version).00:30
fynn Ilari: I thought branch tracking is a fairly old feature?00:31
is "push tracking" something different?00:31
Ilari fynn: Directing pushes to tracked branch. It is new feature.00:32
← plediii_ left00:32
fynn awesome. in any case, we're generally running latest git, so it's not a problem.00:32
← robinr left00:32
tomoj Ilari: are you talking about remote.origin.push?00:32
Ilari At least 1.6.3 required. Nope, push.default00:34
fynn so now, "git checkout --track origin/foo" would cause not only future pulls to pull from origin/foo by default, but also future pushes?00:35
Ilari fynn: It isn't default for pushes.00:36
fynn Ilari: so, the only thing is that you can now set push.default, whereas before you had to always specify an explicit push target?00:36
← parasti left00:37
tomoj before 'git push' did matching00:37
now you get to pick00:37
Ilari fynn: The old behaviour is one of the values of that setting (the default one).00:37
← sps left00:37
→ fujin joined00:37
tomoj fynn: I am reluctantly accepting that you may be onto something00:37
I always felt like I had to push to origin/master from master, but don't know why I felt that way00:38
← w|shimmy left00:39
tomoj still, I keep my master clean, so it doesn't get rebased, and I can rebase foo onto master, fast-forward master, and push00:40
but then, what's the point in even having a master branch if it's always clean? :(00:40
Ilari fynn: Regarging merges: One can draw certain number of distict maximal paths through history (often that number gets huge, Linux history some time ago had ~2^2074 such paths). Each such path describes a possible observer: What commits are made locally and what is merged in?00:42
← MisterN left00:42
Ilari fynn: Maximal meaning that the starting point has no children and ending point has no parents.00:45
← asksol left00:46
← kpreid_ left00:48
→ kpreid_ joined00:48
→ paltman94 joined00:48
tomoj fynn: why exactly do you have a need for a master branch?00:50
fynn tomoj: regular development, unrelated to foo features.00:51
trivial example being trunk bug fixes, or any sort of urgent trunk patches.00:52
tomoj but why not just branch off origin/master and fix in there?00:52
fynn well, seeing the way we work, we could do that every time.00:52
← cemerick left00:53
fynn every pull --rebase is exactly the same a branching origin/master locally, committing, and FF back.00:53
there's basically no reason to keep a persistent master if you always pull --rebase to it.00:54
← wsc left00:55
Ilari The number of paths for n commits is O(e^(n/e)).00:55
fynn finds Ilari strangely fascinating.00:55
fynn your description of this stuff is really high level. it's a unique way of looking at something as practical as a VCS.00:56
← melvinram left00:56
fynn tomoj: (in fact, I'm not sure there's any reason to keep a persistent local branch if you always push any local commit to origin as soon as its done - which is what we generally do here)00:57
the only justification for a local branch is to a place to keep commits that for any reason you're not ready to push to its remote origin yet.00:58
tomoj fynn: is a commit "done" when it's committed?00:58
fynn yes, "as soon as it's made"00:58
tomoj fynn: so you push without running tests? :O00:58
or only commit after running test?00:58
fynn tomoj: indeed we do; our test suite takes an hour to fully complete.00:59
tomoj damn00:59
fynn so it only really runs on the integration server.00:59
so up to an hour after the commit, you may get a message that your commit broke the build.00:59
← werdnativ left01:00
→ pinpinlelapin joined01:00
→ Roman2K_ joined01:01
tomoj I am consoled that my workflow is not totally stupid, it appears hashrocket does it: http://reinh.com/blog/2009/03/02/a-git-workflow-for-agile-teams.html01:02
← pinpinlelapin left01:03
dfr|work → dfr|side_project01:03
→ sylvainfriquet joined01:03
tomoj also, lol @ interactive rebasing "will allow us to pretend that we wrote the tests first"01:04
fynn heh01:06
→ melvinram joined01:06
→ spaceonline joined01:07
← brizly1 left01:07
fynn tomoj: regarding "I keep my master clean, so it doesn't get rebased" - does that mean you always merge-pull into it from origin/master?01:09
tomoj yeah01:09
→ ProperNoun joined01:10
tomoj though the workflow in that article has a problem I believe we talked about: if someone commits to origin after you rebased foo onto master but before you managed to fast-forward master to foo and push, you will have to rerebase01:10
fynn: of course the merge-pull is always a fast-forward01:11
...unless someone commits AFTER I fast-forward master to foo but before I push :(01:12
context maybe. if hulu will stop acting retarded01:12
← ethercrow left01:12
fynn tomoj: right, so you sacrifice origin/master's history, to keep master clean01:13
tomoj I am a one-man team though, so it's ok :)01:13
fynn well, right, that simplifies things quite a bit :)01:13
in a full team, your origin/master history would be a) non-linear, b) littered with merge commits.01:13
tomoj unless I don't pull and instead fetch so I can see if I need to rerebase01:15
which ends up way more complicated than your idea01:15
context i dont use topic branches enough :x even on code only i touch :/01:15
← willb1 left01:15
fynn tomoj: "unless someone commits AFTER I fast-forward master to foo but before I push" <- how does that change things?01:15
tomoj well, really it doesn't I guess, since if that happens my push will just fail01:16
fynn I mean, if you're the only one pushing/pulling to origin/master, all your pushes would be fast-forwards01:16
→ willb1 joined01:16
fynn if you you're not, what happens here is that you need to pull again01:16
yeah, exactly.01:16
tomoj and if the push fails then I know I need to go back and rerebase01:16
fynn you pull again, which is a FF, then you push your commits.01:16
hm, I thought you never rebased?01:16
in that workflow, anyway.01:17
→ bcardarella joined01:17
→ wsc joined01:17
← Roman2K left01:17
tomoj here's the problem. I do some work in foo and decide it's time to push it. so I rebase foo onto origin/master, switch to master and fast-forward to foo. if someone commits right now, before I push, then my push of master will fail01:18
and pulling master wouldn't be a fast-forward01:18
makes much more sense to push directly from foo I guess01:18
→ ethercrow joined01:18
fynn yup...01:18
← djgera left01:19
tomoj one of the steps in that hashrocket article is merging back to master before pushing, like I used to think. I wonder why?01:19
they never work in master anyway, so why is it even there?01:20
fynn btw, pulling master may be a FF01:21
that is, pulling origin/master into master01:21
tomoj but it's fast-forwarded to foo, and foo changes haven't been pushed yet..?01:21
fynn oh, hm, sec.01:21
right, it is not.01:22
there may even be conflicts.01:23
in fact it can't possibly be a FF.01:23
tomoj indeed01:23
but really I wouldn't have pulled, I would've just tried to push and failed01:23
then I'd have to do something like: fetch, rerebase foo onto origin/master, reset master to foo, try to push again01:24
fynn and then reset hard to just before the rebase point, and repeat the process?01:24
tomoj and by that time someone may have committed again :D01:24
← foobarfoo left01:27
← aziz left01:30
Pieter hmm, I'm trying to apply a reject file created with git apply --reject, but it complains01:30
any thoughts?01:31
http://pastie.org/51896401:31
→ Chillance joined01:32
tomoj Pieter: isn't there supposed to be a linebreak after the @@...@@?01:32
Pieter no, that's just metadata01:32
tomoj I have never seen a diff where there wasn't a newline after the range info01:34
but I haven't seen a lot of diffs I guess01:34
→ qhoxie joined01:35
tomoj and git's complaint seems consistent with the fact that it thinks that line isn't range info because there's no newline01:36
← rudi_s left01:39
← paltman94 left01:43
← Gitzilla left01:43
← willb1 left01:46
→ willb1 joined01:46
→ Gitzilla joined01:47
← chris2 left01:49
← ereslibre left01:51
Ilari Are't there supposed to be --- and +++ lines before the hunk headers?01:51
SamB tomoj: git always puts some text there that describes (to a human) the context in which the hunk occurrs01:52
tomoj SamB: ah, I see, thanks01:54
SamB or at least, that's what it's supposed to do ;-)01:55
sometimes it doesn't tell me much, but it's usually pretty helpful01:55
← bdowning left01:58
→ bdowning joined02:00
← ProperNoun left02:01
Pieter Ilari: yeah, that's what I was thinking02:02
Ilari: but it's a bit silly I can't reapply the reject stuff, right? :)02:02
→ murph_ joined02:04
← fynn left02:04
→ ProperNoun joined02:05
← DrAk0 left02:05
→ ericindc_ joined02:06
← spearce left02:08
← fujin left02:10
← Chillance left02:12
← marvil07 left02:13
← murph left02:19
→ cytrinox_ joined02:20
← cytrinox left02:20
→ alanhaggai joined02:23
→ Jinzhu joined02:25
← willb1 left02:27
→ willb1 joined02:27
← aspotashev|eeepc left02:32
→ drizzd_ joined02:33
← whaley left02:35
→ GuyFromHell joined02:35
← GuyFromHell left02:35
→ whaley joined02:35
→ hax0r1337 joined02:38
← voker57_ left02:38
→ voker57_ joined02:39
→ boto_ joined02:41
← bcardarella left02:45
← _Vi left02:46
→ kwatz_ joined02:52
→ fujin joined02:52
→ tjafk2 joined02:53
← wsc left02:54
← drizzd left02:54
← boto left02:55
← altrux left03:00
→ scientes joined03:02
← klbate left03:03
→ bobesponja joined03:04
→ djgera joined03:05
← eddyp left03:05
→ altrux joined03:08
← tjafk1 left03:08
← whaley left03:11
→ _Vi joined03:13
→ hydrogen joined03:14
← bugfux left03:16
← JasonWoof left03:18
→ bcardarella joined03:20
dfr|side_project → dfr|work03:20
→ hyperair joined03:22
← Jinzhu left03:22
← altrux left03:25
← k0001 left03:25
→ k0001 joined03:27
← EndlesslyCurious left03:28
→ wsc joined03:30
← wsc left03:31
→ wsc joined03:31
← makeworkeasy left03:34
→ Jinzhu joined03:34
← alezandro left03:35
→ jasber_ joined03:35
→ candybar_ joined03:37
← jasber left03:37
← candybar left03:37
candybar_ → simplexity03:37
← jasber_ left03:38
→ madewokherd joined03:42
→ hostels joined03:43
← teolicy left03:44
← pulpfiction left03:45
← _Vi left03:46
→ _Vi joined03:46
→ yhager_ joined03:51
← hydrogen left03:51
simplexity → candybar03:57
→ ben_h joined03:59
← djgera left04:00
← fujin left04:00
← demio left04:01
→ eyevee99 joined04:02
cpg → cpg|biab04:04
← jfkw left04:06
→ kennethlove joined04:07
← bcardarella left04:09
← eyevee99 left04:10
→ eyevee99 joined04:11
← qhoxie left04:11
← Roman2K_ left04:11
→ Roman2K joined04:12
← eyevee99 left04:12
← hax0r1337 left04:14
← voker57_ left04:15
→ voker57_ joined04:15
← bobesponja left04:20
→ pulpfiction joined04:20
→ fujin joined04:24
← willb1 left04:27
yhager_ → yhager04:28
← fujin left04:29
← voker57_ left04:41
→ voker57_ joined04:42
← ericindc_ left04:43
← melvinram left04:43
→ willb1 joined04:44
→ bugfux joined04:45
← leachim6 left04:45
→ ericindc_ joined04:46
→ GothAlice joined04:46
GothAlice I'd just like to say damn, thank you for Git. It's awesome, and importing form svn is very smooth.04:46
← Jinzhu left04:47
← kwatz_ left04:50
← sakyamuni left04:52
fr → fr|04:55
→ simplechat joined04:55
← hostels left04:56
simplechat hi, i've just added a few things to my info/exclude. is there any way of removing them from the tree automatically?04:56
→ toxx__ joined04:57
→ Predominant joined04:58
→ bcardarella joined05:00
cloudhead I have a branch I'd like to merge into master, but I only want to merge certain commits, is this possible?05:02
cehteh cherry-pick05:04
rebase --interactive to some extent too05:04
→ hax0r1337 joined05:08
→ qhoxie joined05:09
cloudhead cehteh, thanks, I'll look into that05:10
→ Kirill_Krasnov joined05:10
← namelessnotion left05:10
← [KK]Kirill left05:11
Kirill_Krasnov → [KK]Kirill05:11
← ilteris__ left05:11
← voker57_ left05:13
→ voker57_ joined05:13
← toxx_ left05:14
→ pulpfiction- joined05:16
← hax0r1337 left05:16
→ ianmcorvidae joined05:22
← pulpfiction left05:23
← ianmcorvidae left05:26
→ ianmcorvidae joined05:26
→ fujin joined05:29
← GothAlice left05:31
← qhoxie left05:33
→ sakyamuni joined05:36
→ alanhaggai_ joined05:36
← kennethlove left05:37
← wshimmy left05:38
→ Kirill_Krasnov joined05:39
simplechat hey, how do i get git exclude to actually exclude things from the git tree?05:44
Pieter 'git exclude'?05:44
← ben_h left05:44
→ ben_h joined05:48
← ben_h left05:49
→ ben_h_ joined05:49
→ g3d joined05:50
← rpag left05:50
→ rpag_ joined05:50
→ absentia joined05:51
absentia hi.. anyone awake ?05:51
g3d yes05:51
← alanhaggai left05:52
absentia I'm doing a git push.. and my object 140 is like 30mb... is there any way to find out what file that is? or get a list of files that will be pushed?05:52
→ ben_h joined05:52
absentia git -v push git --verbose push git push -v git push --verbose05:52
don't seem to do anything... 'cept give usage.05:52
← sunoano left05:53
simplechat Pieter, git: 'exclude' is not a git-command. See 'git --help'.05:54
→ sunoano joined05:54
← absentia left05:54
← [KK]Kirill left05:56
← ben_h left05:56
→ ben_h joined05:56
→ ben_h__ joined06:03
← ben_h__ left06:03
← torbenh4 left06:04
→ eletuchy joined06:04
→ torbenh3 joined06:05
→ ben_h__ joined06:05
← ben_h__ left06:05
← ben_h_ left06:10
← abns left06:11
→ ijuma joined06:12
← ijuma left06:12
→ ijuma joined06:13
← ericindc_ left06:13
← bcardarella left06:14
→ qhoxie joined06:14
→ ben_h_ joined06:16
→ fujin_ joined06:18
← fujin_ left06:18
→ Shadou joined06:19
← ben_h left06:21
ben_h_ → ben_h06:21
hyperair how do i push all branches except one?06:25
