IRCloggy #git 2011-09-18

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.

2011-09-18

→ khaije|minder joined00:00
← vvsh left00:01
→ cybersphinx joined00:01
→ subbyyy joined00:02
→ Fudge joined00:02
→ janiczek joined00:04
→ lacrymol1gy joined00:04
← lacrymol1gy left00:04
← ludde left00:04
→ Targen joined00:05
→ unreal joined00:05
← the_cybersphinx left00:06
→ Samual joined00:06
← Cipher-0 left00:09
← janiczek left00:10
← kadoban left00:13
← bigkm left00:14
→ Yannick_Gagnon joined00:15
← field_it left00:16
← Yannick_Gagnon left00:17
← heppy left00:23
→ heppy joined00:23
→ maestrojed joined00:24
← berserkr1 left00:25
→ Ivoz joined00:26
← Ivoz left00:26
→ Ivoz joined00:26
→ tshauck joined00:27
→ Heimidal joined00:28
→ metcalfc joined00:31
← legumbre left00:32
→ legumbre joined00:33
← maestrojed left00:35
← metcalfc left00:39
← Heimidal left00:40
→ eletuchy joined00:41
← henriquev left00:43
→ mysphyt joined00:44
← pom left00:47
mysphyt Okay. I'm having brain problems, here. I've got four repos: working_a pushes to bare_a. bare_b is a clone of bare_a. working_b pulls from bare_b. I'm having problems getting changes in working_a all the way over to working_b. When I run git fetch on bare_b, it says "master -> FETCH_HEAD", but the most recent changes in bare_a aren't in git log. What am I missing?00:48
cbreak mysphyt: just make working_b pull from bare_a00:50
← Klif1980_ left00:51
cbreak after that you can push to bare_b00:51
alternatively, reset after fetching00:51
mysphyt cbreak: Thanks. I knew there was a simple solution. I still feel like there's something important I'm not understanding.00:51
← wokkawokka1 left00:53
← tdebat left00:53
→ kadoban joined00:53
→ tdebat joined00:54
← pen_ left00:55
← ZapZ left00:57
← jbrokc left00:58
sitaram mysphyt: is there ever going to be a push sequence going the other way? (wb -> bb -> ba -> wa)?00:58
mysphyt sitaram: Nope. In fact, the whole reason for all of this is to get rid of wa and ba.01:00
sitaram oh ok01:00
mysphyt This is a web app. Working copies are the live files, bare repos are pushed to by my numerous working computers. I'm switching hosts, so I'm trying to get stuff across. Probably I'm making things too difficult.01:01
→ rbuck joined01:04
→ ZapZ_ joined01:04
→ frogonwheels joined01:05
← hsingh left01:05
← c00kiemon5ter left01:05
← thiago left01:05
→ guns joined01:05
← mandric left01:07
← tdebat left01:11
→ tdebat joined01:12
→ c00kiemon5ter joined01:12
← c00kiemon5ter left01:12
→ c00kiemon5ter joined01:12
→ mlncn joined01:12
→ JamesLeeds joined01:15
← sbell left01:15
← rendar left01:15
← mysphyt left01:18
→ maestrojed joined01:18
← rbuck left01:19
← AaronMT left01:21
sythe-away → sythe01:23
→ kodie__ joined01:23
← kodie left01:24
Spyraway → Spyro01:24
→ blaenk joined01:27
→ bolden joined01:30
← JamesLeeds left01:31
← chrisf_ left01:31
→ eigenburak joined01:31
← kadoban left01:33
← tshauck left01:33
← SimonNa left01:35
→ Weiss joined01:37
→ albel727 joined01:37
← JasCo left01:37
→ bigkm joined01:38
← p3rror left01:42
← avsej left01:42
← notbrent left01:42
← maestrojed left01:43
← brian_g left01:47
← normanrichards left01:49
← warthog9 left01:51
→ SimonNa joined01:52
← sh_sh left01:52
→ warthog9 joined01:53
→ notbrent joined01:54
→ avsej joined01:55
→ chrisf_ joined01:56
← gusnan left02:01
→ normanrichards joined02:01
← notbrent left02:01
← bitkiller left02:02
→ ISF joined02:03
← mattdipasquale left02:04
← eletuchy left02:06
→ envi joined02:06
← warthog9 left02:08
→ akosikeno joined02:08
→ gnufied joined02:10
← SimonNa left02:10
← fisted left02:13
← gnufied left02:15
← chuck left02:15
→ chuck joined02:16
→ fisted joined02:17
← eigenburak left02:18
← bigkm left02:23
→ roue joined02:27
roue hola02:27
I'm trying to download a particular release of linux from github via the git command line. "git clone https://github.com/torvalds/linux.git" seems like it's getting me the latest patches, but how do I request the last stable release (3.0.4)? thanks.02:28
← Ivoz left02:30
→ dexter_e joined02:30
cirwin once you've downloaded it, "git tag" will give you the list of tags; and then "git checkout <tagname>" will give you that version02:30
roue cirwin thanks much.02:31
← Nedly left02:31
→ tos9 joined02:32
tos9 Anyone have a tip on why this is happening? http://paste.pound-python.org/show/12677/02:32
roue cirwin - oddness here. Git tag shows 3.0, then 3.0-rc1,2,3,4,5... but no 3.0.4 . Maybe I'm not understanding how these releases work.02:32
tos9 (changing to dirty doesn't ignore either)02:32
cirwin roue: in that case you'll have to ask someone who knows something about linux.git02:33
← tureba left02:33
→ soulcake joined02:34
rue I just see 3.0 and 3.102:35
← mlncn left02:36
→ henriquev joined02:39
← sythe left02:39
→ nadavoid joined02:43
← chrisf_ left02:44
→ Ivoz joined02:46
← Ivoz left02:46
→ Ivoz joined02:46
cybersphinx roue: The stable kernels are in a different repo, git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-3.0.y.git for 3.0.x (don't know if/where a non-kernel.org repo is).02:47
← Weiss left02:47
→ sythe joined02:49
→ bburhans joined02:50
→ mattdipasquale joined02:52
← Chillance left02:53
← dexter_e left02:53
→ tureba joined02:55
← Targen left02:55
← ISF left02:56
→ warthog9 joined02:57
→ Weiss joined02:59
→ dexter_e joined03:00
← tureba left03:02
roue cybersphinx thanks for the pointer.03:03
← bolden left03:04
roue cybersphinx unfortunately kernel.org is out of commission for now (and the reason I'm looking at github...)03:04
cybersphinx http://thread.gmane.org/gmane.linux.kernel/1192147/focus=1192152 sounds like there is no other repo.03:08
→ ISF joined03:10
→ ericbarnes joined03:10
→ tureba joined03:11
← kodie__ left03:12
→ kodie joined03:12
← dexter_e left03:15
← tureba left03:15
← Bass10 left03:17
→ tureba joined03:21
→ kvanderw joined03:21
← tureba left03:27
← kukks left03:28
→ cjs joined03:28
→ haydenmuhl joined03:30
← kvanderw left03:30
→ kvanderw joined03:31
← ericbarnes left03:32
→ tureba joined03:35
muneeb → muneeb|aft03:35
← muneeb|aft left03:36
→ alester joined03:38
→ infid joined03:39
← tureba left03:42
← nadavoid left03:45
→ tureba joined03:46
→ nadavoid joined03:47
← madewokherd left03:48
← guns left03:48
EugeneKay linux kernel? Linus registered on github and put it up, last I saw....03:49
← akosikeno left03:50
← gbacon left03:51
← juvenal left03:51
← soulcake left03:51
→ akosikeno joined03:52
← bonhoffer left03:52
← tureba left03:53
→ guns joined03:54
→ gnufied1 joined03:55
← caseymcg left03:56
← ISF left03:56
← heppy left03:59
→ macmartine joined03:59
← guns left04:01
→ tureba joined04:03
→ n8o-mba joined04:03
→ metcalfc joined04:08
← tureba left04:08
← cakehero left04:09
→ tureba joined04:10
← gnufied1 left04:11
→ dexter_e joined04:12
→ cytrinox_ joined04:13
→ heppy joined04:13
→ gnufied joined04:14
← cytrinox left04:16
cytrinox_ → cytrinox04:16
← tureba left04:17
→ dfr|mac joined04:19
Tac_Home → atdiehm_home04:19
→ tureba joined04:21
← macmartine left04:22
← mattdipasquale left04:23
→ RichardBronosky joined04:23
← qian left04:23
← subbyyy left04:24
→ randomubuntuguy joined04:24
→ subbyyy joined04:25
→ mattdipasquale joined04:26
← dexter_e left04:26
← dfr|mac left04:30
← randomubuntuguy left04:33
← gnufied left04:34
→ gnufied joined04:36
→ w09x joined04:37
← larie left04:39
← w09x left04:39
→ pharvey joined04:40
→ kennethreitz joined04:42
← ZapZ_ left04:42
→ topeak joined04:43
→ mlncn-agaric joined04:46
← sivy left04:49
→ sivy joined04:50
EugeneKay → EugeneKaway04:50
→ Coriolan joined04:53
← alester left04:54
→ wolftankk joined04:54
← sivy left04:55
→ iband joined04:56
→ aminpy joined04:57
← Coriolan left05:00
→ cakehero joined05:01
→ macmartine joined05:03
→ Coriolan joined05:07
→ dexter_e joined05:07
← gretch left05:09
← gnufied left05:09
← n8o-mba left05:10
← miguet left05:11
→ bolden joined05:12
→ kingfishr joined05:13
→ abec0 joined05:13
→ Nedly joined05:14
← tureba left05:15
→ vmil86 joined05:15
← nadavoid left05:17
→ tureba joined05:18
← asakura left05:19
← subbyyy left05:19
← rane_ left05:19
→ asakura joined05:22
← aalex left05:24
← johnkpaul left05:25
→ syphar joined05:26
→ rchavik joined05:27
← rchavik left05:27
→ rchavik joined05:27
← madprops left05:32
→ LouisJB joined05:33
← Error404NotFound left05:34
→ gretch joined05:37
→ dmac joined05:40
→ jwpeddle joined05:41
← LouisJB left05:41
→ harsh joined05:42
← toobluesc left05:42
→ madsy joined05:42
← madsy left05:42
→ madsy joined05:42
→ toobluesc joined05:42
→ notbrent joined05:42
jwpeddle ok what am I missing? I have a submodule in directory foo, git status tells me nothing is awry. I run git status later and it tells me that "modified: foo (untracked content)". How could my submodule dir be "untracked content"?05:43
oh, I should have run git status inside the submodule. d'oh05:45
Now I see the problem05:45
← kvanderw left05:45
← macmartine left05:47
← madsy left05:49
← cakehero left05:49
→ jmah joined05:50
→ madsy joined05:51
← madsy left05:51
→ madsy joined05:51
← sythe left05:52
← metcalfc left05:52
← kingfishr left05:53
→ m4dc0d3r joined05:56
→ Jon47 joined05:57
→ flavius joined05:57
← m4dc0d3r_ left05:57
→ kingfishr joined05:59
→ jonshea joined06:00
← Ivoz left06:00
→ Cromulent joined06:01
← Jon47 left06:01
→ shiba_yu36 joined06:02
→ ZapZ joined06:03
← ehsan left06:03
← chuck left06:04
← blaenk left06:04
← Coriolan left06:04
→ chuck joined06:06
→ caseymcg joined06:07
→ ManDay joined06:08
ManDay Can someone recomment a tutorial for Git which covers things like branching, what tracking is, origins etc.?06:09
← orafu left06:09
← wolftankk left06:09
selckin google will gladly direct you to a million of those06:10
→ towski joined06:10
ManDay google is pretty bad at making recommendations06:10
but I guess you knew that and just had the urge to say *something*06:10
→ orafu joined06:11
tango_ ManDay: pro-git06:12
ManDay thanks tango_06:12
← mastro left06:13
sitaram ManDay: not precisely what you asked but http://sitaramc.github.com/gcs/ may help also06:15
*I wrote it recently and would like opinions)06:15
← dmac left06:15
ManDay sitaram: Thanks. I will try to review it but now I've started with pro-git and it looks good. I'll make it my second choise :)06:16
← segher left06:16
ManDay sitaram: from the index it looks like a tutorial which is not bottom up06:17
I'll surely read it later then.06:17
I guess it offers more detail06:17
← urbanmonk left06:18
→ dfr|mac joined06:19
← heppy left06:19
← adamm left06:19
ManDay sitaram: actually, I'll try yours first now. It looks a little more condensed, probably just what I need.06:19
← hyperair left06:19
→ jimmy1980 joined06:20
→ ReekenX joined06:20
→ hyperair joined06:21
sitaram ManDay: prog-git is a book, this URL I gave is one article, and mostly pictures. pro-git also will teach you commands and options, this URL is only concepts but might help understand everything else much easier06:22
ManDay sitaram: in this picture here: http://sitaramc.github.com/gcs/images/fig000906:23
It would be interesting to be told what happens if you do a commit on the abandoned_branch06:23
(I guess I know the answer but it would probabaly be more interesting than a commit on master)06:24
→ segher joined06:24
sitaram ManDay: yes but that would require knowing how to *get* to the abandoned branch. Anyway see later for the part on detached HEAD06:24
oh sorry06:24
wrong answer; I forgot I had a *name* there06:24
well nothing really unsurprising happens; it just grows from there... (from commit 3)06:25
ManDay yes06:25
sitaram: yes, I know, yet it would erphaps be more telling to let that branch grow rather than master (or both)06:25
tos9 Anyone have a tip on why this is happening? http://paste.pound-python.org/show/12677/06:26
→ heppy joined06:26
sitaram ManDay: what would be more useful is to simply repeat fig0005 because it already shows a non-leaf node having a branch name06:27
← atdiehm_home left06:28
→ chrisf_ joined06:28
← harsh left06:30
jmah tos9: what's unexpected about it?06:31
→ Tac_Home joined06:31
tos9 It's marking a submodule as dirty even though I have it ignored06:32
Or am I making an error?06:32
→ maestrojed joined06:32
→ Ivoz joined06:33
← Ivoz left06:33
→ Ivoz joined06:33
→ tewecske joined06:33
ManDay sitaram: You know where you come to speak of the "really *really* important point"...06:34
jmah tos9: it should ignore untracked files in the submodule, but not different commits06:34
maybe you can ignore = all06:34
tos9 jmah: I'll try that. Thanks.06:35
← trevorgreen_ left06:37
tos9 Yup that did it. Thank you.06:37
→ trevorgreen_ joined06:38
← heppy left06:39
← Swimming_Bird left06:42
← maestrojed left06:42
→ defendguin joined06:43
defendguin using git how can i look at the difference between a file from last week and a file from today?06:44
cirwin git diff "master@{1 week}" -- <file>06:44
← kingfishr left06:45
→ kingfishr joined06:45
kevlarman cirwin: that produces slightly unexpected results06:45
Ivoz tos9: vim-pathogen seems to be a lot simpler than that06:45
sitaram ManDay: ...yes?06:45
tos9 Ivoz: how so? With pathogen you have a submodule to manually manage for each of your plugins, rather than managing none of them (and just hiding one for OCD's sake :)06:46
cirwin kevlarman: if you want more help, you'll have to be more specific about the unexpectedness06:46
kevlarman cirwin: you shouldn't advise people to use that06:46
cirwin why not?06:47
→ whitman joined06:47
← tewecske left06:48
kevlarman e.g. you didn't pull for a month, then you pull, and immediately look up master@{1 week ago}, you get a 1 month old commit06:48
ManDay sitaram: well it does not become quite clear *what* is actually so important about it. in the concluding sentence you even remain a bit vague06:49
sitaram: over all your tut is a good introduction to the concepts of distributed vcs06:49
cirwin kevlarman: how would you do it more robustly?06:49
kevlarman cirwin: don't use the reflog06:49
cirwin what does git log --after use?06:50
kevlarman not the reflog06:50
that would do it06:50
sitaram ManDay: maybe it needs more emphasis but the words are there. (People from Hg don't get this part at all, since Hg encodes the branch name inside the commit so it matters where it came from!)06:50
cirwin git diff $(git rev-list --after="1 week" --reverse -1) -- <file>06:50
a little verbose perhaps06:50
→ tewecske joined06:50
→ nicoulaj joined06:51
→ dSebastien joined06:52
cirwin and you can't combine --reverse and -1 :(06:52
kevlarman uh06:52
lorenl does git have any mechanism to stash away changes without a commit so I can do things like change to another branch to apply a fix?06:52
kevlarman yes you can06:52
i do it all the time06:52
lorenl: it's named git stash06:52
DrNick lorenl: yes. git calls it the "stash"06:52
cirwin kevlarman: for me it gives the same 1 as the non-reversed version06:52
sitaram ManDay: I'll be back in about an hour; sorry...06:52
ManDay np06:52
Ivoz tos9: Actually, read it a bit more... didn't see that it's pulling plugins from git automatically. That's pretty cool06:53
tos9 Ivoz: Yep. The only module you touch is vundles the first time you clone it and then leave it to manage itself too.06:54
cirwin http://dpaste.org/8eXOi/ ^^ kevlarman ?06:54
← dfr|mac left06:58
→ dr_lepper joined06:58
← dr_lepper left06:58
→ dr_lepper joined06:58
kevlarman cirwin: weird06:59
same here06:59
→ harsh joined07:00
kevlarman i swear i've seen it work though07:00
cirwin maybe it was broken recently, I'm on current git.git master07:00
→ robotmay joined07:00
→ werdan7 joined07:03
→ drizzd joined07:05
← jwpeddle left07:05
syphar → syphar|away07:05
← d0k left07:06
lorenl what's the best way to compare two distinct git repositories? Should I first add the other one as a remote?07:08
cirwin lorenl: that's what I'd do07:09
then it's just a case of comparing stuff within one git repository, which is easier07:09
→ Daniel0108 joined07:09
← Daniel0108 left07:09
→ Daniel0108 joined07:09
← henriquev left07:16
← chrisf_ left07:16
← ManDay left07:19
← shiba_yu36 left07:19
← nyuszika7h left07:20
← mlncn-agaric left07:22
← kingfishr left07:25
← dr_lepper left07:27
→ kingfishr joined07:31
← harsh left07:36
→ nd___ joined07:37
→ ManDay joined07:38
ManDay Which manpage describes the syntax/notation used to describe commits or ranges thereof?07:39
cirwin man gitrevisions07:39
← dim_ left07:39
← nd__ left07:40
ManDay very good!07:40
thanks cirwin07:40
cirwin np07:40
← cirwin left07:42
→ gnufied joined07:44
← bolden left07:45
← haydenmuhl left07:45
→ Squarism joined07:46
← kmc left07:46
→ johnkpaul joined07:46
← gnufied left07:46
← Cromulent left07:47
← topeak left07:49
→ kmc joined07:49
→ shiba_yu36 joined07:50
← johnkpaul left07:51
← towski left07:51
→ chrisf_ joined07:51
→ jutaro joined07:53
→ topeak joined07:53
← LiohAu left07:54
→ soc42 joined07:55
← shiba_yu36 left07:55
→ dvaske joined07:58
→ workmad3 joined07:59
← ManDay left08:00
← bgerber left08:03
→ stodan joined08:04
← pharvey left08:06
→ bgerber joined08:06
→ tobago joined08:07
← cjs left08:09
← bgerber left08:11
← makubi left08:13
→ makubi joined08:13
← zamabe left08:13
← flavius left08:13
→ ManDay joined08:14
→ airborn joined08:14
← Squarism left08:15
← tewecske left08:17
ManDay Guys imagine the following situation: A repository with a complicated history. I want to rewrite the history in a manner that the result are MULTIPLE repositories which do not share history. Example: Let the directories be named with letters according to depth in the tree, postfixed with numbers to distinguish them within one parent. The new repositories shall be:08:18
First repository: A (the original root of the original repository), but without A/B1/C1...C5 (the history and everything associated to those contents of B1 shall not be in the history of the first repository).08:18
Second repository: A/B1/C1 (the stuff which has been excluded from the first repository). Of course, only the history relevant to that C1 shall be recorded.08:18
Third repository: A/B1/C2. And so forth.08:18
← ManDay left08:18
→ ManDay joined08:19
ManDay Guys imagine the following situation: A repository with a complicated history. I want to rewrite the history in a manner that the result are MULTIPLE repositories which do not share history. Example: Let the directories be named with letters according to depth in the tree, postfixed with numbers to distinguish them within one parent. The new repositories shall be:08:19
First repository: A (the original root of the original repository), but without A/B1/C1...C5 (the history and everything associated to those contents of B1 shall not be in the history of the first repository).08:19
Second repository: A/B1/C1 (the stuff which has been excluded from the first repository). Of course, only the history relevant to that C1 shall be recorded.08:19
Third repository: A/B1/C2. And so forth.08:19
Is there anything besides filter-branch that I would need to untangle the history based upon directories, so that each respective repository has no more notion of those other directories and history thereof?08:19
→ thiago joined08:20
← robotmay left08:27
→ miguet joined08:29
ManDay ping?08:30
→ bgerber joined08:30
thiago who are you pinging?08:30
ManDay anyone. i have problems with the atheros card i thought i was disconned08:31
since obviously someone must have had replied by now :P08:31