i'm maintaining a local stacked git branch, and i don't want to push that one since it doesn't look like anyone else can use it easily06:25
→ hax0r1337 joined06:27
← qhoxie left06:28
→ mithro joined06:30
→ stuffcor1se joined06:31
← stuffcorpse left06:31
→ ask joined06:32
→ DrAk0 joined06:32
ask → \ask06:33
← bugfux left06:33
← fujin left06:34
→ bugfux joined06:34
→ qhoxie joined06:39
→ blaenk joined06:46
blaenk sorry for the stupid question, but is it possible to fetch a specific folder/files from a repository? or is it necessary to fetch the entire thing06:47
er, badly worded question06:47
basically there's only one folder (two files in it) from a repository that I want to use, but I dont want to simply copy the files because then I would lose the ability to update the source files should they change, is this possible?06:48
for example in svn I would do svn co myrepos/trunk/the/specific/folder/, I guess06:49
voker57_ you can't06:54
cehteh blaenk: no06:54
blaenk ah okay, thanks guys :)06:55
cehteh you can do a shallow clone06:55
prolly even a shallow fetch of a single branch06:55
but no single dir yet06:55
blaenk okay, thanks I appreciate the response06:55
← blaenk left06:55
← cloudhead left06:57
→ sea-gull joined07:05
→ voker57__ joined07:09
← cpg|biab left07:11
← Shadou left07:15
→ fujin joined07:18
→ cilly joined07:19
← voker57_ left07:24
← flazz left07:32
→ A6T joined07:33
← hax0r1337 left07:34
→ Jinzhu joined07:36
→ WinterWeaver joined07:37
← cilly left07:39
→ bizhat joined07:41
mina86|aw → mina8607:43
→ Octalot joined07:45
→ khmarbaise joined07:46
← alanhaggai_ left07:47
→ brizly joined07:47
← thommey left07:48
→ tatsuyao joined07:53
→ hax0r1337 joined07:55
→ sunoano1 joined07:56
← sunoano1 left07:56
→ sunoano1 joined07:59
Pieter other = xcalloc(sizeof(*other), num_head);08:01
isn't that the wrong way round? :P08:01
paakku yeah but pointers returned by calloc are suitable aligned for any type anyway08:03
→ ilteris__ joined08:03
ben_h hi all - what's the most efficient way to return the commit id of the current head?08:08
efficient as in processing time - i'd like to include it in my prompt08:08
← sunoano left08:08
→ CoderX joined08:11
← kenboo left08:11
paakku ben_h: git rev-parse HEAD seems to take 0.02s here. But if you want the branch name as well then look in git/contrib/completion/git-completion.bash.08:11
ben_h paakku: nice, thanks - i'll check it out08:12
→ memiux joined08:12
← willb1 left08:12
ben_h git rev-parse --short HEAD - perfect08:13
cheers.08:13
← qhoxie left08:15
← eletuchy left08:15
→ eletuchy joined08:16
→ Sigma joined08:18
→ cilly joined08:18
← boto_ left08:18
← tatsuyao left08:19
← hax0r1337 left08:21
→ eddyp joined08:21
→ ph^ joined08:28
→ willb1 joined08:29
→ thommey joined08:30
→ murph__ joined08:34
→ ppawel joined08:34
→ apeiros_ joined08:35
← cilly left08:38
→ elmob joined08:44
→ kumbayo joined08:44
← bugfux left08:45
→ sitaram joined08:47
← murph_ left08:50
← _Vi left08:51
→ Grahack joined08:53
← brizly left08:53
→ cilly joined08:55
→ Tuomas joined08:55
→ rudi_s joined08:56
→ Cromulent joined09:01
→ yann joined09:04
→ robinr joined09:04
← ph^ left09:09
→ ph^ joined09:10
→ chris2 joined09:10
← memiux left09:12
← A6T left09:13
← xenoterracide left09:15
→ Moult joined09:16
→ xenoterracide joined09:17
← Moult_ left09:17
→ A6T joined09:22
→ asksol joined09:22
→ akitada joined09:23
→ lbt joined09:25
→ xflw joined09:26
→ _Vi joined09:26
← DrAk0 left09:27
← xflw left09:27
→ DrAk0 joined09:29
→ resmo joined09:30
← hyperair left09:33
→ patrikf joined09:33
← julm left09:34
→ julm joined09:35
← sunoano1 left09:35
→ sunoano joined09:38
→ ferdy joined09:38
→ kn1tt1 joined09:40
← DrAk0 left09:41
← xenoterracide left09:41
→ xenoterracide joined09:43
← cilly left09:46
→ corni joined09:47
→ shri420 joined09:47
shri420 i am trying to create a new git repository and unable to do so successfully as i get the error : fatal: The remote end hung up unexpectedly09:48
← jefferai left09:48
shri420 can anyone help me to create a simple git repo on ubuntu09:48
→ jefferai joined09:48
paakku what command did you run to get this error?09:48
shri420 git clone09:48
paakku did you want to create a new repository from scratch, or clone an existing one?09:49
shri420 i want to create a new repo from the scracth and want to test it by doing git clone09:49
← akitada left09:50
paakku IIRC git clone is unable to clone empty repositories. Try putting some commits in there first.09:50
drizzd_ paakku: this has been fixed in recent versions09:50
since 1.6.2, to be exact09:52
paakku shri420: which version do you have then?09:53
drizzd_ even so, earlier versions would report "cannot clone empty repository"09:53
shri420 i taken a tar ball and made it a git repo by using git - 1.6.109:55
i have git 1.6.109:55
what is the problem?09:55
i followed the git user manual i have a tar ball i want to create a git tree and provide access to my collegues09:56
can you guys help me09:56
← jast left09:56
→ jast joined09:56
→ hyperair joined09:56
lbt shri420: which docs are you using?09:58
shri420 http://www.kernel.org/pub/software/scm/git/docs/user-manual.html#setting-up-a-public-repository09:59
i am sure the docs are correct. but i am missing something10:00
stupid10:00
lbt usually :)10:00
is git-daemon running10:01
and what user is it running as10:01
shri420 how do i find out. i run it through xinetd10:01
lbt OK, what distro10:01
shri420 ubuntu10:02
lbt first, ps -ef | grep git-daemon10:02
shri420 yes it is running as root10:02
lbt sec10:03
paakku under "Setting up a public repository", why would it matter whether git-daemon is running? The command is "git clone --bare ~/proj proj.git" so it should just read the files directly.10:03
shri420 no you need to export your git repo10:04
so that user can access it10:04
drizzd_ paakku: public presumably means over the network10:04
paakku: proj.git is supposed to be the public repo in this context10:05
paakku I guess I'm confused about which step of the instructions given in the manual is failing.10:06
drizzd_ shri420: using the wrong url will also give you that error message10:07
shri420 yeah i thought so, but i dont think i use wrong url10:08
i will try all over again10:08
lbt back now shri42010:08
→ klbate joined10:08
lbt OK, I assumed the 1st 2 commands went OK. So I assume git-daemon is running as per distro10:09
so in debian it's looking in -base-path=/var/cache10:09
by default10:09
I guess ubuntu is the same10:09
→ Sigma[Mtp] joined10:09
shri420 base path could be anything10:10
lbt yes10:10
http://www.kernel.org/pub/software/scm/git/docs/git-daemon.html10:10
shri420 but you have to tell in git dameon file in xinetd10:10
lbt yes10:11
shri420 that is paramters to git dameon10:11
Ilari shri420: It is specified in command line. If no --base-path is present, the default is '/'.10:11
shri420 yes i see that10:11
→ aziz joined10:11
← IRSeekBot left10:11
lbt so where is your git repo on the filesystem?10:12
have you specified a base-path (it should be on the ps -ef line)10:13
Ilari shri420: Also, anything in system logs?10:13
→ IRSeekBot joined10:13
← julm left10:14
→ d0k joined10:17
← asksol left10:17
→ plediii_ joined10:19
→ reprore joined10:20
→ Yuuhi joined10:20
← Sigma left10:21
shri420 i dont see anything10:24
← reprore left10:26
← up_the_irons left10:26
← plediii_ left10:26
→ reprore joined10:26
shri420 when i do a du -sh kernel.git i see it is only 104K10:31
i think it is a empty repository10:31
could that be a problem10:31
← reprore left10:34
→ ereslibre joined10:36
→ mbroeker joined10:39
← comp left10:42
← is_null left10:44
→ pipingcats joined10:48
← Cromulent left10:53
pipingcats Is there a way to "import" the history of a git repository into another one?10:53
→ moonflux joined10:54
pipingcats I first started using git to manage a project of mine about a year ago, and I ended up deleting the git repository to make a new one.10:54
I did it because I screwed something up that I could have fixed, and now I want to reconnect the two repositories if I can10:55
they're just two repositories of the same project, but with two different initial git commits10:55
← kn1tt1 left10:56
lbt pipingcats: do you care about the git commit ids?10:57
pipingcats over the gap between them, not really, I guess10:57
lbt no, afterwards?10:57
because the latest ids are created using the earlier ids10:58
pipingcats As long as the commits are still accessible, it doesn't matter what they're called, I guess10:58
Ahh.10:58
lbt and if you rebase the later repo on the earlier one then the current 'latest' will change10:58
this only matters if you share10:58
pipingcats yeah, I haven't shared anything yet10:58
I want to make it nice and tidy before I do, though10:59
→ parasti joined10:59
lbt in that case you probably don't care10:59
any branches ?10:59
pipingcats so I can use git rebase?10:59
yeah, a bunch11:00
but they all eventually end up in the same one11:00
lbt it depends what history you want to keep...11:00
I would start with the old repo11:00
(copy)11:00
← bizhat left11:00
lbt and add the new one as a remote11:00
and then pull the remote branches in11:01
pipingcats that's what I tried to do, but it told me they didn't have any commits in common11:01
lbt yes11:01
pipingcats I must have used the wrong command11:01
lbt you may have to cherry pick them in11:02
pipingcats Ahhh11:02
→ up_the_irons joined11:02
patrikf git clone new-repo; cd new-repo; git pull old-repo master:old; git checkout old; git rebase --root --onto master11:02
er11:03
git clone new-repo; cd new-repo; git pull old-repo master:old; git rebase --root --onto old11:03
pipingcats thanks patrikf, I'll try that11:03
lbt does that cherry pick automatically patrikf11:03
git rebase --root --onto old11:03
patrikf theoretically that should do the trick. untested though11:04
→ wshimmy joined11:07
→ ndim joined11:14
pipingcats Hey, the git rebase --root --onto old command isn't working11:15
→ ashH joined11:15
pipingcats everything else seems to have worked, though11:16
← ashH left11:19
→ aspotashev|eeepc joined11:21
→ DiegoMax joined11:21
→ ftf joined11:21
DiegoMax hello there, can somebody help me a little to set up some external repositories on my own repo ?11:21
← shri420 left11:23
← pulpfiction- left11:24
→ pulpfiction joined11:24
→ montylounge joined11:24
→ Sho_ joined11:25
voker57__ DiegoMax: man git-remote11:25
Gitbot DiegoMax: the git-remote manpage can be found at http://git.or.cz/man/git-remote11:25
DiegoMax thanks11:26
→ DaveIngram joined11:27
→ RaceCondition joined11:27
RaceCondition I have a branch in which I have a few commits that should really be in master, even before I merge the branch into master. Is my only option to get them out of the branch and to master to use git-rebase?11:28
← elmob left11:29
→ ashH joined11:30
mbroeker RaceCondition, how many commits do you want to apply to master?11:31
RaceCondition mbroeker, not many... about 3 or 4 of them out of 6 or 7 that are in the branch11:31
mbroeker git format-patch HEAD~8; ls; rm -f (alltheunwantedcommits) and git checkout master && for i in *.patch; do git am $i; done;11:32
RaceCondition whoa11:32
mbroeker, this way they'll still stay in the branch as well, right?11:33
← jksz left11:34
→ plediii_ joined11:34
mbroeker RaceCondition, sure.11:34
RaceCondition mbroeker, so later then I rebase the branch onto master and do a fast-forward merge in master, won't I get conflicts?11:35
→ kn1tt1 joined11:35
RaceCondition if I have the same commits in both branches11:35
mbroeker RaceCondition, it will most likely result in small conficts...11:35
RaceCondition mbroeker, still, what about rebasing and editing commit order?11:36
I've never done it so I'm a bit afraid, but I could try11:36
mbroeker just create a testing branch and apply your three commits and then merge the entire branch into testing. i guess that you won't get any conflicts which cannot be autoresolved...11:37
→ brizly joined11:37
← DaveIngram left11:38
DiegoMax im not sure if i understand this correctly, and sorry but im new to GIT11:38
RaceCondition mbroeker, OK, thanks11:38
DiegoMax i mantain a project that uses FFMPEG as a dependency, is a "remote" what i have to set up just to keep ffmpeg's code synced in my repo ?11:38
← Sigma[Mtp] left11:38
→ torbenh4 joined11:39
← torbenh3 left11:40
sitaram DiegoMax: you want submodules, and hopefully someone with more experience at *using* them will chip in11:41
DiegoMax okay thanks11:41
← montylounge left11:41
→ Juan joined11:43
→ bcarlyon|laptop joined11:44
→ robinsmidsrod1 joined11:45
← Juan left11:45
← BarryCarlyon left11:45
bcarlyon|laptop → BarryCarlyon11:45
→ ferai joined11:45
← ashH left11:46
→ mithraic_ joined11:46
→ ricky_ joined11:46
← ricky left11:47
ricky_ → ricky11:47
← aspotashev|eeepc left11:47
← jefferai left11:47
← corni left11:47
← xenoterracide left11:47
← ph^ left11:47
← sitaram left11:47
← apeiros_ left11:47
← murph__ left11:47
← thommey left11:47
← WinterWeaver left11:47
← voker57__ left11:47
← tjafk2 left11:47
← spaceonline left11:47
← __iron left11:47
← qrush left11:47
← Predominant left11:47
← joevano left11:47
← warthog9 left11:47
← devendra_l left11:47
← FreakGuard left11:47
← jbwiv left11:47
← ericindc left11:47
← hatseflats left11:47
← dmlloyd left11:47
← dfr|work left11:47
← eb_ left11:47
← robinsmidsrod left11:47
← jmspeex left11:47
← shellsage_ left11:47
← coldboot left11:47
← tchan left11:47
← eldragon left11:47
← xckpd7 left11:47
← metze_away left11:47
← sylvainfriquet left11:47
→ xenoterracide joined11:47
→ voker57__ joined11:47
→ aspotashev|eeepc joined11:47
→ jefferai joined11:47
→ corni joined11:47
→ ph^ joined11:47
→ sitaram joined11:47
→ apeiros_ joined11:47
→ murph__ joined11:47
→ thommey joined11:47
→ WinterWeaver joined11:47
→ Predominant joined11:47
→ tjafk2 joined11:47
→ spaceonline joined11:47
→ __iron joined11:47
→ FreakGuard joined11:47
→ qrush joined11:47
→ shellsage_ joined11:47
→ joevano joined11:47
→ jbwiv joined11:47
→ warthog9 joined11:47
→ hatseflats joined11:47
→ devendra_l joined11:47
→ ericindc joined11:47
→ coldboot joined11:47
→ dmlloyd joined11:47
→ dfr|work joined11:47
→ eb_ joined11:47
→ jmspeex joined11:47
→ tchan joined11:47
→ eldragon joined11:47
→ xckpd7 joined11:47
→ cods joined11:47
→ pastie joined11:47
→ Fissure joined11:47
→ jester2 joined11:47
→ cbreak joined11:47
→ osfameron joined11:47
→ coopsh joined11:47
→ Jippi joined11:47
→ metze_away joined11:47
→ spb joined11:47
→ mitsuhiko joined11:47
→ thijso joined11:47
→ shana joined11:47
→ Grum joined11:47
→ mpr joined11:48
← jefferai left11:48
← kn1tt1 left11:49
← Roman2K left11:50
→ Roman2K joined11:51
→ knittl joined11:52
pipingcats Is there an "alias" for the very first commit in a repository?11:52
Kind of like the reverse of HEAD11:52
← stuffcor1se left11:52
→ stuffcorpse joined11:53
sitaram pipingcats: no, because there need not be only one11:53
← sakyamuni left11:53
sitaram I can fetch a completely unrelated branch from some repo and stick it in my repo, no you have two "root" commits11:53
pipingcats ahh11:54
→ Roman2K_ joined11:54
pipingcats I've still kind of been using git as a glorified cookie trail11:54
← Roman2K_ left11:54
pipingcats because I'm the only programmer11:54
→ Roman2K_ joined11:55
→ shri420 joined11:55
shri420 guys, still i am unable to setup a git repo on my machine11:55
please help me11:55
pipingcats What OS are you running, shri420?11:56
shri420 ubuntu11:56
→ Superbaloo joined11:56
pipingcats http://book.git-scm.com/4_setting_up_a_public_repository.html11:56
shri420 i follow that but i am unable to setup11:56
sitaram pipingcats: there are ways to get that, but nothing elegant11:57
shri420 since i get fatal: The remote end hung up unexpectedly11:57
pipingcats Right now I'm still trying to figure out how to connect the two histories of these repositories together11:57
sitaram pipingcats: one way is "git log --format="%p-%h"|grep ^-11:57
→ drdave joined11:57
pipingcats shri420: did you run git-daemon?11:57
← _Vi left11:58
pipingcats go to where you put the repository and run git-daemon11:58
shri420 yes it is running11:58
sitaram shri420: I suspect something more basic; I remember you asking this in the morning, though I got disconnected later11:58
shri420 pipingcats: i run git daemon via xinted11:58
xinetd11:58
pipingcats how are you trying to fetch from the repository?11:59
← aspotashev|eeepc left11:59
shri420 using git clone11:59
sitaram and the URL?12:00
← ftf left12:00
sitaram is it an ssh: URL or a git: URL12:00
?12:00
shri420 git clone git://<IPAddress>/kernel.git12:00
→ FunkeeMonk joined12:00
pipingcats it should look something like "git clone git@myrepository:~/project.git/" if you put the project folder in the home folder of a git account you made12:00
shri420 pipingcats: i put inside /pub/scm directory and have given this directory as argument to git-daemon12:01
sitaram shri420: as pipingcats says, try putting in the full path to your repo in there, but I suspect xinetd is the problem -- you'll have to check the logs to see if it's doing it's stuff properly12:01
there are ways to debug xinetd issues (though I don't know them well enough)12:02
shri420 pipingcats: what is git@myrepository?12:03
username?12:03
sitaram shri420: can you paste your xinetd.conf and a sample session with a few different URLs?12:03
pipingcats git@<ipaddress>:/pub/scm/project.git12:03
sitaram shri420: that syntax is only for ssh access not plain git access12:03
pipingcats: he wants to run git daemon via xinetd which means public (unauthenticated) access12:04
pipingcats Ahhhhh12:04
I've never used xinetd12:04
shri420 full pathname works12:04
sitaram no userid involved12:04
me neither12:04
shri420 if i use full pathname it is working12:04
thank you12:04
sitaram shri420: works? meaning you can clone?12:04
pipingcats np12:04
shri420 yes12:04
pipingcats then you can do "git remote add myrepoalias git@<ipaddress>:/pub/scm/project.git"12:05
sitaram shri420: then you need to figure out how to opitimise it, or leave it as is -- after all anyone can add a remote12:05
pipingcats so you can just do "git fetch myrepoalias" later12:05
sitaram pipingcats: again that "git@" :-)12:05
pipingcats :))12:05
shri420 why do i need to add remote repo12:05
pipingcats it's just an alias12:05
sitaram shri420: if you did a clone, it automatically created a "remote" called originn12:05
origin*12:05
pipingcats so you don't have to type the whole thing12:05
oh, or that12:05
you can just use origin, like sitaram says12:06
shri420 how do i see that - using git branch -a?12:06
sitaram shri420: a "remote" in git is just a convenient, one-word, way to denote some repo with a long URL -- nothing more12:06
git reomte12:06
git remote12:06
shri420 okay12:06
sitaram then you can also say "git remote show origin" to see details of the remote called "origin"12:07
← ph^ left12:07
sitaram or any other remote that "git remote" shows you of course12:07
doener sitaram: well, it is a bit more. It can also provide the defaults for fetch and push12:07
sitaram: the default refspec for fetch being a major part12:08
shri420 now tell me how do i provide access using webserver git web12:08
→ aspotashev|eeepc joined12:08
drdave If I've got a file /folder commited in the repository, and i decide I want to remove it from the tracked files - adding it to .gitignore won't work (only works with uncommited files. Do I use git rm --cached ? Want to check before messing things up :)12:08
→ abw__ joined12:10
← abw_ left12:10
sitaram shri420: basic instructions for gitweb are available somewhere; most important is to put "gitweb.cgi" in a directory that your webserver recognises as having CGI scripts. Then you have to configure the script to look in the right places for the right things. That's al12:10
drdave: do you want it removed all the way *back* in history also?12:10
shri420 okay i will check that, thank you12:10
← Roman2K left12:11
drdave um.... not that bothered so yes.12:11
shri420 i am trying to design my git tree and i want to design in modular fashion, which one is good - git rebase or git merge12:11
sitaram drdave: the simpler case is handled by what you said "git rm --cached". The complex case (all the way back in history) needs filter-branch, and *will* mess up everyone who pulled from you or you pushed to12:11
shri420: both have their uses; there's no single right answer there (otherwise the other one wouldn't exist!!)12:11
shri420 sitaram: thats is the standard answer12:12
lol12:12
drdave k - I'll remember that. Even if I've just done it locally and not pushed?12:12
sitaram hey I'm a pretty standard guy :)12:12
→ UUncia joined12:12
shri420 how are usually kernel trees maintained they use git merge or rebase12:13
→ _Vi joined12:13
sitaram drdave: if you've just done it locally and not pushed, of course there's no problem12:13
→ knitt1 joined12:14
drdave So if I've not pushed is there a way to cleanly remoe it all the way back in the history? if so some pointers would be good to learn from?12:14
sitaram shri420: rebase until you are ready to push (while your topic branch is still under development) and merge otherwise of course12:14
drdave Coz if I push - I'll push my messed up commit12:14
sitaram drdave: man git-filter-branch12:14
Gitbot drdave: the git-filter-branch manpage can be found at http://git.or.cz/man/git-filter-branch12:14
drdave thanks12:14
sitaram drdave: you likely want "--index-filter 'git rm --cached --ignore-unmatch bad.file.path'"12:15
→ jceb joined12:15
sitaram drdave: test it out. Git makes it easy. Just make a copy of your current branch as a tag so you can always come back to it, and start with a clean working tree (no uncommitted changes)12:16
drdave Does anyone ever feel that playing with git - is like your messing with time travel :)12:16
→ DavidKlein joined12:16
sitaram drdave: I've lately started saying "a branch is like an alternate reality for your project" so yes :)12:16
sitaram has a son who write Harry Potter fan fiction, with time loops in it :-)12:16
sitaram writes*12:17
← Roman2K_ left12:17
shri420 sitaram: as i understand you say: we have master branch and topic1 topic2 topic3 branches which are need to be pushed into master. after development is it rebase or merge12:17
you say merge?12:17
sitaram shri420: rebase is always within one repo. Once you cross repos it better be merge hadn't it? Or I did not understand your question12:18
shri420: let me describe this in more detail12:18
shri420 you mean repo => which has several branches or one single branch12:18
sitaram shri420: you start topic1 off of master: git checkout master; git pull; git checkout -b topic112:18
drdave anyone know of any interesting projects that hook git in as a backend - I'm messing with using it for a distributed story telling tool - so its hooked into the script editor and also the text / media content. Any other similar projects - I heard about a wiki... couldn't find much.12:18
← pulpfiction left12:19
shri420 sitaram: go ahead i am listening12:19
sitaram shri420: now you make a few commits on topic1: git commit...12:19
shri420: meanwhile master on the "origin" has moved on, it got a few important things you need12:19
shri420: you have not yet pushed your topic1, so you can (and in many cases should) rebase: git checkout master; git pull; git checkout topic1; git rebase master"12:20
shri420: ok so far?12:20
shri420 yes12:20
→ Roman2K joined12:21
pipingcats Is there a way to automatically make a new local branch for each remote branch?12:21
← knittl left12:21
sitaram shri420: if your topic1 takes a long time to come out of development and be in a pushable state you will almost certainly rebase a lot12:21
pipingcats if I'm say, fetching 20 new branches from a remote12:21
← paakku left12:21
sitaram shri420: when you're ready to push, you... push! After that, that SHA (the top of topic1) is now sacred12:22
shri420: git push origin topic1:topic112:22
shri420 now what happens to master12:22
it has changes on top of the last commit to master before push or the changes are rebased to master (reverse order)12:23
sitaram shri420: I'm assuming you're not controlling master -- if you are, you can certainly merge topic1 in if it is deemed to be stable enough to do so12:23
shri420 so you say - if i am inside topicX branches, i rebase from master12:24
if i want to push the changes inside topic1 and not in master i merge it with master12:24
when i merge it creates only one commit (or patch) isnt it?12:25
sitaram shri420: so someone else (the PL) integrates your current topic1 into master. Then, when you bring down this "master" (git checkout master; git pull), and then try and rebase your topic1 onto it (git checkout topic1, git rebase master), git notices the commits that already made it to master and puts only any new ones in topic1 on top of the new master12:25
shri420 sitaram: does my PL do a merge or rebase12:26
(since i dont have a PL) ;-)12:27
sitaram shri420: depends. The bulk of the development on open source happens with patches emailed, which is effectively a rebase12:27
the point is, a rebase changes the SHA of a commit. If that is a problem (someone else already has that SHA) then use merge. Otherwise use rebase12:27
shri420 and what happens to topic1 tree. it is deleted after wards12:28
When you say SHA of a commit? which commit?12:28
sitaram shri420: whatever commit got rebased.12:28
shri420 all commits which git rebased?12:29
sitaram yes12:29
shri420 got rebased12:29
oh, i though sha is only for the commit (depending on its contents)12:29
wereHamster contents and all its parent commits12:30
shri420 and it changes only if contents changes12:30
okay12:30
sitaram SHA depends on (1) commit message (2) tree (3) parent commits (4) author/committer info12:30
when you rebase, parents are changing so SHA changes12:30
→ reprore joined12:32
← Roman2K left12:33
shri420 okay i will you my setup here - we have a master linux branch - This linux branch is common for 2.6.27 Linux for my pxa processor. now i created a branch out of master for board1, my company makes similar boards (board1, board2... of same processor) so most of the stuffs are similar. so i want to group similar stuffs in one or more branches12:34
now tommorrow one more board arrives i just need to do (rebase/merge, which i need to decide now)12:35
thats is why i am asking these questions12:35
sitaram: you got what i try to do here12:35
DiegoMax this is too complex12:35
sitaram reading...12:35
← pcapriotti left12:36
shri420 okay i will make simple - We have a linux master branch and lots of common stuffs (android, some other bug fixes for linux 2-6.27 and so on)12:37
← wshimmy left12:37
sitaram shri420: conventional wisdom is to keep common stuff in a common branch and so on, otherwise if a change applies to board 1/2/3 and no 4/5 you'll be repeating that change everywhere by cherry-pick at least12:38
→ wshimmy joined12:38
sitaram shri420: I can't look too deep into your situation, but I'd recommend a good read of man git-workflows12:38
DiegoMax is there any way to tell submodules to always update to the latest version ?12:38
(automagically)12:38