syphar|away → syphar08:34
← m4dc0d3r left08:35
sitaram ManDay: I think filter-branch is the only thing that you need. Unless there are lots of renames crossing your new boundaries, it should even be fairly easy08:36
→ j416 joined08:36
← warthog9 left08:36
← syphar left08:36
→ warthog9 joined08:37
← drev1 left08:37
← caseymcg left08:38
ManDay sitaram: Thank you, gives me confidence I can do it :)08:38
→ awallin joined08:39
← drizzd left08:41
→ mlncn joined08:42
sitaram ManDay: gfb (short for git-filter-branch) is picky about its arguments, and you should read the entire manpage before trying it08:43
for example, what happens to tags in your case you need to decide...08:43
you also need to decide if the repo containing A/B1/C2 should have A and A/B1 etc or simply start at the contents of C208:43
expect to spend some time, dont panic and it will all work out08:44
anyway its git... clone it somewhere and work on that if you like08:44
→ shiba_yu36 joined08:45
ManDay sitaram: Yes, I planned to be thorough with it08:45
Thanks for the heads up08:45
→ Lemon|mbp joined08:46
sitaram you're welcome!08:46
ManDay I've a questing regarding mv - how does GIT figure out that I moved a file if I do so with my systems mv? How does Git distinguish between one of the files being deleted and the other created and the rename, if I don't explicitly let git handle it with git-mv?08:47
thiago ManDay: it compares the contents08:47
sitaram ManDay: even the so-called explicit command is just a convenience front; internally it's the same as a separate add and rm08:47
thiago ManDay: if it's exactly the same or similar, it concludes it's a move08:47
ManDay Aha. So after all the semantics aren't meant to matter to GIT at all - it just seeks the most efficient way to track the changes, is that right?08:48
(the changes in the tree)08:48
soc42 hi #git08:49
ManDay In other words: One may assume that no renames take place. It is only that *if* I actually renamed a file it is likely that GIT's preferred way of tracking the changes is saving the change as a rename, but still semantically presenting it as a delete and a create.08:49
sitaram ManDay: "saving the change as a rename" does not occur in git08:50
→ d3vic3 joined08:50
ManDay sitaram: It does not save the changes as a complete diff (meaning saving the content of the file which is "created")08:51
sitaram ManDay: look at it like this. "git mv a b" results in an *identical* git repo to "cp a b; git rm a; git add b"08:51
ManDay sitaram: Yes I know.08:51
← d3vic3 left08:51
→ d3vic3 joined08:51
sitaram ManDay: since a and ba have the same *content* they have the same SHA, so no diffing happens. A SHA changes in the tree object that's all08:52
ManDay I just meant: One can not rely on GIT faithfully representing the semantics of history (if I actually delete and create a file, instead of renaming the former, Git will store it as a rename) - so one may equally say GIT has only one semantic presentation of the change, though it has several means to save it.08:52
sitaram excuse me, a filename changes in the tree object08:52
thiago if you delete and re-add a different file with the same name, Git may not realise it's a new file either08:53
ManDay thiago: Yes. But I spoke of deleting one file, creating another with the same contents and a different name.08:53
thiago by the same token, if you modify a file too much, Git may conclude it is a rewrite08:53
ManDay Git will "falsely" take it as a rename.08:53
thiago: Yes. We are clear on what happens.08:53
I just wanted to stress that from the user perspective there is no reliable way to distinguish between renames and creating in the history, hence one should assume there is only one of them (necessarily the delete-and-create-type of change)08:54
→ mabrand joined08:55
→ gnufied joined08:56
← mabrand left08:56
→ mabrand joined08:57
← mabrand left08:57
→ Chib joined08:57
wereHamster ManDay: git tracks content, not files. If you delete one file and create another one with the same content, it's move (rename) from gits perspective. The content moved from one file to another08:57
→ Vile joined08:57
sitaram ManDay: you are right there is only one way git looks at it, but I dont see why the user should care that it is not being recorded as 2 separate activities08:58
← topeak left08:58
ManDay wereHamster: I don't grasp that.08:59
wereHamster: What does "tracking contents" mean, if not "tracking contents of files"?09:00
wereHamster ManDay: contents of files09:00
ManDay Sounds like a motto - not really technical09:00
wereHamster you said: 'One can not rely on GIT faithfully representing the semantics of history'. What do you mean by 'semantics of history'?09:00
ManDay wereHamster: That's what I meant by two technical means to track.09:00
wereHamster: Okay, let me elaborate again:09:01
wereHamster I didn't follow the conversation too much, I just got up09:01
→ _aeris_ joined09:01
→ jigal joined09:03
wereHamster git does not even tracks 'delete' and 'create' if you want to be strict. Git only tracks: the tree looked like this in commit A, and like this in commit B. Git then infers the changes by comparing the two trees09:04
→ m4dc0d3r joined09:04
sitaram ManDay: you MUST google for and read 'git from the bottom up' -- you won't regret it, I promise09:04
wereHamster it does not track changes. it tracks states (trees)09:04
ManDay I say that GIT has two technichal way to track the changes: One is patch'ing the changes between two commits of one file and the other is moving the file (and possibly additionaly patching it). It employs those means in the most efficient way. Meaning if the "distance" (as in levenshtein distance) between two commits is smaller with either of that methods, it will take that method to track the change. One09:05
may plausibly called a "rename", the other an "edit". However, the method which it chooses does not directly relate to the semantics of the change, as the user did it - meaning depending on the exact structure of the change an "edit" might be tracked as a "rename" and vice versa. Hence I say the user shall not rely on GIT faithfully restoring the semantics and should generally assume that GIT tracks the09:05
changes by the "patch"/"edit" method and only internally chooses the most efficient way to store that change.09:05
→ apocalyptiq joined09:05
ManDay Do you agree with that sitaram wereHamster ?09:05
wereHamster ManDay: as I said before, git does not track changes. What you described could be seen as two possible means to *display* changes between two trees09:06
and assuming that, yes, I think it's an acurate description..09:07
→ nazgul101 joined09:07
← kingfishr left09:07
wereHamster ManDay: practical question: when would the user prefer to record a rename not as a rename but as an delete/add?09:08
→ harsh joined09:09
ManDay wereHamster: The user musnt care.09:11
sitaram ManDay: I dont quite agree. The object *model* is of complete objects. The delta, Lev distance, and such is implementation and quite irrelevant to the concepts a git user needs09:11
ManDay He should assume that everything is recorded as a delete/add, to make it easy for himself to reason.09:11
→ Squarism joined09:11
sitaram ManDay: they only become factors when running say 'git repack' with custom window/depth, otherwise no09:11
← Daniel0108 left09:12
wereHamster ManDay: easy, use -M100% in git commands09:13
then git won't display any renames09:13
and -C100%09:13
ManDay wereHamster: ?09:14
← asakura left09:14
ManDay To which command?09:14
wereHamster actually, log doesn't display renames by default, you have to enable rename detection with -M09:15
ManDay wereHamster: So what is the advantage for the user two distinguish between renamed and add/deleted?09:16
s/two/too09:16
s/too/to09:16
→ asakura joined09:17
← f0i left09:17
wereHamster less cognitive load when reviewing code. If I see a rename, I don't have to review the code, if I see an add and delete, I have to review the whole added file.09:17
ManDay True.09:18
iband → iband|away09:20
wereHamster ManDay: if you say that the user mustn't care about rename vs add/delete. The SCM has to choose one representation. Which one should it use?09:22
→ ouah joined09:23
wereHamster and it has to choose one way to store it in history. Which one should it use there?09:23
→ Rvl joined09:23
← workmad3 left09:24
ManDay wereHamster: I did just elaborate on that!09:24
I said the SCM uses the *technical* representation which is most efficient.09:24
jigal hell can i get some help with this error : jigal@jigal-laptop:/var/www/nrka3$ git commit -a Error reading /home/jigal/.nano_history: Permission denied09:24
ManDay Which for an ordinary rename is a rename and for a normal delete/add is a delete/add. wereHamster09:24
However, if you perform a delete/add which resembles a rename, it will choose a rename.09:25
And if you rename and edit the file a lot, it will choose a delete/add.09:25
wereHamster ManDay: let's not debate what is technically more efficient. Because renames in git, even if not stored as such, are pretty damn efficient.09:26
→ flijten joined09:26
ManDay wereHamster: Perhaps. I assumed if NOT stored as a rename, it will store the diff for each of the individual files.09:26
wereHamster anyway, there are drawbacks when storing a rename each way. Linus wrote a nice email about it once. I'll try to dig it up09:26
→ field_it joined09:27
wereHamster actually, there are only drawbacks if you store it explicitely inside history :)09:27
← madsy left09:28
← Liquid-Silence left09:28
→ jj- joined09:28
→ Liquid-Silence joined09:28
← gnufied left09:29
wereHamster ManDay: here it is: http://article.gmane.org/gmane.comp.version-control.git/21709:29