shri420 sitaram: the problem is from master what you do (rebase/merge)12:39
sitaram shri420: and the following two thumb rules: (1) fork at the earliest point it makes sense and (2) make a change in the oldest branch that it applies to and merge upward12:39
shri420 i want to make sure of this how we merge the changes in common branch into a new board branch12:39
← ben_h left12:39
shri420 you say if it is in early stages do a rebase otherwise do a merge?12:40
→ fridim_ joined12:41
→ shadowhand joined12:45
→ blkf joined12:48
→ ben_h joined12:49
→ ben_h_ joined12:49
→ tvw joined12:50
← ben_h left12:50
→ masak joined12:50
pipingcats How do you make a mailbox file?12:53
for use with the git am command12:53
Grahack Hi, how can I edit the initial commit? `git rebase <initial-commit-sha1>` lets me edit from the second commit only12:54
offby1 I think there's a special command-line option for that12:54
dunno what it is, though12:55
drizzd_ pipingcats: using a mail user agent12:55
wereHamster Grahack: git rebase --root12:55
← sitaram left12:55
→ Bass10 joined12:55
Grahack thansk wereHamster12:56
→ kukks joined12:59
→ bx2 joined12:59
pipingcats drizzd_: What is that?13:00
Grahack wereHamster: thanks but it doesn't work here, do I need a recent git ? mine is 1.5.6.313:00
← shri420 left13:00
→ shri420 joined13:01
drizzd_ Grahack: you need 1.6.213:01
← shri420 left13:02
drizzd_ Grahack: with earlier versions, you need to checkout the root commit, amend it, and from your old root --onto the new root.13:02
→ jargan joined13:04
← jast left13:04
offby1 pipingcats: "mail user agent" is a stuffy, formal name for what normal people call 'an email client'.13:07
RaceCondition mbroeker, I think it screwed it up - I did as you told and applied the patch files to master, then created another branch from master, but I merged the original branch into the new branch assuming only the patches that arent in the new branch are pulled13:07
offby1 pipingcats: a typical example would be "mutt"13:07
RaceCondition but they were all pulled, and now I have duplicate patches in the history of the new branch13:07
and I deleted the original branch before I discovered this13:07
offby1 challenges RaceCondition to a bug-duel!13:07
mbroeker never ever test on master, RaceCondition :)13:07
offby1 maybe we can get Mrs Fencepost over there to join in13:08
RaceCondition should I just create yet another branch and cherry pick only non-duplicate branches from the fucked up branch?13:08
pipingcats Ahh13:08
RaceCondition mbroeker, I did not test on master13:08
bremner, I only applied the patches to master, then created a new branch for testing13:08
bremner, sorry, I meant mbroeker13:08
drizzd_ Grahack: http://colabti.org/irclogger/irclogger_log/git?date=2009-05-27#l176513:08
Ilari And mr. integer overflow, mr. stack overflow and mr. buffer overflow? :-)13:08
RaceCondition I think cherry-pick will help probably, or do a format-patch again and just appliy those 2 that I need to a new branch13:10
← CoderX left13:10
offby1 Ilari: they're too busy13:11
→ saurabhverma joined13:14
→ montylounge joined13:14
makies_bot → makies13:17
RaceCondition can I have git tell me where I've got whitespace errors?13:24
I'm getting whitespace warnings all the time when committing, but I never find them13:25
ferai → jefferai13:25
rudi_s RaceCondition: git diff should mark them in red if you use [color] ui = auto in your .gitconfig.13:25
RaceCondition oh, I see13:25
rudi_s, what about existing commits?13:26
→ Roman2K joined13:26
→ ftf joined13:26
RaceCondition I could just do a diff on the all commits starting from initial commit13:26
but that's a killer :P13:26
rudi_s RaceCondition: You could use your editor to find problematic ones now and fix them. Then there is no need to inspect old commits.13:27
→ jksz joined13:27
← BarryCarlyon left13:28
RaceCondition yeah13:28
← reprore left13:28
RaceCondition can I tell git to ignore trailng whitespace on lines that just contain whitespace and nothing more?13:28
I'm used to indenting all lines, even empty ones13:28
it makes sense to me to indent them13:28
because they're still part of the code block, even though they're empty13:29
offby1 I assume the warnings are coming from a hook, and that you can edit the hook to do whatever you want.13:30
DiegoMax well... i give up :)13:30
→ ben_h joined13:31
DiegoMax what could be the reason why im pushing the files to github, it says "all ok", but cant see the updates on the remote server ?13:31
any hint ?13:31
offby1 maybe you're not pushing the branch you want to push?13:31
DiegoMax i have only one branch13:32
→ reprore joined13:32
← aspotashev|eeepc left13:32
→ julm joined13:33
← Kirill_Krasnov left13:34
RaceCondition it's not possible to replace the authors/commiters e-mail address on already existing commits, right?13:34
without rebuilding the whole history13:34
← knitt1 left13:34
→ thiago joined13:34
rudi_s RaceCondition: No. That would change the sha113:34
RaceCondition yeah, I know13:35
I've nothing against that actually :P13:35
← bx2 left13:36
rudi_s RaceCondition: Then you could use git rebase -i or filter-branch (but I have no idea how filter works, never used it).13:36
If you've published your repository then you shouldn't do this.13:37
RaceCondition yeah, well screw it :P too much effort13:37
I've not13:37
I just thought correcting old author e-mail aadresses would be a nice thing to do13:37
patrikf hmm. what's the easiest way to get a list of all paths tracked by git?13:37
I know I can pipe ls-tree to cut, but maybe there's a simpler variant13:38
offby1 git ls-files13:38
patrikf ah :) thanks13:38
← qrush left13:39
← jceb left13:41
← julm left13:42
→ julm joined13:42
→ Sigma joined13:44
→ abw_ joined13:49
← abw__ left13:49
← ben_h_ left13:50
→ whaley joined13:51
Grahack wereHamster told me about "git rebase --root" but how can I set the --onto option (mandatory) to just edit the initial commit ?13:52
← apeiros_ left13:57
→ apeiros_ joined13:57
drizzd_ Grahack: looks like you'll have to use the method I proposed instead.13:58
← Scurz left14:00
→ bremner_ joined14:01
→ remy joined14:03
Grahack thanks drizzd_ I give this a try14:04
← ProperNoun left14:08
→ ProperNoun joined14:08
← reprore left14:09
→ reprore_ joined14:09
← brizly left14:11
→ JasonWoof joined14:13
→ memiux joined14:16
← apeiros_ left14:17
→ jmou joined14:19
→ tatsuyao joined14:19
← julm left14:20
→ marvil07 joined14:20
→ shri420 joined14:22
→ sitaram joined14:23
→ aspotashev|eeepc joined14:24
← shri420 left14:26
← corni left14:27
→ kenboo joined14:27
→ knitt1 joined14:31
← UUncia left14:32
→ ph^ joined14:34
→ CoderX joined14:35
→ lucideer joined14:38
→ bobmcw joined14:38
→ tatsuyao_ joined14:42
← tatsuyao left14:42
← simplechat left14:44
← Jinzhu left14:46
← aspotashev|eeepc left14:47
← thiago left14:49
← tvw left14:51
← tatsuyao_ left14:52
← ivanoats left14:52
← reprore_ left14:53
→ reprore_ joined14:53
→ simNIX joined14:54
← sea-gull left14:59
← chris2 left14:59
← bremner_ left14:59
→ tatsuyao joined14:59
makies → makies_bot15:00
→ thiago joined15:02
→ apeiron joined15:02
← kpreid_ left15:02
→ kpreid_ joined15:03
drdave → Guest4441015:03
→ bcardarella joined15:06
→ apeiron_ joined15:08
← apeiron left15:10
→ ntoll joined15:12
→ _wwwald joined15:12
← wwwald left15:12
→ cilly joined15:13
← mbroeker left15:15
DiegoMax well i knew I couldnt be that stupid15:17
→ name joined15:17
DiegoMax the problem was github not working15:18
← abw_ left15:18
DiegoMax geez what a waste of time15:18
i signed up for a gitorious account and everything just works as supposed15:18
→ SlipperyChicken joined15:19
← remy left15:20
← kenboo left15:21
← ph^ left15:22
← lucideer left15:22
→ voker57 joined15:23
← ia left15:23
→ JNRowe joined15:24
offby1 surprising; github seems reliable to me15:24
DiegoMax they have been with issues today15:25
i just saw their status page15:25
→ ia joined15:25
→ bcarlyon|laptop joined15:25
DiegoMax also, i like gitorious more tbh15:25
→ mbroeker joined15:25
→ tango_ joined15:26
offby1 competition++15:27
thiago gitorious is getting sponsored to make improvements15:28
and it's open source15:28
AGPL15:28
← WinterWeaver left15:29
← Grahack left15:29
← blkf left15:30
← Guest44410 left15:31
← montylounge left15:31
→ asksol joined15:32
→ blkf joined15:32
→ montylounge joined15:32
← cilly left15:32
← k0001 left15:33
→ k0001 joined15:34
← voker57__ left15:38
← JNRowe left15:38
→ JNRowe joined15:38
← JNRowe left15:38
← saurabhverma left15:38
← reprore_ left15:39
→ JNRowe joined15:39
→ ph^ joined15:39
→ reprore_ joined15:39
→ voker57_ joined15:41
← ericindc left15:42
→ altrux joined15:42
← montylounge left15:42
← eletuchy left15:46
← JNRowe left15:47
→ JNRowe joined15:47
→ Cromulent joined15:47
← bobmcw left15:49
← voker57_ left15:49
→ voker57_ joined15:49
← sunoano left15:53
← bcardarella left15:54
← DiegoMax left15:54
→ ericindc joined15:55
← voker57 left15:55
→ dodo_the_last joined15:58
← CoderX left15:59
← sitaram left16:00
→ bobmcw joined16:00
← JNRowe left16:01
→ JNRowe joined16:01
← shadowhand left16:01
→ shadowhand joined16:02
→ ivanoats joined16:04
← ivanoats left16:06
jmou → julm16:06
← mbroeker left16:06
→ ivanoats joined16:07
→ bobmcw_ joined16:07
→ kabus joined16:08
← reprore_ left16:08
→ reprore_ joined16:08
← ivanoats left16:08
→ ethercrow_ joined16:09
→ ivanoats joined16:10
← bobmcw_ left16:10
← ethercrow left16:11
apeiron_ → apeiron16:11
← ivanoats left16:12
→ sunoano joined16:12
→ kabus_ joined16:13
→ ivanoats joined16:13
→ kumbayo_ joined16:14
→ radarek joined16:14
← ivanoats left16:15
← kabus_ left16:15
→ kabus_ joined16:15
← ftf left16:16
← kabus_ left16:16
→ kabus_ joined16:16
→ ftf joined16:17
→ ivanoats joined16:17
← doener left16:18
← kabus left16:18
→ doener joined16:18
← ivanoats left16:19
← pipingcats left16:20
← ftf left16:22
← bobmcw left16:24
→ dimsuz joined16:25
dimsuz hi guys! noob question: when I do a git push, it warns me like this: "you did not specify any refspecs to push and the current remote has not configured any push refspecs..."16:26
what does this mean?16:26
it doesn't happen when i do push on the other host16:26
why? and what to do?16:26
Fissure hm, is that related to push.default?16:29
← up_the_irons left16:29
Fissure are the hosts running the same version?16:29
← kumbayo left16:30
← moonflux left16:30
← fujin left16:35
dimsuz Fissure: not sure about same version, and yes, warning mentions push.default16:35
→ ciskje joined16:37
→ up_the_irons joined16:40
← FunkeeMonk left16:42
Fissure then set push.default to whatever you prefer and continue on your way16:43
→ CoderX joined16:44
← sgrimm left16:45
← up_the_irons left16:50
dimsuz like what? "master"?16:52
→ apeiros_ joined16:52
→ ftf joined16:52
→ qrush joined16:52
→ sea-gull joined16:53
→ WinterWeaver joined16:54
ftf → the_zeno16:54
← ciskje left16:55
← the_zeno left16:55
→ up_the_irons joined16:58
mina86 → mina86|aw16:59
→ dbaker1 joined17:05
dbaker1 when I'm commiting do I need to be in the root path of my project in the terminal?17:05
offby1 nope17:07
← WinterWeaver left17:07
offby1 this is the sort of question you can answer yourself, using the time-honored technique known as "Try It And See™".17:07
→ bizhat joined17:08
dbaker1 offby1: so the active branch is global on my system?17:09
→ corni joined17:09
dbaker1 if I have several projects only one of them can use git ?17:09
wshimmy dbaker1: ah. you need to be *under* your root path17:09
← dimsuz left17:09
offby1 dbaker1: gosh, I don't understand your question; but certainly you can have many git projects on a given box17:10
I think wshimmy read your mind better than I did.17:10
wshimmy dbaker1: if you have ~/projects/foo and ~/projects/bar, each as distinct git repos, you can commit to foo from ~/projects/foo/baz/buz just fine, but you can't commit to bar from ~/projects/foo17:10
← masak left17:10
dbaker1 wshimmy: ok17:10
wshimmy: can I have nested git projects?17:10
wshimmy: ~/projects/foo/bar and ./git inside foo and another ./git inside bar?17:11
wshimmy dbaker1: or you can confuse yourself by using GIT_DIR="~/projects/foo/.git" git commit or similar from outside of that dir17:11
dbaker1: not unless you are using submodules, which is slightly different than just "nested git projects"17:11
dbaker1 wshimmy: I'm not trying to confuse myself :)17:11
wshimmy: great. thanks17:11
← _Vi left17:13
→ pjan joined17:18
→ mbroeker joined17:19
← hyperair left17:20
→ hyperair joined17:20
→ tchan1 joined17:20
offby1 wishes he could wshimmy like his sister kate17:21
← tchan1 left17:22
→ mnob joined17:24
← reprore_ left17:25
→ tchan1 joined17:26
← mnob left17:26
→ willb2 joined17:26
→ mnoble joined17:27
← tchan1 left17:28
Pieter what's a wshimmy?17:29
← willb1 left17:32
offby1 wshimmy (noun): an IRC user, active on the #git channel. Synonyms: Enright17:32
← klbate left17:34
wshimmy That wins most bizarre highlight notification of the week!17:34
← mnoble left17:35
← tchan left17:35
→ reprore joined17:36
offby1 cashes in his prize and flies to Aruba17:36
Nugget heh17:37
→ ivanoats joined17:37
→ DaveIngram joined17:38
→ EndlesslyCurious joined17:40
→ ivan_ joined17:42
→ cpg joined17:44
← ivan_ left17:45
← ivanoats left17:48
→ zooko joined17:52
zooko Is it possible for two repos to have the same files and directories but different hashes, because the history of patches that created those files and directories differ?17:53
Or is the version identifier the hash only of the files and directories, excluding the history of how they got there?17:53
offby1 not only is it possible; it's likely17:53
the commit hash effectively includes the entire history17:54
patrikf zooko: if they have all files and paths identical, then the sha1 of the tree will match (which you can see with git cat-file -p HEAD)17:54
offby1 true true17:55
but the commit includes, among other things, the time of day17:55
zooko But the sha1 of the tree isn't what is normally used for an identifier?17:55
offby1 not too often.17:55
mina86|aw → mina8617:55
offby1 It's more common to refer to the sha1 of a commit.17:55
patrikf zooko: no. the commit is.17:55
zooko Does the sha1 of the tree include files that are in there but which aren't under git's supervision?17:55
patrikf no.17:56
zooko Is it often the case in practice that some minor change like an automatically generated timestamp or version number or "$CvsId$" type thing in a file causes the sha1 of the tree to be different?17:56
patrikf of course.17:56
offby1 well, _if_ you have such a change in your repository, then yes, the sha1 will differ.17:57
patrikf the SHA-1 will be completely different, and this is by design. it wouldn't be a good hash function otherwise...17:57
offby1 But I'm not sure that happens often -- git users tend to ensure their build process doesn't modify their source files.17:57
(At least, _I_ avoid that)17:57
zooko Thanks for the answers!17:57
← reprore left17:57
→ reprore joined17:57
mina86 I would imagine automatically generated timestamp or $Id$ is changed when file content's changes so it doesn't matter anyways17:59
ie. set a hook which updates file's last modification date when you save it in editor -- in such situation the date would change only if you change something else in the file as well18:00
offby1 I don't _think_ git has a mechanism to expand keywords like '$Id$', and even if it does, I suspect the more sophisticated git users avoid it.18:00
mina86 it has18:00
offby1 ya learn something new every day ...18:01
mina86 but it is somehow useless (IMO) as it works a bit differently then thing in CVS or SVN18:01
wereHamster yeah, the $id$ expands to the blob id and not the commit id18:01
mina86 what's worse it gets expanded only on checkout or something like that18:02
offby1 it's probably just as easy to have a build system insert timestamps and commit Ids and that kinda thing.18:02
that's what I do at work anyway.18:02
wereHamster offby1: that's what most projects stored in git do, see for example GIT-VERSION-GEN in the git repo18:03
zooko offby1: you have the build system insert timestamps or commit ids?18:07
So then if you ask for the sha1 of the tree state after that it will differ.18:07
offby1 wereHamster: yep. GIT-VERSION-GEN is way fancier than what we use.18:07
zooko: well, our build system doesn't modify the _source_ files; it instead modifies the build products.18:08
it's a bad idea in general for a build to modify source.18:08
zooko offby1: okay, so the sha1 of the files which are managed by git will be unchanged.18:08
offby1 correct18:08
which is, of course, a Good Thing™.18:08
← DaveIngram left18:10
→ cemerick joined18:10
zooko Thanks, folks! Here's the ticket I've been working on to implement some git-like features on top of darcs: http://allmydata.org/trac/darcsver/ticket/318:11
← thiago left18:13
→ demio joined18:13
← BeerSerc left18:14
→ BeerSerc joined18:14
← JasonWoof left18:16
→ pulpfiction joined18:19
← aziz left18:19
→ aziz joined18:21
→ dsch04 joined18:23
← shadowhand left18:23
← willb2 left18:24
→ Pistahh joined18:24
← demio left18:25
→ bobesponja joined18:25
← aziz left18:27
→ aziz joined18:27
← Cromulent left18:28
→ tatsuyao_ joined18:31
→ robomalo joined18:33
← reprore left18:34
← tatsuyao left18:36
← julm left18:37
→ julm joined18:38
← JNRowe left18:39
← tatsuyao_ left18:39
→ ericindc_ joined18:40
→ MisterN joined18:42
→ ivanoats joined18:43
→ bx2 joined18:44
← Octalot left18:44
→ \ask_ joined18:45
← CoderX left18:45
← dreiss left18:45
← cemerick left18:45
→ CoderX joined18:45
← ivanoats left18:46
← memiux left18:50
→ cloudhead joined18:50
← _wwwald left18:50
← ericindc_ left18:51
murph__ → murph18:52
← Suhail left18:52
← neoeinstein left18:52
→ _Vi joined18:54
→ dreiss joined18:57
→ sakyamuni joined19:00
→ comp joined19:01
← comp left19:01
→ comp joined19:01
→ Suhail joined19:02
→ rudi_s_ joined19:03
→ bugfux joined19:03
→ hpoydar joined19:04
→ chris2 joined19:04
→ hax0r1337 joined19:05
← __iron left19:07
← rudi_s left19:07
→ __iron joined19:09
→ moconnor joined19:15
← ben_h left19:16
→ JasonWoof joined19:19
→ tvw joined19:19
← mbroeker left19:25
→ yrashk joined19:26
→ memiux joined19:27
→ solydzajs joined19:28
← solydzajs left19:28
← zooko left19:28
→ cemerick joined19:30
→ zacharyc joined19:31
→ tcurdt joined19:33
tcurdt Hey folks, I am a little confused. I have a local-async-pd branch following a remote async-pd svn branch. I just did a 'git rebase master' on the local branch. Which did rebase the changes. Then did a 'git svn dcommit' ...but I would have expected it to also move the branch point. Hard to explain. Have a look here: http://vafer.org/pub/picture1.png ...understand what I mean?19:37
essentially I wanted to get my changes from master over to the async-pd branch19:38
so trunk and async-pd don't get too different19:38
← __iron left19:39
tomoj tcurdt: looks like you rebased changes that were already pushed out19:39
tcurdt yes ... to the svn branch19:40
argh ... I see your point19:40
→ owen1 joined19:40
tomoj I dunno anything about git-svn19:40
but in regular git, you never never rebase commits that have already been pushed19:40
tcurdt will remember that for sure! but ... hm ... wondering how I could fix this now19:41
seems like I should have used a merge19:42
→ __iron joined19:42
tomoj well, you could push local-async-pd to async-pd19:43
but that would be a force push19:43
→ buffi joined19:43
buffi is it possible to get the latest commit logs from a GIT repository without cloning it locally?19:44
tcurdt that's what I thought ... but no idea on how to do this with git-svn ... as usually only use git-svn dcommit19:44
tomoj tcurdt: oh, yeah, I dunno anything about git-svn, sorry19:44
buffi that is, doing the equivalent of "svn log svn://not.my.machine.com/repo"19:44
→ seanmccann joined19:44
tcurdt tomoj: nw ... thanks anyway ... still hoping to find someone here that knows though :)19:45
→ klbate joined19:46
← CoderX left19:46
buffi anyone? :)19:46
← kabus_ left19:48
g3d buffi: yes, just a sec19:48
buffi I've been reading man pages like crazy and searching for the answer, but I only located another guy with the same question on stackoverflow :)19:49
g3d buffi: see http://stackoverflow.com/questions/615816/git-how-to-get-a-snapshot-of-a-git-repository19:49
→ qhoxie joined19:50
WinterMute g3d, I think he wants logs without a local copy19:50
g3d buffi: ah heh, sorry, I totally misread your question19:50
buffi yeah, that is something completely different19:50
g3d yeah, sorry19:50
buffi no problem19:50
cehteh buffi: no such thing19:51
buffi :(19:51
g3d buffi: what do you mean by "without" cloning?19:51
cehteh there is git ls-remote to see the heads19:51
g3d buffi: as in "no access to the remote repository" ?19:51
buffi no, without checking out all files locally19:51
cehteh and if the remote has gitweb or such installed you can browse the log19:51
buffi well, if you have a SVN repo at a remote machine. you can do "svn log svn://somewhere.com/repo" to get the commit messages without having to check out all of the actual data19:52
cehteh yes19:52
buffi so no such thing in GIT?19:52
cehteh thats a difference .. git is distributed19:52
buffi so?19:52
cehteh you have to clone a repo, or have access to the remote machine19:52
buffi if it's possible to clone the repo, then it should be quite possible to fetch the logs from a repo without having to grab all of the actual data?19:53
or am I missing something?19:53
cehteh or rely on the services the remote provides, such as gitweb or so19:53
→ LCIDFire joined19:53
buffi oh well, thanks anyways :)19:54
← seanmccann left19:54
cehteh not really possible19:54
← bizhat left19:54
cehteh logs are constructed from commit objects19:54
LCIDFire folks - I'm stuck here with 2 files that I cannot commit and not revert - the console just gives me "error: no desc for state={_U}"19:55
cehteh iirc there is no 'clone commits only' .. that will leave a broken repository anyways19:55
LCIDFire: what OS/filesystem?19:55
LCIDFire Ubuntu 9.04/ext319:56
cehteh strange .. dunno what causes this error19:56
git status gives?19:56
→ RhysU joined19:56
LCIDFire btw: is there any decend gui yet - I'm not a friend of console19:56
cehteh git gui19:57
wereHamster LCIDFire: gitk19:57
RhysU Dumb question-- how do I remove the execute bit on a tracked file?19:57
LCIDFire I said decend ;)19:57
← cemerick left19:57
→ moconnor_ joined19:57
cehteh they are decend19:57
just configure your tk installation to look pretty19:57
wereHamster RhysU: chmod 664 $file; git add $file; git commit19:57
LCIDFire how would I get the status of a file in git gui19:57
RhysU Thank you19:57
← moconnor left19:57
cehteh it shows that19:58
LCIDFire where?19:58
cehteh prolly in 'unstaged'19:58
and then right on the text box it has a headline telling about the file you selet19:58
→ tatsuyao joined19:59
→ lobolife joined20:01
mina86 → mina86|aw20:01
→ jrmuizel joined20:02
→ jasber joined20:02
← _Vi left20:03
→ _Vi joined20:03
← scientes left20:03
→ WinterWeaver joined20:04
← ericindc left20:05
→ Octalot joined20:05
← marc_kdab left20:05
mina86|aw → mina8620:14
← tcurdt left20:16
← dsch04 left20:17
← eMBee left20:23
→ eMBee joined20:23
← eMBee left20:24
← wagle left20:24
→ toxx_ joined20:25
← jrmuizel left20:26
← DavidKlein left20:26
→ mattikus joined20:26
→ eMBee joined20:26
← sunoano left20:31
← wsc left20:31
→ boto joined20:34
← toxx_ left20:37
← RhysU left20:37
→ neoeinstein joined20:39
→ doki_pen joined20:44
← Roman2K left20:46
→ sunoano joined20:46
→ Grahack joined20:47
→ fynn joined20:48
→ pulpfiction- joined20:48
fynn OK, some simple questions:20:48
→ Juan_ joined20:48
fynn I'm on branch foo, how do I show ./bar.py on branch master?20:48
→ bcardarella joined20:48
fynn I.e. cat the contents of ./bar.py as they are on master's HEAD20:48
← buffi left20:49
← Tuomas left20:50
→ djgera joined20:50
→ pmurias joined20:51
fynn OK, looks like it's 'git show master:bar.py'20:51
← SlipperyChicken left20:52
← Grahack left20:52
← g3d left20:53
→ burto joined20:53
← memiux left20:53
→ sobersabre joined20:53
sobersabre hi.20:54
I want to allow others to commit into my git repository.20:54
what document shall I read ?20:54
← ppawel left20:55
bremner sobersabre: do you want to give them shell access? If not, look for gitosis20:55
Juan_ http://toolmantim.com/articles/setting_up_a_new_remote_git_repository20:55
→ DrAk0 joined20:55
LCIDFire sobersabre: I'd just go for github20:55
sobersabre bremner: I prefer NOT to, this is the last resort.20:55
bremner sobersabre: gitosis seems to work pretty well then, check it out.20:56
sobersabre hm.. lemme google up github... is it some kind of public service ?20:56
→ bobmcw joined20:56
Ilari sobersabre: And giving others commit access to your repo is bad idea. The usual way is via push access to bare (no working tree) repo.20:56
→ dcolish joined20:56
LCIDFire sobersabre: the best service I know so far20:56
sobersabre Ilari: I somehow avoided knowing all these conepts whilst using git for myself :)20:57
dcolish is there a way to aggregate rss for code changes via gitweb?20:57
sobersabre "bare", etc.20:57
→ fujin joined20:58
sobersabre I think I'm gonna rtfm :)20:58
thanks guys.20:58
dcolish: I don't know, but trac does this.20:59
← pulpfiction left20:59
Ilari sobersabre: Bare repo means repo without working tree.20:59
sobersabre I mean generates rss for "version control backend"20:59
← lobolife left20:59
← burto left20:59
sobersabre Ilari: you mean an empty versioned location ?20:59
Nugget weird, the torrent crashes rtorrent. working with vuze and ctorrent though.20:59
erp20:59
sobersabre and to merge from it into my master or other branch.. right ?20:59
day change! :)21:00
← LCIDFire left21:00
sobersabre Ilari: did I "dig" your idea correctly ?21:01
have I grokked it ?21:01
← klbate left21:02
→ klbate joined21:05
owen1 i try to clone the gitosis-admin with git clone git@<server-ip>:gitosis-admin.git and being asked for a password even though git account has no password. any clues?21:05
sobersabre owen1: does it work via ssh ?21:06
owen1 good question. let me see.21:06
sobersabre check if ~/.ssh/* are accessible...21:06
and make sure /etc/ssh/sshd_config ... doesn't block the user for some reason.21:07
like membership in a denied or no-membership in an allowed group.21:07
owen1 sobersabre: i can't ssh! it ask for pass.21:07
sobersabre well, check ... ssh configuration.21:07
owen1 sobersabre: le me see21:07
i did chmod 700 for my home folder recently...21:08
sobersabre this doesn't matter.21:08
owen1 so no guest acccouts will read any of the files in my home.21:08
Ilari sobersabre: TZ Charlie?21:08
cpg → cpg|biab21:09
← dcolish left21:09
sobersabre Ilari: no idea.21:09
or I'm not concentrated.21:09
Ilari owen1: Ah, that dreaded password prompt of death. Nasty diues to amount of of possible causes.21:10
← radarek left21:10
Ilari sobersabre: TZ Charlie => UTC+3h.21:10
owen1 Ilari: i am clening the .ssh folder21:10
sobersabre I'm not used to UTC at all... I'm referring to GMT :)21:10
but, yes.21:10
during the summertime21:10
anyway, do I understand the idea of letting evil committers to push into a bare repo:21:11