→ hsingh joined09:29
← Rvl left09:30
jigal hell can i get some help with this error : jigal@jigal-laptop:/var/www/nrka3$ git commit -a Error reading /home/jigal/.nano_history: Permission denied09:30
thiago jigal: not a git error09:30
looks like your editor is nano and it doesn't work09:30
iband|away → iband09:31
jigal thiago, but it happend after i do i do a git commit09:32
thiago jigal: git commit wants to run the editor to get a commit message09:32
jigal ok09:32
thiago jigal: does the same problem happen with git commit -a -m "Message here" ?09:32
ouah are HEAD^ and ORIG_HEAD always the same?09:32
thiago ouah: no09:32
← Nedly left09:32
→ dim_ joined09:33
thiago ouah: commit will cause that to happen, but most other commands will have different results.09:33
ouah so I don't understand them::(09:33
← defendguin left09:33
thiago ORIG_HEAD is what HEAD was before the last operation09:33
HEAD^ is the first parent of the current HEAD09:34
ORIG_HEAD is equivalent to HEAD@{1}09:34
→ cjs joined09:34
← jutaro left09:36
ouah ok thanks09:36
→ f0i joined09:36
← Matth1a31 left09:36
→ systemclient joined09:36
systemclient should I tag my stuff with "1.4.5" or "v1.4.5"?09:37
thiago systemclient: whatever you prefer09:37
systemclient thiago: I always see the former style everywhere, although I think the latter says more …09:37
jigal what do i need to do after a commit? to get it to my online repository?09:37
systemclient k, than I just use my style09:37
jigal: like github?09:37
jigal: usually a push09:37
thiago jigal: push it09:37
jigal systemclient, yes github how to push it09:38
push origin master?09:38
thiago jigal: that would do it, yes09:38
jigal cool it worked09:38
thiago, can i ask a different question?09:38
thiago jigal: that was already a different question :-)09:39
jigal lol09:39
thiago, i have setup a git repositor on my laptop. Now i want to check it out from my pc and be able to commit from my pc. How do i have to setup my pc in terms of the certificate and so on? Do i need a different certificate to commit?09:40
thiago certificate?09:40
I'd simply suggest you perform the same operations: clone from the same original repository, then start working09:40
← cjs left09:41
← dexter_e left09:41
jigal thiago, but how do i get the rights to commit to that repository?09:42
thiago you mean "right to push to that repository"09:43
you can always commit in the clones you create -- because you created them09:43
to push, you need to identify yourself in a way that the server will recognise and authorise you09:43
jigal thiago, yes the right to push09:43
but do i need to download the ceftificate or so?09:44
thiago one way is to copy the private SSH key. I actually recommend you generate a new key in the old PC and upload the public key to the server.09:44
jigal so then i so to say act like two different users thiago09:44
→ JamesLeeds joined09:44
← shiba_yu36 left09:44
→ Matth1a3 joined09:44
← JorgeRuiz left09:45
← f0i left09:45
← harsh left09:46
→ Dave^| joined09:46
iband → iband|away09:46
← systemclient left09:47
iband|away → iband09:48
→ sbell joined09:48
→ rendar joined09:49
→ brian_g joined09:50
← iband left09:53
→ gnufied joined09:54
→ stoffus joined09:54
← chrisf_ left09:55
→ sbell_ joined09:57
→ cesc joined09:59
← sbell left09:59
sbell_ → sbell09:59
→ f0i joined09:59
→ syphar joined10:01
syphar → syphar|away10:01
syphar|away → syphar10:02
→ Ben1980 joined10:02
→ dnivra joined10:03
→ Deesl joined10:05
← dnivra left10:07
← f0i left10:08
← gnufied left10:09
syphar → syphar|away10:10
← kennethreitz left10:15
→ jutaro joined10:17
← jutaro left10:19
← soc42 left10:22
→ f0i joined10:22
→ seivan joined10:24
← stoffus left10:29
→ stoffus joined10:29
→ adnasa joined10:30
← albel727 left10:30
→ shiba_yu36 joined10:30
← f0i left10:32
→ ludde joined10:33
← shiba_yu36 left10:34
← alnewkirk left10:36
← adnasa left10:37
← aminpy left10:39
→ aminpy joined10:39
→ albel727 joined10:40
→ f0i joined10:45
→ malumalu joined10:47
← hsingh left10:49
ManDay can you see that picture? http://progit.org/figures/ch3/18333fig0313-tn.png10:50
cbreak it looks like a valid png10:51
ManDay ok my question is what is the difference between being on master and merging hotfix and the other way arround?10:51
(id est being on hotfix, merging master)10:52
wereHamster being on hotfix and merging master is a noop10:52
ManDay i see10:52
thanks10:52
wereHamster it will tell you 'already up to date' because there is nothing to do10:52
→ pantsman joined10:53
← pantsman left10:53
→ pantsman joined10:53
cbreak if you mean being on hotfix and merging issblah and vice versa: the difference is in which branch the commit ends up, and the order of the parents10:53
← dvaske left10:54
← shennyg left10:54
ManDay cbreak: with regard to your scenario: in this case http://progit.org/figures/ch3/18333fig0314-tn.png the order doesnt matter, correct?10:54
cbreak it does10:55
ManDay (merging iss53 with master=hotfix or the other way arround)10:55
what will be the difference ?10:55
cbreak if you mean being on hotfix and merging issblah and vice versa: the difference is in which branch the commit ends up, and the order of the parents10:55
wereHamster he already explained10:55
cbreak same as before10:55
ManDay i dont understand10:55
assume everything is committed10:56
cbreak if you are on the hotfix branch10:56
→ cjs joined10:56
cbreak and you merge the other... in which branch is the merge commit?10:56
← j416 left10:56
ManDay in both10:56
wereHamster *bzzzt* wrong10:56
cbreak in the current branch! Because git merge always changes the current branch and NO OTHER10:56
ManDay if I merge here, http://progit.org/figures/ch3/18333fig0314-tn.png, doesnt the result look like a diamond arround the c2,c4,c3 and the merge commit (c5) ?10:57
no matter, which branch i am on when I do the merge?10:57
wereHamster merwe what10:57
merge10:57
ManDay c3 and c4. merge hotfix=master with iss5310:58
wereHamster .. into what10:58
checkout master; merge iss53 or the other way around?10:58
cbreak ManDay: imagine you are on a branch10:58
ManDay with. i said "with". which implies symmetry - doesnt tit?10:58
cbreak let's call it HEAD10:58
wereHamster no10:58
cbreak and you make a commit10:58
in which branch is the commit?10:58
wereHamster ManDay: a git commit can have multiple parents. There has to be a first and second parent.10:59
cbreak (I am trying to guide him to the answer like a teacher)10:59
cbreak searches for a bamboo cane10:59
wereHamster cbreak: it doesn't look like you are a good teacher :P10:59
← f0i left10:59
cbreak :(10:59
cbreak searches for an iron pole11:00
cbreak So Mr. ManDay, we're waiting for an answer!11:00
if you make a commit, in which branch does it end up?11:00
wereHamster raises his hand11:00
wereHamster I know the answer. pick me mr. teacher.11:00
albel727 me too me too!11:01
cbreak I know you know the answer. But he has to solve the problem on his own11:01
albel727 jumps and waves his hand wildly.11:01
cbreak It's a basic concept in git11:01
ManDay cbreak: I've found an online sketchpad which I've been sketching on to illustrate my assumption11:02
cbreak alright, wereHamster, you may answer.11:02
ManDay just to find out it doesnt save11:02
cbreak the answer is one word11:02
ManDay HEAD11:02
of course11:02
cbreak yes!11:02
ManDay (though HEAD is a tag, not a branch, eh)11:02
cbreak no11:02
wereHamster no, HEAD is not a tag11:02
cbreak it's a symbolic ref11:02
ManDay well, yes.11:03
cbreak it usually points to the current branch11:03
ManDay ok, but i knew that11:03
cbreak ok, so you know where fresh commits end up11:03
so what happens if the commit has two parents?11:03
(if it is a merge commit). Where does it end up?11:03
ManDay cbreak: look11:04
maybe if you would correct my assumption which is:11:04
wereHamster cbreak: that approach doesn't really work with certain people :)11:04
cbreak hmmm :(11:04
→ shiba_yu36 joined11:04
cbreak maybe it's just my brain that works with deductive and inductive reasoning based on similarity assumptions11:04
ManDay if I merge in the following sit http://progit.org/figures/ch3/18333fig0314-tn.png , i assume that i get a diamon shaped form (ending in the newly created commit C5) - are we clear on that?11:04
← Deesl left11:04
→ soulcake joined11:05
ManDay cbreak: you are simply adressing something that is not my problem in understanding11:05
cbreak ManDay: yes... but there are some differences11:05
wereHamster yes, the shape is a diamond11:05
ManDay cbreak: ok, i was not yet finished:11:05
And do we agree that then, all branches point to C5 ?11:05
(because i do so)11:05
wereHamster no, we don't11:05
ManDay i see. so only the branch which I have been in while merging will point to C5 i assume?11:06
wereHamster .. which is what cbreak was trying to explain to you11:06
ManDay so it's a special case of fast-fwd merges that both branches than point to the same commit11:06
cbreak the merge commit will only end up in HEAD11:06
no11:06
ManDay bah11:06
wereHamster no, git well *never* modify anything else than the current branch11:07
cbreak almost every command not explicitly for changing arbitrary refs works only on HEAD11:07
ManDay i see11:07
→ SimonNa joined11:07
ManDay now i understood11:07
thanks cbreak wereHamster11:07
cbreak also: do you think c5 is the same wether made on c3 or c4?11:07
ManDay you mean by the data it holds?11:08
yes. it references two parents and specifies how they are merged11:08
cbreak no11:08
ManDay (it only belongs to two different branches in the two cases)11:08
cbreak incorrect assumption11:08
ManDay ok11:08
→ hsingh joined11:08
cbreak you know a merge commit is a commit with two parents, right?11:08