Ilari owen1: The usual causes are: 1) SSHD configuration preventing login, 2) Locked account. 3) Wrong key imported. 4) Non-default key imported. 5) SSHD doesn't like permissions.21:11
sobersabre let them put it htere, and I'll pick it up laters, compare, consider and merge whatever I want.21:11
owen1 Ilari: ok. btw, i can ssh with other accounts.21:12
fynn Ilari: what was that change in "git push" on 1.6.3 you mentioned yesterday?21:12
sobersabre owen1: let's start with the simple: do you have ANY kind of successfull ssh access to that machine ?21:12
← univate_ left21:12
sobersabre owen1: :) missed your post.21:12
owen1 sobersabre: yes21:12
sobersabre owen1: the 1st step would be check the server's configuration.21:12
mina86 → mina86|aw21:12
owen1 sobersabre: ok21:13
sobersabre usually it's located at /etc/ssh/sshd_config21:13
run: egrep -n -e "Allow|Deny" /etc/ssh/sshd_config21:13
→ Moult_ joined21:13
sobersabre this will show you who's allowed or not on group/user basis.21:13
→ magera joined21:14
sobersabre has egrep found anything ?21:14
magera → mager21:14
Ilari owen1: One trick to exclude 1 and 2 is to temporarily give the account a password, then try interactive login with it and then lock the password (but not the account!).21:14
← kumbayo_ left21:14
← dbaker1 left21:14
owen1 sobersabre: 72:# Allow client to pass locale environment variables21:15
→ SlipperyChicken joined21:15
sobersabre Ilari: 5) sshd likes permissions set correctly :)21:15
owen1: this means either you're on case 3)/4) which is same-same21:15
→ AdamaBot joined21:15
AdamaBot is there some command to add and commit at once?21:16
sobersabre owen1: do you have root access to that server ?21:16
← hax0r1337 left21:16
patrikf AdamaBot: commit FILENAME21:16
Ilari owen1: Note that not only permissions of .ssh/authorized_keys matter, but also permissions of ALL directories on path from root to it. None may have g+w or o+w.21:16
patrikf AdamaBot: or commit -a for all changes to tracked files21:16
owen1 sobersabre: ok. let me try creating the public key again.21:16
yes. got root.21:16
it's my desktop.21:16
Ilari AdamaBot: Or 'git commit <files>...'. But that doesn't work with untracked files.21:16
AdamaBot patrikf: i have a directory where i just added multiple directories/files. "git commit -a" brings the same prompt that asks me to manually add them21:16
sobersabre owen1: when you do this, make sure you're running it on the right machine :)21:16
AdamaBot 'git add .' ?21:17
owen1 sobersabre: ok. i deleted the .pub files on .ssh a few minutes ago.21:17
sobersabre AdamaBot: 1 command would be:21:17
Ilari AdamaBot: Or 'git add <directory>'21:17
AdamaBot Ilari: i added multiple directories and files in the current main directory21:17
sobersabre git add .; git commit -m "blabla added whatever was untracked :) "21:17
AdamaBot sobersabre: i see.21:17
→ tchan joined21:18
sobersabre AdamaBot: you can write an alias "supercommit" which would do this in 1 command.21:18
sucommit..21:18
Ilari owen1: I don't know does gitosis handle reinitialization correctly.21:18
owen1 Ilari: it worked before. give me a sec.21:18
sobersabre owen1: also, when you run ssh-keygen, you must be correct user on the source system.21:18
( I mean src system = ssh client, dst system = ssh server)21:19
← bobmcw left21:19
Ilari And ssh-keygen should only be rerun if you for some reason need to change the keys or you need key for some _non-interactive_ use.21:19
sobersabre gitosis sounds like a diagnosis.21:19
Ilari sobersabre: more like 'mitosis'.21:20
offby1 sobersabre: the prescription is: mouthwash21:20
owen1 before running sudo -H -u git gitosis-init < /tmp/id_rsa.pub, does it matter where do i loccate this file?21:20
and can i delete it after i run it?21:20
Ilari owen1: It needs to be readable by current shell process.21:20
offby1 my hunch is a) no and b) yes21:20
sobersabre offby1: you've got an acute gitosis, my lad. you've got to take a break, and treat it swiftly.21:21
offby1 I'm 60% sure gitosis-init copies the file to the right place21:21
I floss every morning21:21
offby1 nods gravely21:21
→ Fullmoon joined21:21
sobersabre takes off the glasses, and looks through offby121:21
sobersabre such a young man ...21:22
Ilari owen1: And also apporiate domains need the fd:use permission to shell domain to inherit the file descriptor.21:22
:-)21:22
sobersabre what happens these days, where a everybody is vulnerabls to such things, as gitosis.21:22
much more dangerous than the infamous swine flue.21:22
fynn Are the changes to git push in 1.6.3 fully reflected in the man page?21:23
sobersabre I'm sorry, but we'll have to amputate.21:23
Ilari correction: First wave of Influenza A(H1N1). :->21:24
→ stragerLN joined21:24
stragerLN Hello. Is there a way to merge only one commit without any other commits before that?21:24
Ilari sobersabre: No. But There's cherry-pick.21:25
fynn stragerLN: nope, you want cherry-pick21:25
sobersabre Ilari: you're too fast. and I'm not native english speaker, so I do miss some stuff.21:25
like cherry-pick.21:25
→ univate_ joined21:25
sobersabre what's that ?21:25
stragerLN sobersabre: He meant to beep me21:26
(I think)21:26
fynn yeah, he did.21:26
← doki_pen left21:26
Ilari sobersabre: Get diff of some change, apply it to current commit and commit the change.21:26
owen1 my home folder is 700, the same for ./ssh and ./ssh/known_hosts21:26
← wshimmy left21:26
sobersabre <|oo|>21:27
→ doki_pen joined21:27
owen1 do i need to give read/write permissions to everyone?21:27
Ilari owen1: No.21:27
owen1 i still can't ssh with git account.21:28
but can do it with others.21:28
Ilari owen1: SSHD uses DAC Override permissions to read the required files.21:28
owen1 Ilari: i am afraid i don't understand what u just said..21:29
mager Anybody use oauth-php here?21:29
owen1 i'll google for it21:29
← Moult left21:29
← doki_pen left21:29
sobersabre owen1: are your users - regular unix local users ?21:29
→ doki_pen joined21:29
Ilari owen1: Pastebin one of {~/.ssh/id_dsa.pub, ~/.ssh/id_rsa.pub, ~/.ssh/identity.pub} and ~git/.ssh/authorized_keys.21:29
sobersabre I mean stored in /etc/passwd,shadow,group,gshadow ?21:29
owen1 sobersabre: yes21:30
sudo adduser --system --shell /bin/sh --gecos 'git version control' --group --disabled-password --home /home/git git21:30
→ jrmuizel joined21:30
owen1 i followed: http://scie.nti.st/2007/11/14/hosting-git-repositories-the-easy-and-secure-way21:30
sobersabre owen1: if so, your user must be able to login with the keys, because password is disabled.21:30
Ilari owen1: (and if possible to same paste, makes it easier to compare).21:30
sobersabre owen1: what distro are you using ?21:31
I mean linux distro.21:31
stragerLN fynn: If I cherry-pick a commit then later merge the branch which has the commit I cherry-picked, will this be handled properly? I.e. will git detect it doesn't need to reapply the commit?21:31
sobersabre if linux....21:31
← robomalo left21:31
owen1 i use linux.21:31
sobersabre which one ?21:31
fynn sobersabre: nope, it will not.21:31
owen1 sobersabre: what do u mean? ubuntu. is there any other (-:21:31
fynn err, that was for stragerLN21:31
→ lobolife joined21:31
stragerLN fynn: Aww21:31
fynn: Is there any way?21:31
fynn stragerLN: as you can immediately see, applying a change in this way takes it out of context.21:32
sobersabre fynn: stragerLN, Ilari sorry I have s* nick :-\21:32
fynn so it doesn't have the same history, so the SHA1 changes.21:32
stragerLN fynn: Can I maybe pull the commit in the other branch out and put it into master?21:32
fynn stragerLN: I think you can rebase it onto a branch of master21:32
sobersabre owen1: are you able to login without password with another users ?21:33
stragerLN fynn: There are some small bug fixes in an experimental branch and I want them into master21:33
fynn: hmm?21:33
owen1 sobersabre: don't know. i always use passwords. i can try if u want me to.21:33
sobersabre no21:33
tomoj stragerLN: is the experimental branch local or has it been pushed somewhere?21:33
sobersabre I am just asking for comparison.21:33
stragerLN tomoj: Not pushed, none has yet21:33
sobersabre so, let's say you're trying to login as user git into gitserver.21:34
while you're a user1 on host gitclient21:34
ok ?21:34
fynn stragerLN: so you have master and foo, and you have some commits on foo that you want to merge to master right now, and some that you want to merge to master later?21:34
sobersabre owen1: are you there ?21:34
owen1 sobersabre: yes21:34
listening21:34
stragerLN fynn: Basically yes21:34
sobersabre you're user1@gitclient, trying to login AS git@gitserver.21:34
fynn stragerLN: OK, what you should have done is make those commits in a separate branch21:34
stragerLN fynn: I know =[21:35
sobersabre do you21:35
owen1 sobersabre: i am confused. let me think21:35
fynn stragerLN: right now, I think your base option is to rebase those commits onto master, then rebase foo onto master21:35
stragerLN fynn: I want to correct it somehow but I don't know how, really21:35
sobersabre have ~/.ssh as user1@gitclient ?21:35
tomoj stragerLN: one way to do it would be to use rebase to chop those commits out, then cherry-pick them21:35
stragerLN fynn: Okay, I'll look into that21:35
← moconnor_ left21:35
stragerLN tomoj: mmk =]21:35
fynn tomoj: why cherrypick?21:36
tomoj fynn: what would you do?21:36
sobersabre owen1: you maybe have an athrophied case of gitclient==gitserver...21:36
owen1 sobersabre: in ~/.ssh i only have 1 file: known_hosts (i am on the server)21:36
sobersabre owen1: is that the case ?21:36
offby1 if git hands you cherries, make cherry-ade.21:36
sobersabre owen1: be with me... :)21:36
← pjan left21:36
owen1 sobersabre: ok.21:36
sobersabre is gitserver==gitclient?21:36
fynn tomoj: hm, actually it may be the way to do it.21:37
→ wsc joined21:37
owen1 sobersabre: let me think...21:37
sobersabre owen1: there is only 2 possible answers: "yes", or "no".21:37
fynn I think he needs to use rebase to move the commits on top of the branching point21:37
then cherrypick them to master21:38
sobersabre is the server you're trying to connect via ssh the SAME as the machine you're trying to connect from ?21:38
owen1 sobersabre: are u asking if my git client is the same machine as the server?21:38
tomoj fynn: if he's cherry-picking the commits need to be gone from the branch, no?21:38
← bcardarella left21:38
sobersabre owen1: yes. that's the question.21:38
fynn tomoj: not if they're right on top of the branching point...21:38
hm, sec.21:39
stragerLN fynn/tomoj, If only one could drag-and-drop commits. xD21:39
owen1 sobersabre: i have 2 computers:21:39
sobersabre ok.21:39
so let's call them: gitserver and gitclient, for the sake of this discussion.21:39
is it OK for you ?21:39
owen1 sobersabre: desktop and laptop. the desktop will act as the server. and i'll have two clients21:39
→ memiux joined21:40
fynn tomoj: well, depends on what he can do. if he can rebase foo onto master, just making sure the commits he wants to merge are on top of master's HEAD, he can keep them in foo.21:40
sobersabre whatever.21:40
owen1 sobersabre: one on the desktop and the other on the laptop.21:40
sobersabre owen1: all I want is common *terminology*21:40
← Juan_ left21:40
owen1 sobersabre: agree21:40
tomoj fynn: and then fast-forward through just those commits?21:40
fynn tomoj: exactly.21:40
tomoj that would probably be easier than chopping them out21:40
fynn right. as long as he can rebase foo.21:41
WinterMute omg, are you two *still* analysing this?21:41
→ so_ joined21:41
tomoj WinterMute: no, different problem :)21:41
← djgera left21:41
stragerLN fynn/tomoj, So any conclusion? xD Still reading how rebase works21:41
AdamaBot → mederEsite21:42
sobersabre owen1: let's call your gitclient's user "user1"21:42
so_ upgraded to 1.6.2.2 on fedora, having a slight problem with the *~ pattern in my .gitignore, doesn't seem to be working21:42
sobersabre and gitserver's user "git".21:42
ok ?21:42
← mederEsite left21:42
fynn stragerLN: are you willing to rebase foo onto master?21:42
sobersabre owen1: agreed ?21:42
owen1 sobersabre: 1 sec (-:21:42
stragerLN fynn: Well I dunno what that means21:42
sobersabre you need to be now user1@gitclient21:42
stragerLN fynn: You mean make it as if I didn't branch at all? (foo is branched from master's HEAD)21:43
owen1 sobersabre: when u say gitclient's user, do u mean the user on the laptop?21:43
fynn stragerLN: it means your experimental branch will get all the commits made to master since you branched it off.21:43
tomoj stragerLN: may want to read this http://gitready.com/advanced/2009/03/20/reorder-commits-with-rebase.html21:43
→ wshimmy joined21:43
owen1 sobersabre: the laptop has user named yuka.21:43
sobersabre owen1: don't you laptop me.21:43
stragerLN fynn: There are no extra commits so that's fine21:43
tomoj: Will do21:43
owen1 sobersabre: on the propmt on the laptop is : yuka@yuka-laptop:21:44
fynn stragerLN: OK, so just read that blog from tomoj, reorder those commits to be the first21:44
sobersabre owen1: we agreed "gitserver" and "gitclient", so use only these terms.21:44
owen1 sobersabre: is it the phisical machines?21:44
fynn stragerLN: then you can fast-forward master through those commits.21:44
→ bcardarella joined21:44
sobersabre owen1: you don't want to use my language.21:44
owen1 sobersabre: or unix users?21:44
sobersabre :)21:44
stragerLN fynn: K I think I understand21:44
owen1 sobersabre: just want to make sure.21:45
sobersabre you're confusing yourself & me too :)21:45
fynn stragerLN: basically, what you're doing when you reorder those commits is change them as if they were made on master21:45
tomoj fynn: how would the fast-forward go? reset?21:45
owen1 sobersabre: gitclient=laptop21:45
sobersabre ok.21:45
fynn tomoj: yes21:45
sobersabre and gitserver=desktop ?21:45
owen1 sobersabre: gitserver=desktop t21:45
sobersabre and what's21:45
owen1 ye-21:45
yes21:45
sobersabre: let continue21:45
sobersabre and what's the username on "desktop" you're trying to connect as ?21:45
"git" right ?21:46
owen1 sobersabre: correct21:46
sobersabre ok.21:46
so_ nm, found somehow a developer checked in a ~ file into the repository ... booo21:46
sobersabre and yuka on laptop.21:46
owen1 sorry (-:21:46
sobersabre now: be yuka on laptop.21:46
tomoj stragerLN, fynn: so, I guess, do 'git rebase -i master' from your topic branch, move all the bugfix commits to the top, then 'git reset --hard XYZ' where XYZ is the sha of the last bugfix commit?21:46
sobersabre owen1: are you yuka on laptop now?21:46
owen1 sobersabre: yuka is the user on the laptop but when i ssh i do git@uoaiuaou21:46
sobersabre: yes21:46
← doki_pen left21:46
fynn tomoj: yes, that's how I think it should be done.21:47
sobersabre owen1: I understand this!21:47
→ doki_pen joined21:47
owen1 sobersabre: sorry..21:47
sobersabre when you're now yuka@laptop, run this command:21:47
cd; cd .ssh21:47
and tell me if you have the files:21:47
stragerLN tomoj: Q: What is the "top" ?21:47
← eddyp left21:47
sobersabre (id_rsa and id_rsa.pub) OR (id_dsa and id_dsa.pub) ?21:47
owen1 sobersabre: i see id_rsa and id_rsa.pub and known_hosts21:48
sobersabre ok21:48
now.21:48
← Sigma left21:48
tomoj stragerLN: the top of the editor window that pops up with all the commits21:48
sobersabre what are the permissions for the file id_rsa ?21:48
stragerLN tomoj: Heh, K21:48
← mager left21:48
sobersabre it should be mode should be at MOST 60021:48
owen1 sobersabre: -rw------21:48
sobersabre ok.21:49
does it belong to your yuka user ?21:49
(file owner)21:49
owen1 sobersabre: r u sure it's not id_rsa.pub that we should look at?21:49
sobersabre: yes, yuka owns it.21:49
← doki_pen left21:49
fynn tomoj: I'm starting to wonder how mercurial and bazaar do this kind of stuff without rebase21:49
→ doki_pen joined21:49
owen1 sobersabre: that's the folder where the client hold the private keys?21:50
sobersabre ok. now you need to copy id_rsa.pub to the desktop /tmp21:50
owen1: we're currently done with client side.21:50
owen1 sobersabre: why do i have two files?21:50
← corni left21:50
owen1 sobersabre: is it normal+21:50
sobersabre did you copy the file to owen1 id_rsa is private key for user yuka on laptop.21:50
yes, it's normal.21:51
owen1 sobersabre: got it. one is public and the other is private.21:51
← khmarbaise left21:51
owen1 sobersabre: let me copy the public one.21:51
sobersabre anyway, DID YOU COPY THE FILE id_rsa.pub TO THE desktop /tmp already ?21:51
owen1 sobersabre: give me a sec..21:51
← fujin left21:51
sobersabre when you did so, become root on desktop.21:52
→ khmarbaise joined21:52
owen1 mv id_rsa.pub /tmp/ (i did it without root)21:52
i can change the ownership to root if that's what u mean.21:53
stragerLN tomoj/fynn: K, I think I need a little help. One of the bug fixes (well, really an addition) changed one of the experimental parts so there's a merge conflict (changing something which doesn't exist)21:53
owen1 -rwx------ 1 oren oren 398 2009-06-14 16:37 id_rsa.pub21:53
sobersabre owen1: were you yuka@laptop, and did you copy /home/yuka/.ssh/id_rsa.pub to desktop:/tmp ?!21:53
tomoj stragerLN: yeah, that sucks21:53
stragerLN tomoj: Any resolution?21:53
sobersabre owen1: answer my question please... luckily it has only 2 answers: yes or no.21:54
tomoj stragerLN: well, you certainly can't pull that commit into master21:54
sobersabre owen1: ...21:54
stragerLN tomoj: Not in it's current state no21:54
owen1 sobersabre: i copied the file to a shared folder first,21:54
stragerLN tomoj: Can I split the commit?21:54
sobersabre owen1: ok.21:54
can you become root on desktop ?21:54
owen1 sobersabre: that i went to the desktop machine and moved it to /tmp (oren did it. oren is the user on the desktop)21:55
tomoj stragerLN: yes, see man git-rebase21:55
Gitbot stragerLN: the git-rebase manpage can be found at http://git.or.cz/man/git-rebase21:55
owen1 sobersabre: sure21:55
sobersabre report when ready, please.21:55
Ilari owen1: Run md5sum on both files (the one in /tmp and the one in ~/.ssh) and compare values?21:55
stragerLN tomoj: That can split commits too? xD21:55
owen1 sobersabre: i use sudo all the time if that's what u mean become root.21:55
cpg|biab → cpg21:55
sobersabre owen1: I mean become root. if using sudo, then run: sudo su -21:55
owen1 sobersabre: sure. let me do it.21:55
sobersabre then, after giving "oren"'s pass you're root.21:56
owen1 sobersabre: i am root21:56
Ilari: one sec. sobersabre is in the middle of something...21:56
sobersabre ok. run: ls -al /home/git | grep ssh21:56
and show me the output (should be 1 line). I want to make sure .ssh folder exists.21:56
adn the perms, etc.21:57
owen1 drwx------ 2 git git 4096 2009-06-21 14:21 .ssh21:57
sobersabre ok.21:57
← tatsuyao left21:57
← Bass10 left21:57
sobersabre now run:21:57
tomoj stragerLN: there's a splitting commits section in the manpage. you'll have to split the commit with an interactive rebase, leaving everything in order, then after that rebase again to move them around, I think21:57
sobersabre ls -l /home/git/.ssh21:57
this is several lines, what I want to see if you have a file named authorized_keys there21:58
← memiux left21:58
← bcardarella left21:58
owen1 total 421:58
-rw-r--r-- 1 git git 171 2009-06-21 14:21 authorized_keys21:58
i see only those 2 lines21:58
← bx2 left21:58
sobersabre what is the size of /tmp/id_rsa.pub ?21:58
stragerLN tomoj: K, thanks21:59
→ clug joined21:59
sobersabre owen1: what is the size of /tmp/id_rsa.pub ?21:59
owen1 39821:59
clug How do I merge two commits into one commit?21:59
sobersabre owen1: do this:21:59
stragerLN clug: git-rebase -i, then squash them, I think21:59
owen1 ls -lh22:00
sobersabre cp -p /home/git/.ssh/authorized_keys /home/git/.ssh/authorized_keys.BKP22:00
and then:22:00
cat22:00
sh1te22:00
clug no, git rebase deletes everything22:00
sobersabre cat /tmp/id_rsa.pub > /home/git/.ssh/authorized_keys22:00
2 commands.22:00
Ilari That breaks gitosis!22:00
sobersabre owen1: got it ?22:00
owen1 sobersabre: yes22:00
sobersabre ran it ?22:01
owen1 sobersabre: done22:01
sobersabre now run ls -l /home/git/.ssh/authorized_keys22:01
owen1 -rw-r--r-- 1 git git 398 2009-06-21 15:00 /home/git/.ssh/authorized_keys22:01
tomoj clug: rebase deletes everything?22:01
clug yeah you like do git-rebase -i and then this random stuff and then everything is gone22:02
subversion didn't delete stuff...22:02
sobersabre owen1: now, go back to become yuka @ laptop, and run:22:02
ssh git@desktop (well, the hostname of desktop or its IP)22:03
tomoj clug: I think you must not understand how rebase works...22:03
sobersabre owen1: have you logged in ?22:03
cpg → cpg|biab22:03
clug I thought it meant delete everything22:03
what does it really do?22:03
owen1 sobersabre: no. it asks for password22:03
sobersabre what is the command you're running, show me.22:04
← pmurias left22:04
tomoj clug: http://www.gitready.com/advanced/2009/02/10/squashing-commits-with-rebase.html22:05
owen1 sobersabre: ssh git@<server_ip>22:05
sobersabre ok22:05
can you become root back on the desktop ?22:05
owen1 sobersabre: yes22:05
sobersabre chmod 600 /home/git/.ssh/authorized_keys22:05
and then retry the ssh command.22:05
OK ?22:05
owen1 sobersabre: sure22:05
clug does "git remove ." remove hidden files?22:06
owen1 sobersabre: still the same22:06
tomoj clug: I don't think 'git remove .' does anything...22:06
stragerLN Uh I think I messed something up. I did 'git reset 1234567' (with a real commit). How do I get back to the latest commit?22:07
sobersabre owen1: did you run the command as yuka@laptop ?22:07
clug oh, thanks22:07
owen1 sobersabre: ssh? yes22:07
wereHamster stragerLN: git reset HEAD@{1}22:08
clug tomoj, oh maybe is was "git remove .*"22:08
tomoj clug: what I'm saying is I don't think "git remove" exists22:08
at least it doesn't exist for me22:08
clug oh, let me see22:09
it's git delete?22:09
stragerLN wereHamster: THanks =]22:09
clug oh maybe git-r-m22:09
stragerLN *Thanks =X22:09
parasti clug: it's git rm, no dashes22:09
clug so "git rm .*"? thanks, hidden files are annoying me22:10
sobersabre what is the output of: ls -l /home/git/.ssh/authorized_keys (on the server, the desktop)22:10
?22:11
owen1:22:11
I meant you ...22:11
owen1 sobersabre: one sec22:11
← sea-gull left22:11
clug is there a quick command to squash all commits into one commit?22:11
Then I could keep the log clean22:11
owen1 -rw------- 1 git git 398 2009-06-21 15:00 /home/git/.ssh/authorized_keys22:11
sobersabre ok.22:11
now, restart ssh server, just for the sake of the ...restart of the server, and retry logging from the laptop.22:12
if that didn't work, run this command:22:12
tail -f /var/log/auth.log (on the server)22:12
owen1 sobersabre: rebbot the computer?22:12
sobersabre NO!!!!22:12
run as root:22:12
owen1 sobersabre: ok22:12
sobersabre invoke-rc.d ssh restart22:12
did you ?22:12
owen1 yes22:13
sobersabre (on the server)22:13
owen1 let me try to connect22:13
sobersabre no.22:13
run the tail command already.22:13
why wait..22:13
owen1 ok22:13
sobersabre when you did, on the server, it will be "stuck", i.e. will wait for input.22:13
then you go to the client, and try to connect.22:13
during this observer server's terminal, and see what errors you will have there.22:13
stragerLN tomoj: Okay, I got the commits at the top of 'foo'. Now, what do I do with rebase to put it on 'master'?22:14
sobersabre paste the errors into pastebin.ca somewhere.22:14
owen1 Failed password for git from 192.168.1.1 port 46073 ssh222:14
sobersabre (sshd related errors)22:14
← WinterWeaver left22:14
tomoj stragerLN: so, foo looks like: master <- bugfixes <- other commits ?22:14
stragerLN yes22:14
sobersabre owen1: did you touch /etc/ssh/sshd_config at any time ?22:14
→ David-T joined22:14
tomoj stragerLN: so go into master and git reset --hard <SHA of last bugfix>22:14
owen1 sobersabre: no22:15
sobersabre I meaen did you edit it ?22:15
→ ereslibre_desk joined22:15
owen1 sobersabre: let me see22:15
stragerLN tomoj: Done22:15
← jrmuizel left22:15
sobersabre compare it's date with /etc/ssh/ssh_config22:16
if it's same, tell me.22:16
if it's not.... tell me :)22:16
fynn BTW, am I the only one who created shortcut aliases for status and checkout?22:17
clug is there a quick command to squash all commits into one commit?22:17
Then I could keep the log clean22:17
→ mederEsite joined22:18
owen1 sobersabre: sshd was changed on oct 08 and ssh on may 0822:18
stragerLN tomoj: And now it's done? =S22:18
sobersabre ok.22:18
you need to have this option enabled in sshd_config:22:18
owen1 sobersabre: btw, is it ok to have PermitRootLogin??22:18
mederEsite i accidently added say, a dozen .pyc files that i dont want anymore. i know there's a .gitignore type file i can add.. how can i already get rid of all the pyc files from my current repo?22:18
sobersabre RSAAuthentication yes22:19
mederEsite i would want to get rid of them recursively in my root diretory22:19
sobersabre do you have it ?22:19
stragerLN mederEsite: Do you mean destroy from your history?22:19
mederEsite perhaps22:19
sobersabre owen1: got it ?22:19
mederEsite or get rid of in my current commit22:19
as in22:19
tomoj stragerLN: yes, when you push the remote will fast-forward22:19
mederEsite not really my current commit but22:19
owen1 sobersabre: got what?22:19
mederEsite the present, heh :p22:20
current state of my repo, whatever22:20
i dont really mind history-wise22:20
sobersabre does the /etc/ssh/sshd_config file contain a line I wrote "RSAAuthentication yes" ?22:20
mederEsite if i can get rid of it in history too, fine22:20
owen1 sobersabre: yes22:20
sobersabre does it have:22:20
PubkeyAuthentication yes22:20
?22:20
owen1 sobersabre: yes22:20
sobersabre: btw, this is commented: #AuthorizedKeysFile %h/.ssh/authorized_keys22:21
sobersabre owen1: what's the size of /home/yuka/.ssh/id_rsa.pub on the client side ?22:22
stragerLN tomoj: Many thanks. It works now. =]22:23
← ph^ left22:23
owen1 sobersabre: 39822:23
→ Peaker joined22:24
Peaker hey - before using git, lots of people told me that git tracks line moves between files. how do I see such "smart diffs" that show me which lines moved between files?22:24
now that I've used git a while, I've never seen that supposed feature in action, so I wonder :-)22:25
clug so do you do "git rm something" or "git `rm something`"22:26
offby1 the first one22:26
no backticks22:26
clug: your "quick command to squash all commits into one commit" is "git rebase"22:27
Peaker btw, I'd really like to have my build/testing process create automatic commits and then reset --soft those commits. And the git staging area is really standing in the way :-( When I was a git newbie people told me everyone learns to eventually like the staging area, but I see it as simply a clumsy UI element22:27
owen1 sobersabre: did u give up?22:30
← fridim_ left22:30
sobersabre no. try running the ssh client with -v option.22:30
ssh -v git@<desktopIP>22:31
clug ooh, it would be cool to make it so every 5 minutes git added my work, commited it, and squashed all the commits into one commit22:31