ManDay yes11:09
thats what I said11:09
cbreak how are the parents stored?11:09
ManDay as trees11:09
cbreak no11:09
ManDay (commits which point to trees)11:09
jeez...11:09
cbreak inside a commit object, they are stored with their hash11:09
but... in which order?11:10
ManDay http://progit.org/book/ch3-1.html11:10
cbreak this is quite important for history traversal, because, as you surely know, HEAD~3 for example means: follow the first parent three times11:11
ManDay cbreak: I didnt know that11:11
cbreak: I dont know in which order11:11
cbreak the current HEAD commit is parent 111:11
ManDay By the way, when you first asked me, I assumed you were referring "how the parents are stored in git" , not "in the merge" , as you meant11:12
cbreak all other merge heads are listed in the order in which you mention them on the cli11:12
← Squarism left11:12
cbreak so for example, if I am on c0, and I type git merge c2 c1 foo11:12
ManDay so the first HEAD is that which has been previously the HEAD of the branch?11:12
cbreak then the resulting merge has heads c0, c2, c1 and foo11:12
ManDay yes.11:12
why does the order matter?11:13
cbreak this is quite important for history traversal, because, as you surely know, HEAD~3 for example means: follow the first parent three times11:13
→ bonhoffer joined11:13
cbreak for example, if you work with a "only merges in master" workflow, you can just follow the first heads to see all merge commits11:13
and ignore all real commits in the second parent11:13
→ kingfishr joined11:14
ManDay hm - so its just a convenience - technically the order has no impact on the merge, right?11:14
cbreak the resulting tree is equivalent for most algorithms11:15
BUT the resulting DAG is not11:15
history traversal is important, so the order of parents is important as well11:16
ManDay topologically the DAG will be the same. and to all what matters (technically, again) it will also be equivalent11:16
cbreak no11:16
→ f0i joined11:16
cbreak see above11:17
→ tewecske joined11:17
ManDay you mean it matters for history traversal? yes, but only as a convenience, that's what i mean11:18
by "it doesn't really matter"11:18
are we clear?11:18
→ hafos joined11:18
cbreak no11:20
not for convenience11:20
for actual effect11:20
think of git bisect11:20
think of git rebase or git filter-branch11:20
→ martinjlowm joined11:21
→ dvaske joined11:21
→ toabctl joined11:23
→ w0bni joined11:23
ManDay but what if you don't want to traverse along the "primary inheritance line"?11:24
I assume there must be a way to specify that, too?11:24
← ouah left11:27
→ gnufied joined11:28
← gnufied left11:28
← f0i left11:31
→ Raging_Hog joined11:31
← dvaske left11:33
→ alnewkirk joined11:33
→ gnufied joined11:33
→ ruskie joined11:34
← bittin left11:35
→ Rvl joined11:36
cbreak ManDay: there's the ref^parentnum syntax, but it only goes one parent far11:38
so it's more cumbersome to use11:38
ManDay thank you cbreak11:39
cbreak np11:40
→ johnkpaul joined11:41
← jimmy1980 left11:43
→ zamabe joined11:43
→ f0i joined11:44
→ ouah joined11:49
→ jimmy1980 joined11:51
← johnkpaul left11:51
← shiba_yu36 left11:52
← f0i left11:54
← bonhoffer left11:55
→ f0i joined11:56
→ harsh joined11:59
→ chungwen joined12:01
← chungwen left12:01
→ sythe joined12:02
← sythe left12:02
→ sythe joined12:02
← JamesLeeds left12:02
← tos9 left12:03
→ nyuszika7h joined12:04
← f0i left12:07
← mattdipasquale left12:07
→ mattdipasquale joined12:08
→ the_metalgamer joined12:09
→ f00li5h joined12:10
← cesc left12:10
← Rvl left12:13
→ bittin joined12:14
→ f0i joined12:16
→ Drakonite joined12:16
← bittin left12:19
→ bittin joined12:19
→ subbyyy joined12:19
→ flaguy48 joined12:20
← stoffus left12:20
→ Sajaki joined12:21
← kingfishr left12:21
Sajaki hi, i'm on osX, and when i change the case of a filename, Git doesn't see it as a change. how to fix?12:22
← rchavik left12:23
rudi_s Hi. Where is the current git repository located? git.kernel.org doesn't work for me.12:24
Sajaki http://git-scm.com/12:24
← f0i left12:25
Sajaki kernel.org is down for a few days now12:25
rudi_s Sajaki: That points to git.kernel.org, is there no mirror. I'm just asking because /topic talks about 1.7.6.3 and I'd like to get it (git-scm.com is outdated btw. it mentions 1.7.6.1).12:26
→ logii joined12:27
logii I'm getting this error: No submodule mapping found in .gitmodules for path 'sites/all/modules/admin_menu'12:28
when i do git submodule status12:28
i think the problem arises from the fact that i actually used clone, instead of submodule add12:29
→ pragma_ joined12:29
cbreak Sajaki: rename to something else and then back12:30
logii is there any command for me to update the remotes of the submodules, from the main repo?12:30
cbreak logii: man git-submodule foreach, try that12:30
← harsh left12:33
logii cbreak: cool! but what am i to do after submodule foreach? update?12:34
cbreak no12:34
logii cbreak: it keeps saying: No submodule mapping found in .gitmodules for path 'sites/all/modules/admin_menu'12:34
cbreak git remote update probably12:34
but why do you want to do that anyway?12:34
submodules will be automatically updated with git submodule update12:35
logii cbreak: I think they are not recognized as submodule in the main repo?12:36
→ loic_m joined12:36
cbreak why not?12:36
if they are submodules then they are12:36
logii they were not added as submodule, those repo were cloned12:37
pragma_ How does one get a git log of commits made since the repo was checked out?12:38
→ Praise joined12:38
cbreak then they aren't submodules12:38
← awallin left12:38
cbreak pragma_: repositories aren't checked out12:38
logii is there anything available for me to convert them to submodules?12:38
→ awallin joined12:38
cbreak reformulate your question12:38
logii: git submodule add12:38
but ... read up on what submodules are!12:38
you seem very clueless at the moment12:39
and it might not be the case that submodules are the solution to your problem12:39
thiago pragma_: you need to know the SHA-1 of the time of the clone12:39
pragma_: this might work though: git log @{10.years.ago}..12:40
pragma_ I mean, suppose I have a working directory and I want to see what others have committed since then12:41
cbreak then just do a git remote update12:41
pragma_ ie, what changes I'm missing12:41
cbreak and git diff ..origin/master or what ever12:41
or git log ..origin/master12:42
ouah pragma_: git fetch and then git log master..origin/master12:42
→ f0i joined12:42
← zamabe left12:43
pragma_ I get "fatal: ambiguous argument: unknown revision or path not in working tree12:43
wereHamster pragma_: paste the complete command12:44
→ Vortex34 joined12:44
← Vortex35 left12:46
→ Bass10 joined12:48
→ anaio joined12:49
← f0i left12:50
← alnewkirk left12:51
anaio → alnewkirk12:51
ManDay tango_: the progit.org is really good!12:54
→ caruso_g joined12:55
→ Chillance joined12:55
← d3vic3 left12:59
→ _iron joined13:05
→ f0i joined13:09
→ yeban joined13:10
← Bass10 left13:10
→ jbrokc joined13:12
← yeban left13:13
← flijten left13:14
→ bonhoffer joined13:15
← catsup left13:16
→ catsup joined13:17
→ madewokherd joined13:18
→ goshakkk joined13:21
→ w09x joined13:22
→ johnkpaul joined13:22
← bonhoffer left13:23
← f0i left13:23
← jimmy1980 left13:25
← cjs left13:25
→ Cipher-0 joined13:25
sie If I just want a copy of the project from the bare repo for apache and other services, then I should fetch, instead of pulling, right?13:26
→ JamesLeeds joined13:26
rudi_s sie: If you don't have a repository local, you use clone. fetch/pull is used when you already have a clone of a remote repository and want to get new commits.13:26
sie I do.13:27
I want to update it.13:27
← mattdipasquale left13:27
rudi_s sie: Then git fetch to get the new changes, git pull is git fetch plus git merge.13:27
sie Fetch cannot fail, right?13:27
cbreak sure it can13:27
rudi_s sie: But it won't corrupt your local state.13:28
sie because of conflicts and such, that is?13:28
rudi_s No.13:28
It just stores the new commits in your repository and updates the remote branches.13:28
(Like origin/master.)13:28
→ jimmy1980 joined13:31
← pantsman left13:32
sie Hmm, I did git fetch, but I see no changes in the files. Does it store the changes only in .git or something?13:33
wereHamster yes13:34
m1sc sie: fetch doesn't touch you objects, it's just about getting remotes objects13:34
your13:34
sie Okay, so how can I fetch & merge(or pull) overwriting ALL THE THINGS and such?13:36
wereHamster git pull13:36
sie I have a `error: The following untracked working tree files would be overwritten by merge:13:36
wereHamster sie: oh wait. git fetch && git reset --hard @{u}13:37
← martinjlowm left13:37
→ mattdipasquale joined13:37
→ martinjlowm joined13:38
sie just ... '@{u}'?13:38
← tdebat left13:39
→ tdebat joined13:40
sie wereHamster - ^13:42
→ f0i joined13:42
wereHamster ...13:42
← jimmy1980 left13:42
→ iband joined13:44
← tdebat left13:45
→ tdebat joined13:45
bremner sie: man git-rev-parse13:48
← tdebat left13:49
← jonshea left13:50
← f0i left13:51
→ tdebat joined13:52
← caruso_g left13:52
← lacrymology left13:53
← toabctl left13:54
← tdebat left13:54
→ lacrymology joined13:55
→ ISF joined13:55
→ Octalot joined13:56
→ kukks joined13:56
→ keyzs joined13:57
keyzs http://www.thevenusproject.com/13:57
← keyzs left13:57
→ possibilities joined13:57
← airborn left13:57
← apocalyptiq left13:58
→ CannedCorn joined13:59
→ Prasad joined14:00
→ Cromulent joined14:00
← Lemon|mbp left14:03
→ MUILTFN joined14:04
→ f0i joined14:05
→ cjs joined14:06
→ pantsman joined14:07
← pantsman left14:07
→ pantsman joined14:07
→ ehsan joined14:07
→ toabctl joined14:08
← f0i left14:14
← Cromulent left14:15
→ tdebat joined14:15
→ shiba_yu36 joined14:15
← tdebat left14:16
→ heppy joined14:17
→ Phylock joined14:17
→ sorin joined14:19
→ Lisimba2 joined14:19
← heppy left14:19
→ bonhoffer joined14:20
→ sattu94 joined14:20
iband → iband|away14:20
→ jonshea joined14:21
← jonshea left14:22
← Lisimba left14:22
→ stealthmk joined14:27
← seivan left14:27
→ sivy joined14:28
→ f0i joined14:28
← Ben1980 left14:29
← EricInBNE left14:29
← w09x left14:29
→ Lemon|mbp joined14:30
← the_metalgamer left14:30
← bonhoffer left14:30
→ heppy joined14:32
← fisted left14:32
← heppy left14:33
→ w09x joined14:33
← w09x left14:33
← gnufied left14:33
→ kvanderw joined14:34
← f0i left14:37
→ Daniel0108 joined14:38
→ jogla joined14:38
→ aalex joined14:42
→ fisted joined14:42
← pigeons left14:43
→ the_metalgamer joined14:43
← sorin left14:45
← goshakkk left14:46
→ gusnan joined14:46
→ JasCo joined14:46
→ jhoepken joined14:47
→ codebeaker joined14:47
→ apocalyptiq joined14:47
→ diegoviola joined14:49
→ f0i joined14:51
← f0i left14:51
← normanrichards left14:51
← Lemon|mbp left14:54
→ Lemon|mbp_ joined14:55
→ posciak joined14:56
← stodan left14:57
→ bloopletech joined14:57
→ bitkiller joined14:57
← brian_g left14:57
bloopletech Is there way to get all the changes for a specific part of a file? I want to git log -p a file, but I only care about a small part of a large file14:57
→ brian_g joined14:58
← Cipher-0 left14:58
← Chib left14:59
→ harsh joined14:59
wereHamster only with git blame15:00
→ setmeaway joined15:00
bloopletech wereHamster: can that give me the history as opposed to just the last change on the line?15:02
wereHamster no, but it's the closest you can get15:02
you'll have to drill down through the history manually15:03
→ pen joined15:03
bloopletech aw I hate it when git isn't magic. Oh well thanks15:03
cybersphinx bloopletech: You can use "tig blame file", and then "," to go to the previous commit of the selected line.15:05
→ Deesl joined15:06
bloopletech cybersphinx: ooh I'll check it out15:06
→ anaio joined15:06
← mattdipasquale left15:07
← anaio left15:08
EugeneKaway → EugeneKay15:08
→ masando joined15:08
← jbrokc left15:08
← alnewkirk left15:09
→ mattdipasquale joined15:09
iband|away → iband15:10
bloopletech cybersphinx: tig looks friggin awesome, thanks for the tip15:11
→ replore_ joined15:12
→ dbpolito_ joined15:13
← stealthmk left15:13
→ sorin joined15:15
← codebeaker left15:15
→ dfr|mac joined15:16
→ ch077179 joined15:17
→ _steve joined15:17
→ Swimming_Bird joined15:19
→ flavius joined15:20
flavius → Guest708315:20
← thews left15:20
→ thews joined15:22
← lacrymology left15:22
→ macmartine joined15:23
← Guest7083 left15:24
→ Guest7083 joined15:24
ChanServ set mode: +v15:24
_steve hi all. just converted our hg project to git using fast-export. have installed git for windows & tortoise git. what do I need to do to tell tortoise git where my repository is & where my current working directory is ?15:25
Guest7083 → flavius15:25
← macmartine left15:25
cbreak don't use tortoise :/15:26
but if you want, just use the context menu15:26
EugeneKay Tortoise Git is a shell extension. You open the directory and right-click. I hate it.15:27
_steve used to svn so works for me :)15:27
← pantsman left15:27
← mino left15:27
_steve get a context menu if I right click over the repository directory.15:27
jaeckel _steve: stop using it :) you will be happier when you're used to the commandline!15:28
→ Yuuhi joined15:28
EugeneKay The .git folder? Leave that alone.15:28
cmyers at my company, our recommendations for git on windows are git bash (msysgit) for CLI, and Visual Git Extensions for gui.15:28
→ Nedly joined15:28
_steve the git folder. that's where the version control data is.15:28
how do I tell the system it's there or does it 'just know'15:29
EugeneKay Git operations re done on the working tree, not on the .git dir15:29
← jogla left15:29
→ goshakkk joined15:29
_steve sure, but it needs to know where the repo is surely ?15:29
EugeneKay Git looks in the current folder, then in every parent folder, for a .git dir. If it doesn' find one, it fails.15:29
← dbpolito_ left15:29
wereHamster _steve: it should 'just know'15:29
well, git. Don't know about tortoisegi15:30
EugeneKay I use SmartGit. It suits me well, but it's not FOSS. I'm OK with this.15:30
_steve hmmm so I have drive/repo/.git & drive/dev/source ... will that config work ?15:30
← logii left15:30
EugeneKay No. The .git dir goes alongside your codes.15:31
cbreak _steve: if drive/repo/.git exists, then drive/repo is the repository15:31
_steve yes15:31
cbreak and everything inside it is working dir15:31
(with some exceptions)15:31
drive/dev/source is irelevant15:31
_steve so the working dir sits inside the repo ?15:31
← bgerber left15:31
wereHamster the working dir + the .git dir *is* the repo15:32
EugeneKay path/to/working/dir/.git/15:32
← brian_g left15:32
_steve so all my working direcories have .git subdirs or just the root ?15:32
*directories15:32
wereHamster just the root15:32
← harsh left15:32
_steve ok - we're still good.15:32
→ shtrb joined15:32
→ macmartine joined15:32
cbreak have you never used subversion?15:33
← Nedly left15:33
_steve so I need to position my .git dir in the root of my source tree ?15:33
EugeneKay Just dir/. A file at path/to/working/dir/foo/bar.txt will show up in you repo as being named "foo/bar.txt", wherease path/to/working/dir/README will simply be named "README"15:33
cbreak no15:33
git will do that itself when you git init15:33
cmyers _steve: how are you creating this git repo? You shouldn't have to position anything15:33
_steve it is a conversion of a hg repo.15:34
shtrb is there any limit of revition amount when I use git svn ? git crashes after getting the 1000th rev and I wish to get all the history (60K revisions)15:34
selckin shtrb: "crashes" ?15:34
wereHamster shtrb: shouldn't be. Why is git crashing?15:34
cmyers _steve: I have only done it a few times, but I believe you shoudl create a new git repo using `git init`15:34
then push into it from your hg repo.15:34
_steve done the conversion already, thanks.15:34
shtrb wereHamster , before saying that something have a bug I'm checking if this is not a feature15:34
selckin : exit code 1315:35
→ bgerber joined15:35
cmyers _steve: and it just created a .git dir, right?15:35
_steve yes15:35
wereHamster shtrb: any error messages?15:35
shtrb selckin , aka segmentation fault15:35
nope15:35
cmyers that is because you haven't checked out a branch then probably15:35
or it is a bare repo15:35
shtrb this is why I think it is a feature15:35
selckin shtrb: start with that last one in the future15:35
cmyers if you go to the directory which contains .git, and run `git checkout <branch>`, see what happens15:36
wereHamster shtrb: any idea which command causes that sigsegv?15:36
→ tpiep joined15:36
shtrb wereHamster, It works upto the 1000'th (checked 1 10 100 revisions works on this crashes)15:36
wereHamster , nope15:36
_steve will try that, thanks.15:36
→ brian_g joined15:36
wereHamster shtrb: IIRC git repacks every 1000 revisions, so it's maybe git-repack?15:36
← shiba_yu36 left15:37
→ ajpiano joined15:37
shtrb I knew it some feature ;-)15:38
it is ...15:38
← angelsl left15:38
shtrb how can I check if it's the $@$ problem (nothing in /var/log/syslog)15:38
← redglasses left15:40
→ cesc joined15:41
→ rpg joined15:41
→ SimonNaa joined15:41
→ henriquev joined15:42
shtrb can I continue clone after a crashed repack ? (if it is the issue)15:43
selckin git svn fetch15:43
_steve ok - ran git checkout master & worked as expected except seems top be an old version. had done some file renames in hg & they're not there.15:44
*to15:44
maybe should try the conversion again.15:44
shtrb selckin 10x I'll try15:45
selckin 10x?15:45
shtrb thank you15:45
_steve ran the fast-export script - is this the recommended method of conversion ?15:45
← SimonNa left15:45
→ Davey_ joined15:45
← Davey_ left15:45
→ Davey_ joined15:45
wereHamster _steve: yes15:45
_steve hmmmm.15:45
→ Cromulent joined15:45
→ eeanm joined15:46
_steve so I just ran checkout master from tortoise using its defaullt config. is there a better way to make sure I've got the very latest version in the repo ?15:46
cbreak git remote update15:47
checkout master just checks out the master branch15:48
it doesn't fetch anything from remotes15:48
_steve this is a local repo. I'm the only user.15:48
cbreak then it still doesn't check out the newest commit15:48
just the master branch's tip15:49
_steve so how do I get all the latest commits ?15:49
cbreak look at the branches15:49
find the branch with the latest commit15:49
and switch to it15:49
_steve noob q - how do I look at the branches.15:50
→ mino joined15:50
wereHamster with your eyes. Unless you are blind in which case you need a braille reader15:50
_steve I see the nerdiverse is alive and well.15:51
cmyers _steve: you want to see what branches there are?15:51
cbreak git branch shows branches15:51
cmyers or look at their history?15:51
→ d0k joined15:51
_steve thanks15:51
cbreak surprise!15:51
read the man page for flags15:51
git log shows history graphs, read the man page for details15:52
_steve everything is easy & obvious when you know how to do it.15:52
git branch says master15:52
selckin or if you read the most initial of documentation15:52
→ muneeb joined15:53
cbreak maybe you only have that branch...15:53
or you forgot to read the man page15:53
_steve ok - think I'm going to switch back to hg & make sure everything is commited properly there15:53
thought it was but could be user error.15:53
cmyers _steve: if you prefer a video tutorial, I suggest watchign this: http://www.youtube.com/watch?v=ZDR433b0HJY15:53