sobersabre and send the output to the pastebin.22:31
and send me the link to the paste22:31
ok ?22:31
Peaker clug: why would you want to squash everything?22:32
← fynn left22:32
sobersabre owen1: ok ?22:32
Peaker clug: you mean just the yet-uncommitted changes?22:32
owen1 sobersabre: ok. 1 sec22:32
clug I mean squash entire log22:32
tomoj Peaker: what do you mean "create automatic commits and then reset --soft"? what are you trying to accomplish?22:33
Peaker clug: something like an automatic git commit -a --amend -m"Autosave at %date" ?22:33
clug so I always have 1 commit every 5 minutes and 1 entry in the log22:33
sobersabre I'm logging off will be back in 2-3 mins.22:33
← sobersabre left22:33
clug idk, I don't use those "-" things they delete stuff sometimes22:33
Peaker tomoj: I'd like every build to be labeled with a ref that uniquely identifies the entire contents of the build sources. I would like not to have any private changes that are not accounted for in this ref22:33
tomoj: basically everything that would like to script commits automatically has to deal with the staging area which just complicates such scripts' lives22:34
tomoj Peaker: isn't that what tags are for? maybe I don't understand what you're trying to do22:35
clug I don't really see the point of version control anyway22:36
← ereslibre left22:36
→ sobersabre joined22:36
sobersabre owen1: what's up.. what's the paste link ?22:36
clug I still have to make backup copies of the project22:36
and it randomly deletes stuff22:36
who made git anyway? al gore?22:37
tomoj clug: git does not randomly delete stuff22:37
Peaker tomoj: people might try to build when they aren't fully committed22:38
tomoj: the idea is that anyone can build a system and run tests on it - and I'd like test failures to be able to precisely identify the exact tree that built and tested22:38
tomoj Peaker: oh, I see22:38
Peaker tomoj: it would be nice if I could have a script that tried to make sure "git status" was completely clean before building, and then later used git reset --soft maybe to return to the previous situation22:39
tomoj Peaker: won't these commits get pruned anyway? or do you need them only short-term?22:40
sobersabre owen1: ... u there ?22:40
owen1 sobersabre: http://pastebin.com/m832607322:40
← mattikus left22:42
sobersabre owen1: I have an idea :)22:43
generate a dsa key.22:43
cehteh eek dsa :P22:43
sobersabre btw, you didn't paste the sshd_config file from the server (where git user is)22:43
Peaker tomoj: short-term is ok, also, they can be tagged to remain ungc'd22:43
tomoj: with some random auto-generated date-based tag22:44
owen1 sobersabre: ? i pasted the ssh log from the client22:44
sobersabre: ssh -v git@432423422:44
sobersabre yes.22:44
I've read it.22:44
← lobolife left22:44
sobersabre for some reason there's a skip after sending the rsa key.22:44
owen1 sobersabre: do u want to see the sshd_config?22:44
sobersabre and I want to understand why.22:44
yes.22:44
owen1 ok22:44
sobersabre please paste it and put here the link.22:44
owen1 sure22:44
← name left22:45
→ lucasvo_ joined22:46
owen1 sobersabre: http://pastebin.com/m33bf4e1d22:47
PermitEmptyPasswords no22:47
sobersabre: i think we got it22:47
clug Does get have a command to delete a random file?22:48
like git rm --rand22:48
lucasvo_ I'm trying to set `git config user.name '#{commit_name}'` in a pre-commit hook, it doesn't seem to affect the most current commit but only the one after that. Is there a different way to do that? (I'm using this script: http://tinyurl.com/npc5pw)22:48
Peaker clug: you trolling?22:48
clug no22:48
tomoj owen1, sobersabre: btw, did you notice Ilari's warning?22:49
owen1 sobersabre: i changed it no yes, restarted ssh and it didn't work.22:49
tomoj: no22:49
tomoj owen1: you broke gitosis by using a plain public key for authorized_keys22:50
owen1 tomoj: mmmm...so when i create a keypair i have to type a password?22:50
sobersabre tomoj: which one (there were several comments from Ilari)22:50
← k0001 left22:50
owen1 sobersabre: maybe i have to use a password when i create the pairs.22:51
tomoj sobersabre: the one where he said, 'that breaks gitosis'22:51
← asksol left22:51
sobersabre ok22:51
→ k0001 joined22:51
tomoj owen1: by "plain" I don't mean no password, I mean gitosis needs to add some extra stuff in there22:51
← Moult_ left22:51
→ pulpfiction joined22:51
sobersabre owen1: you can change password on an existing private key.22:52
tomoj: does it change the comment of the public key ?22:52
owen1: can you try using getosis to reset the keys ?22:52
owen1 sobersabre: what exectly should i try?22:52
→ Moult joined22:53
→ jrmuizel joined22:53
tomoj owen1: the guide you used for setting up gitosis should tell you what to do with your public key22:54
owen1: but you need to fix your ssh problem first22:54
owen1 tomoj: ok22:54
sobersabre I am not familiar with getosis22:54
butm, you're supposed to become a git user on the desktop.22:54
it's easy: become root,22:54
and then run:22:54
su - git22:54
owen1 sobersabre: should i change the PermitEmptyPasswords no to yes?22:55
← wsc left22:55
sobersabre no, don't touch that just yet.22:55
owen1 ok22:55
sobersabre anyway, are you git user on the desktop already /22:55
?22:55
owen1 no. 1 sec22:55
sobersabre: i run su - git22:56
i guess i am the git now...22:56
sobersabre run this command:22:56
which gitossi22:56
sorry: which gitosis22:56
sorry!22:56
which gitosis-init22:56
owen1 /usr/local/bin/gitosis-init22:56
sobersabre ok.22:56
now run this:22:57
gitosis-init > /tmp/id_rsa.pub22:57
SORRY!22:57
the oposite!22:57
gitosis-init < /tmp/id_rsa.pub22:57
note the direction of the redirection :)22:57
you're feeding gitosis-init with the public key.22:57
ok ?22:57
owen1 -su: cannot open /tmp/id_rsa.pub: Permission denied22:58
sobersabre oh,..22:58
get back to root: exit22:58
owen1 sudo su22:58
sobersabre nono22:58
exit22:58
you've become git from root shell.22:58
owen1 ok22:58
sobersabre so get back there.22:58
owen1 i am back22:58
sobersabre and now, run chmod 644 /tmp/id_rsa.pub22:58
so git can read it22:58
and then repeat: su - git22:58
and then the gitosis-init < /tmp/id_rsa.pub22:59
owen1 ok. got Reinitialized existing Git repository in /home/git/repositories/gitosis-admin.git/22:59
(twice)22:59
sobersabre ok...23:00
→ lobolife joined23:01
owen1 i still can't ssh with git@4p23343423423:01
← pulpfiction- left23:01
lucasvo_ how can I set the git username before committing in a pre-commit hub?23:01
several people are working on the same git checkout and we want to be able to track who's doing what23:02
tomoj lucasvo_: why??? :(23:03
sobersabre now, the ssh problem...23:03
owen1: I think you should increase server's logging verbosity.23:03
1. backup sshd_config file in /etc/ssh23:03
lucasvo_ tomoj: because merging every half hour is too complicated23:03
owen1 sobersabre: ok23:04
sobersabre are you there ? backed up ?23:04
open the file /etc/ssh/sshd_config (on the desktop!)23:04
→ abw__ joined23:04
owen1 sobersabre: it's open23:04
sobersabre change the line to Logging DEBUG23:05
the "Logging INFO"23:05
owen1 sobersabre: done23:05
sobersabre or .. just make it "VERBOSE", not DEBUG23:05
owen1 ok. 1 sec23:05
→ fujin joined23:06
owen1 done23:06
sobersabre ok ?23:06
tomoj lucasvo_: but.. now they can't even work at the same time, can they?23:06
→ pflanze joined23:06
pflanze How can one proceed a "git svn clone"?23:06
← sakyamuni left23:06
→ Paf joined23:06
offby1 ?23:06
pflanze: please rephrase your question; I didn't understand it23:06
sobersabre restart ssh23:07
owen1 o23:07
ok23:07
sobersabre on the desktop (invoke-rc.d ssh restart)23:07
owen1 done.23:07
sobersabre now poll the logfile:23:07
tail -f /var/log/auth.log23:07
pflanze After two git svn cloning jobs each running for 2.5 hours both have failed with "RA layer request failed: REPORT of '/svnroot/opalvoip/!svn/vcc/default': SSL negotiation failed: Secure connection truncated (https://opalvoip.svn.sourceforge.net)"23:07
owen1 Server listening on :: port 22.23:07
pflanze I guess someone has restarted the servers on sf.23:08
owen1 sobersabre: should i try to ssh?23:08
sobersabre owen1: ok, now, ssh, and during this look at the log.23:08
owen1 ok23:08
sobersabre maybe you can see something fishy.23:08
lucasvo_ tomoj: yes we can23:08
owen1 Connection from 192.168.1.1 port 5848123:08
Failed none for git from 192.168.1.1 port 58481 ssh223:08
cehteh pflanze: for CVS sf once had a tarball download service for the whole repository23:08
owen1 Failed publickey for git from 192.168.1.1 port 58481 ssh223:08
cehteh i wonder if they have such for svn too23:08
pflanze cehteh: I know, rsync. Was (is?) wonderful. But this is SVN.23:09
tomoj lucasvo_: how? what if I need to check out a different branch? everyone else's work will be wiped out23:09
cehteh thats in my experience much faster and more reliable23:09
Ilari owen1: Does it say more about why pubkey fails?23:09
← voker57_ left23:09
owen1 Ilari: no23:09
→ voker57_ joined23:10
sobersabre owen1: edit logging to be debug.23:10
Logging DEBUG23:10
and restart the server23:10
owen1 ok. 1 sec23:10
sobersabre then retry the connection, whilst polling the log23:10
and tell us what's up with the public key.23:11
I suspect it's still permissions/ownership.23:11
on ssh server, but I'm not sure what and where.23:11
owen1: you have a long path until the user's home dir.23:11
← tvw left23:12
→ hax0r1337 joined23:12
sobersabre I mean for user git.23:12
I think some part of this path is world writeable or something like that.23:12
owen1 the log shows the same.23:12
drwxr-xr-x 5 git git 4096 2009-06-21 15:58 git23:13
→ Shadou joined23:13
owen1 this is the git folder23:13
sobersabre chmod to 750 git23:13
owen1 sobersabre: ek23:13
ok23:13
sobersabre and retry to login as git user from the laptop.23:13
← qhoxie left23:14
owen1 sobersabre: it was already 75023:14
i thinkg23:14
drwxr-xr-x23:15
sobersabre this means 75523:15
r-x = 101 = 523:15
rwx = 1 1 1 = 723:15
owen1 sobersabre: so that's how to tell! thanks23:15
sobersabre so, can you login or NOT ?23:16
owen1 sobersabre: no23:16
the same23:16
sobersabre can you run this command on the server:23:17
moment.23:17
pflanze offby1 / cehteh : seems one can continue by "cd $clonedir; git svn fetch" (not yet finished, so can't say for sure whether that works)23:18
← warthog9 left23:19
→ wsc joined23:21
← chris2 left23:22
offby1 aah23:25
pflanze: I've actually done that; it did indeed work23:25
pflanze c00l23:25
← Yuuhi left23:26
→ bremner_ joined23:26
← fujin left23:26
Shadou → fujin23:26
← clug left23:26
← lobolife left23:27
→ rpag joined23:28
← dreiss left23:30
← ProperNoun left23:32
← rettub_ left23:32
← lbt left23:32
→ rettub_ joined23:33
offby1 svn is so slow that, with a large project, the server is likely to time out at some point in the process :-|23:36
← Octalot left23:36
→ sgrimm joined23:36
→ romao joined23:41
→ qhoxie joined23:41
← rpag_ left23:43
Ilari owen1: I know one problem which would cause keypair auth to fail and is essentially impossible to debug...23:46
→ montylounge joined23:46
owen1 Ilari: please share..23:47
Ilari owen1: Mismatched keypair on client side. Really nasty tendency to lead to wild chase...23:48
sobersabre Ilari: you say he's created the keypair on different occasion ? :)23:48
or diff. users ?23:48
← Martz left23:48
Ilari Nope, .pub key being for different key than the private one.23:48
sobersabre it is possible, Ilari indeed...23:48
owen1 sobersabre: maybe. i want to remove all keys23:48
sobersabre: from both machines. where are they stored? let's get rid of them.23:49
sobersabre start on the client only:23:49
be that user on laptop,23:49
and mv the keys into some ~/.ssh/old directory.23:49
Ilari There's way to get comparable IDs: 'ssh-keygen -y -f ~/.ssh/id_rsa | md5sum' dumps it for private key and 'cut -d \ -f 1-2 ~/.ssh/id_rsa.pub | md5sum' for public key.23:49
owen1 sobersabre: ok. what about the * we changed on the server?23:49
← montylounge left23:50
sobersabre it doesn't matter.23:50
since it works for root :)23:50
anyway, run the ssh-keygen -t rsa again, when ~/.ssh has no keys (except the dir, and known hosts file)23:51
owen1 sobersabre: ok. should i delete id_rsa and id_rsa.pub?23:51
Ilari owen1: Move them aside?23:51
sobersabre whatever you wish, you can delete, you can move.23:51
I'm old-fashion person... so I move things.23:51
owen1 sobersabre: delete and also the known_hosts.23:51
sobersabre: one sec23:51
sobersabre I don't think you should ...23:51
known hosts stores all the ssh servers' signatures...23:52
the annoying "somebody's trying to look under your kilt" thingie is annoying.23:52
owen1 sobersabre: why do i care? i am on the client now.23:52
sobersabre anyway, now, you need to sopy the id_rsa.pub file again, but make sure you put the NEW one on the desktop.23:53
known_hosts is the client's file.23:53
and it stores all the signatures of all ssh servers you've connected.23:53
owen1 sobersabre: give me 2 minutes.23:53
← jrmuizel left23:53
sobersabre if the sigs change, the ssh client warns you.23:54
← robinr left23:55
→ lobolife joined23:55
sobersabre Ilari: ... what a tiredsabre I am.23:55
didn't think of the case with mismatch at all.23:55
→ panpainter joined23:56
Ilari sobersabre: Its probaby very rare, one of the nastiest and getting ID out of private key one can compare to public key is not trivial.23:57
← panpainter left23:57
← hpoydar left23:58
→ panpainter joined23:58
← panpainter left23:58
→ memiux joined23:59

Logs Search ←Prev date Next date→ Channels Documentation