90 minutes well spent15:53
_steve ofc I haven't read any man pages - got you guys :)15:53
wereHamster we did'nt expect so15:54
cbreak what do you mean "committed properly"?15:54
do you want an other branch?15:54
or are you missing commits?15:54
_steve Some of my files on a basic repo are out of date after conversion.15:54
cbreak define "out of date"15:55
→ herdingcat joined15:55
_steve not the latest versions15:55
→ caruso_g joined15:55
_steve I'd renamed some files in hg so maybe that's confused the conversion15:55
cbreak how do you know?15:55
_steve because the file names are the old file names15:55
← caruso_g left15:55
← Yuuhi left15:55
cbreak is the commit with the renames in git?15:55
cmyers _steve: you are on windows, right?15:56
_steve yes15:56
cbreak or in other words: is the history there?15:56
cmyers did you change the case of any filenames?15:56
renaming files from like, Foo.java to foo.java, for example, causes big problems15:56
in hg and git.15:56
_steve didn't change case. file.hlsl to file.sh15:56
leave it with me. now I know the basics I can check me repo is behaving properly in hg then will try another conversion.15:57
worst case I could create from scratch & lose the history. wouldn't be the end of the world.15:57
← apocalyptiq left15:57
_steve thanks for your help guys.15:57
← SimonNaa left15:58
← shtrb left15:58
→ SimonNa joined15:58
_steve @cbreak - to be clear - the yes above is yes I'm on windows. haven't looked to see if the renames are in git yet.15:59
→ codetroll joined15:59
_steve checked one of the files - there's some history there but not all. best guess is fast-export doesn't handle renames properly but like I say, could be user error.16:01
→ Yuuhi joined16:01
← [1]codetroll left16:01
→ Dave^|| joined16:01
← aminpy left16:02
_steve ok - back to hg for now. will try again later. thanks for all your hard work on this. hope to use it later.16:02
→ defendguin joined16:03
→ f0i joined16:04
selckin try -C -M --follow16:04
← Dave^| left16:05
_steve @me ?16:05
selckin no this isn't twitter16:07
→ socketwiz joined16:07
→ Rvl joined16:07
← jhoepken left16:08
→ Targen joined16:08
← socketwiz left16:09
→ socketwiz joined16:09
_steve sorry. ok done for now. will leave you guys in peace. help appreciated. l8r.16:10
← _steve left16:10
→ harsh joined16:11
→ berserkr joined16:12
→ moops joined16:13
← ISF left16:15
→ dbpolito_ joined16:15
← Ivoz left16:16
→ drev1 joined16:16
← f0i left16:18
sorin Has anyone got an extensive article on the Git subtree merge strategy?16:19
← toabctl left16:19
→ dnivra joined16:20
← herdingcat left16:21
→ aalex-home_ joined16:21
← Targen left16:21
→ arunce_ joined16:22
← bloopletech left16:22
← arunce left16:23
ManDay Are git commit messages supposed to be line wrapped=16:23
?16:23
cbreak that's the job of the editor16:23
bremner ManDay: git does not mess with them, if that is your question16:23
cbreak or you16:23
→ kadoban joined16:23
ManDay my question was if by convention they are supposed to be16:24
sorin ManDay, the first line should be 58. Leave a blank line. Then write paragraphs at 78.16:25
cbreak in my repositories they have to16:25
ManDay sorin: That's what I meant16:25
cbreak around 80 chars wide16:25
it's not that strict16:25
ManDay My friggin VIM does line wrap everywhere where I dont want it but OF COURSE it does NOT with commit messages16:25
stupid thing16:25
sorin See https://github.com/blog/926-shiny-new-commit-styles16:26
→ DireFog joined16:27
sorin Sorry, I meant 50 then 72.16:27
← possibilities left16:28
← zivester_ left16:28
infid how is there no git pocket reference book on amazon16:29
DireFog I have a problem with filter-branch with a subdirectory filter. It will always come back with "error: Entry '.gitignore' not uptodate. Cannot merge.", and when I run an index-filter which "git-rm"'s files that are both in the new top-level directory and the old root, I get "error: Untracked working tree file 'CONTRIBUTORS' would be overwritten by merge."16:29
sorin infid, because the Git man pages are the size of a phone book?16:29
← moops left16:30
→ zamabe joined16:30
← harsh left16:30
cbreak DireFog: git status16:30
what does it show?16:30
FauxFaux infid: 'cos it's too easy, no reason to have a book!16:31
→ alnewkirk joined16:31
DireFog cbreak: AFAICT the entire old repo as new files, and a few modifieds16:31
cbreak it should show nothing at all to commit16:31
and nothing changed16:31
and nothing untracked16:31
DireFog do you mean before or after the failed filter run?16:32
because I'm starting with a fresh clone16:32
cbreak before you started16:32
← martinjlowm left16:32
→ martinjlowm joined16:32
→ metcalfc joined16:33
DireFog # On branch active_support16:33
nothing to commit (working directory clean)16:33
ManDay ping?16:33
wow, im still here16:33
→ Jon47 joined16:33
← flavius left16:33
← rpg left16:33
← Tabmow left16:33
thiago ManDay: yes, your wireless connection is still working16:33
infid sorin: FauxFaux you say it's too easy, sorin says it's too complicated. which is it?16:34
DireFog cbreak: it fails while rewriting tags AFAICT16:34
← defendguin left16:34
→ scarabx joined16:34
FauxFaux infid: Well, in my opinion, I'm right. Or, more seriously, the bits of the man page that are worth having in a pocket reference are bogglingly tiny.16:34
ManDay thiago: You DONT KNOW how HORRIBLE it is to live like that!16:35
thiago ManDay: connected to the internet? Yeah, horrible.16:35
cbreak DireFog: weird.16:35
→ shiba_yu36 joined16:36
ManDay thiago: No, using an Atheros card which is voted as working superb on linux but which constantly tortures you for over a year.16:36
← goshakkk left16:36
FauxFaux Homeplugs / pla / egthernet over power time.16:36
← Cromulent left16:37
← shiba_yu36 left16:37
→ Tabmow joined16:38
infid FauxFaux: yeah well, there's a cvs pocket reference but not one for svn either. and the svn manual in paper back is a pretty big book16:39
sorin infid, it's both.16:39
→ bloopletech joined16:39
→ adamm joined16:39
sorin Git is not easy, neither is Mercurial, whith pretends to be easy, but it's quite complex.16:40
There is a learning curve to everything.16:40
← posciak left16:40
sekhmet (That said, git really isn't that difficult to pick up for simple use-cases)16:41
→ rpg joined16:41
FauxFaux infid: Also, o'really do those, and they basically publish books at random.16:41
sekhmet IMO if you're already familiar with non-distributed VCSes, you can hop into git using similar workflows within a few minutes16:41
sorin sekhmet, It would have been nice to hide Git's complexity for simple use cases.16:41
sekhmet I suppose16:41
→ f0i joined16:42
→ dcarper joined16:42
sorin I'm sure libgit2-based clients will fix Git's ugliness.16:42
cbreak you found uglyness in git?16:43
bremner as long you write to suit your definitions, I agree. when libgit2 is finished.16:43
ManDay In order to extract one directory and its contents from a repository and turn it into an own repository I first need to clone the reposirtory before I run filter-branch, so that I dont loose the rest, right?16:43
FauxFaux Should put some time into that.16:43
bremner sorin: you can already use mercurial as a git client aiui.16:43
cbreak ManDay: you can just make a branch16:43
→ possibilities joined16:44
sorin bremner, You can, but it doesn't support everything.16:44
ManDay cbreak: and then?16:44
cbreak filter-branch16:44
ManDay After I filtered the branch, how I separate the two?16:44
← dfr|mac left16:45
ManDay So that I got two distinct repos with separate history (both of which have no longer notion of the other)16:45
bremner sorin: well, that's not git's fault ;)16:45
→ toabctl joined16:45
cbreak branches are separate16:45
→ towski joined16:45
cbreak you can push them somewhere else16:45
or just clone before or after the filtering16:45
ManDay cbreak: And that will not clone/push the data that is contained in the branch which has not been filtered?16:46
cbreak depends...16:46
it's git. your choice what to fetch/push16:46
ManDay I don't understand that. If I "filter the branch" and the other branches remain uneffected, that means the history is still haning arround16:47
Doesn't it?16:47
→ LouisJB joined16:47
cbreak sure16:47
sorin bremner, bzr has a pretty decent interface. Hopefully, a libgit2-based client will look at it for inspiration.16:47
ManDay but i want to get rid of everything that is related to that directory16:47
cbreak then just do that16:48
ManDay cbreak: with that new repository i want to extract there should no longer be a bit associated with the old data16:48
cbreak: can you please make an example?16:48
cbreak why?16:48
bremner sorin: well, maybe. Or you could just propose patches to git. Previous separate interfaces have not been sustainable (see e.g. cogito)16:48
cbreak it's all simple git commands16:48
push what you want16:48
← macmartine left16:48
cbreak delete what you don't want16:48
fetch what you want16:48
there's nothing special about it16:48
ManDay cbreak: Of course there is16:49
I want to rid the repository of all the history associated with the parent directory.16:49
cbreak there's nothing special about it16:49
just do it16:49
ManDay *how*16:49
cbreak you've ever looked at man git-branch?16:49
← Rvl left16:49
cbreak it can delete branches16:49
sorin bremner, neah. People who are used to Git's ugliness tend to oppose such patches.16:49
cbreak or git fetch? it can fetch branches16:50
→ Textmode joined16:50
→ macmartine joined16:50
ManDay cbreak: If I fetch a branch, i fetch the history associated with it.16:50
cbreak that's the whole point...16:50
ManDay I dont want the history.16:50
cbreak then DO NOT FETCH THE BRANCH!16:51
ManDay Thats what I meant by16:51
cbreak is that so hard?16:51
fetch what you want16:51
do NOT fetch what you do NOT want16:51
it's like photoshop16:51
if you don't want the picture to be pink16:51
don't paint it pink16:51
← f0i left16:51
← ZapZ left16:51
ManDay cbreak: if its so easy please make an example. I want to extract A/B1/C1 from the A repository as an own repository, how?16:52
cbreak git branch -d branch1 deletes branch116:52
got it?16:52
selckin http://www.google.com/search?q=git+split+repo16:52
cbreak just delete all you don't want16:52
or git fetch remotename branchname16:53
fetches the branch into FETCH_HEAD16:53
→ canton7 joined16:53
← macmartine left16:54
→ f0i joined16:54
ManDay cbreak: this is not helpful16:54
cbreak because you don't read what I write!16:54
I answered your question like 5 times16:54
ManDay selckin: Yes, the thread on StackOverflow does do it the way i suggested.16:54
cbreak: You keep talking about fetching branches which makes no sense to me. thats why i asked for an example16:55
EugeneKay ManDay - you want to split a directory out? filter-branch. The command is misnamed, it's really filter-work-tree.16:55
cbreak do you know git fetch?16:55
→ thoraxe joined16:55
cbreak do you know what it does?16:55
thoraxe if i wanted to grab a previous commit to test what i might have broken, i always forget how to do that16:55
cbreak EugeneKay: wrong16:56
it filters branches, not the work tree, so it's filter-branch, not filter-worktree...16:56
ManDay cbreak: If you could drop your permantent counter-questions and eventually answer straight-ahead i might actually understand it.16:56
cbreak if you want filter-worktree, look at sed or awk16:56
ManDay: ok, I assume you know nothing about git16:56
ManDay then you assume wrong.16:56
selckin ManDay: what part of the stackoverflow awnser do you have problems with16:57
cbreak ManDay: then don't pretend you're a moron16:57
if you know git fetch16:57
EugeneKay cbreak - filtering a "branch" to me means to filter out differing branches. filter-branch works on the repo itself.... this is just how I'm looking at it, within the context of the usual usage.16:57
cbreak then you know that it fetches history16:57
make a new repo16:57
fetch the branch you want16:57
done!16:57
EugeneKay It may be wrong, but it makes sense, within context.16:57
cbreak is that so hard to get?16:57
ManDay none selckin. Since it is exactly what I said. But cbreak started claiming that I would not have to clone that repo a priori and instead fetch something, so I got involved with him.16:57
cbreak EugeneKay: filter-branch means you're filtering a branch16:57
ManDay cbreak: Not per se16:57
cbreak you put a branch into it and get the filtered branch out16:58
← cesc left16:58
cbreak ManDay: not per se?16:58
EugeneKay cbreak - it depends on how oyu look at it. From a git internals perspective that makes sense, but looking at the filesystem, not quite as much.16:58
ManDay cbreak: I'm not entirely convinced (or understood how) the history will be gone from the branch by fetching it.16:58
cbreak EugeneKay: who cares about the filesystem? it's not involved16:59
EugeneKay cbreak - and, as far as "how hard is it to understand", history suggests that unlearning a concept is harder than learning a new one. See the group of people who are convinced that Windows 7 is the invention of the devil, and XP is the one true windows.16:59
cbreak ManDay: you fetch the branch you want to keep!16:59
→ ajpiaNOU joined16:59
← ajpiaNOU left16:59
cbreak the branches you do NOT fetch are not there16:59
got it?16:59
if you only want to keep the new filteredBranch, just fetch that branch17:00
→ ajpiaNOU joined17:00
← ajpiaNOU left17:00
ManDay cbreak: You're really making it hard to understand you if you are deliberately not replying to questions but instead leading a monologue one is supposed to learn from.17:00
cbreak all other branches will not be there because you did not fetch them17:00
I answered your question like 8 times by now17:00
but you're acting like a read only medium :(17:01
→ ajpiaNOU joined17:01
← ajpiaNOU left17:01
→ __iron joined17:01
EugeneKay cbreak - repeating yourself 8 times is doing him no good when he doesn't understand the explanation. ;-)17:01
← canton7 left17:01
cbreak he doesn't say what he doesn't get17:01
sekhmet TYPE LOUDER17:01
ManDay cbreak: I bet you tried. But it really didn't come across. For whose fault it might be.17:01
cbreak ok, noob edition17:01
EugeneKay So, stop wasting your time and let somebody else pound it into his head :-p17:01
→ ajpiaNOU joined17:01
← ajpiaNOU left17:01
ManDay sekhmet: excellent idea, lol17:01
cbreak init a fresh repository17:01
shturm how do i set .gitignore to ignore any files under a folder/ but keep the folder sutrcture ?17:01
cbreak what history do you have?17:02
NO HISTORY17:02
EugeneKay shturm - git doesn't track folders, only files within them.17:02
cbreak because it's a fresh repository17:02
then you fetch the history you want with git fetch17:02
what history do you have now?17:02
the history you want!17:02
anything else? NO!17:02
why not? because you didn't fetch anything else!17:02
horay!17:02
EugeneKay You can touch&add a foo/bar/.folder file within an ignored folder foo/bar/ to make the dir create.17:02
→ canton7 joined17:03
ManDay Does anyone else see a point in what cbreak says with regard to my question?17:04
(Which is how to get rid of history based upon which directory it belongs to - in other words just keep the history of one specific directory)17:04
→ Squarism joined17:04
selckin sides with cbreak and thinks ManDay needs some sleep17:04
← dnivra left17:05
← _iron left17:05
→ harsh joined17:05
EugeneKay ManDay - So, you want to remove foo/ from history? Like it never existed?17:05
→ alexim joined17:05
← ManDay left17:05
← alexim left17:06
→ ManDay joined17:06
ManDay Sorry, you know the story with my ATH card.17:06
DireFog OK, just for dummies, this should work when starting from a freshly cloned repo, right? git filter-branch -d /tmp/lolrebase2 --subdirectory-filter merb-core --tag-name-filter cat --prune-empty17:06
ManDay cbreak: My problem with your suggestion is that I don't see where the "abandoning the other history" comes in.17:06
← bloopletech left17:06
selckin ManDay: the making a new repo part17:06
← canton7 left17:07
cbreak ManDay: man git filter-branch17:07
ever heard of it?17:07
as I said like half an hour ago...17:07
it filters a branch17:07
ManDay cbreak: What happens with the history in the commits which the branch shares with branches which I did NOT filter?17:07
← f0i left17:08
cbreak ManDay: the whole branch gets filtered17:08
the filtered version has no commits in common with other branches17:08
ManDay Aha.17:08
EugeneKay filter-branch usually ends up rewriting all history in the repo, because the SHA1 chain changes from the first commit on up.17:08
cbreak maybe the initial commit, but that'd be a very rare case, which doesn't matter17:08
ManDay cbreak: That explains it.17:08
EugeneKay So you have to throw out all fo the old packfiles, etc.17:09
cbreak EugeneKay: no17:09
→ canton7 joined17:09
← canton7 left17:11
← envi left17:11
← Pupeno left17:13
→ canton7 joined17:13
→ Error404NotFound joined17:13
→ flavius joined17:13
← w0bni left17:13
→ ajpiaNOU joined17:16
← ajpiaNOU left17:16
→ munificent_ joined17:16
→ ajpiaNOU joined17:17
← ajpiaNOU left17:17
→ robotmay joined17:17
→ f0i joined17:17
→ ajpiaNOU joined17:18
← ajpiaNOU left17:18
shturm thanx EugeneKay17:18
→ w0bni1 joined17:18
munificent_ i'm having a "fatal: The remote end hung up unexpectedly" problem pushing to github, which i don't *think* is related to ssh. (https://gist.github.com/1225296) any ideas?17:21
→ j0llyr0tten joined17:21
j0llyr0tten hi, anybody can instruct me on signing a tag?17:21
← Squarism left17:22
FauxFaux munificent_: git branch -avv; git remote -v17:22
j0llyr0tten i have a public/private key pair already for github.com and i thought i might be able to use that pair?17:22
FauxFaux j0llyr0tten: Wrong kind of key; gpg and ssh kkeys are unrelated.17:22
j0llyr0tten FauxFaux: i did not know that ... interesting, i thought all keys are created equal17:23
munificent_ FauxFaux, https://gist.github.com/122530017:23
← toabctl left17:23
FauxFaux j0llyr0tten: Arguably a key is just a random number, and you can probably kind of convert between them, but... :)17:23
→ Targen joined17:24
FauxFaux munificent_: Looks fine; that's me out of guesses. :)17:24
j0llyr0tten FauxFaux: i'll look up how to use GPG on Mac OS X cuz that's what I'm on, if I run into trouble using it the key I generate with git I'll come back here17:24
FauxFaux: thx17:24
munificent_ :( thanks anyway.17:24
ManDay cbreak: The counterpart for the subdirectory-filter will be the index-filter, to remove the extracted directory from the repo, right?17:26
cbreak hmm? it's not reversible17:26
oh, yes17:26
that kind of counterpart17:26
→ rolfb_ joined17:27
→ Saur joined17:27
← Saur left17:27
→ lacrymology joined17:27
canton7 munificent_: try GIT_TRACE=1 git push ?17:29
munificent_ trying...17:29
ManDay cbreak: thanks17:30
canton7 munificent_, and you haven't suddenly gone behind a new firewall, have you? try cloning a random repo over ssh17:31
→ normanrichards joined17:31
munificent_ no, i don't think so. just on my usual home wifi. let me try grabbing a random repo.17:31
canton7 actually that's a dumb suggestion. try sshing into anything, though17:31
→ kumbayo joined17:32
← f0i left17:32
munificent_ i've trying pushing and pulling from other repos of mine and that doesn't seem to work either.17:32
canton7 can you ssh into *anything*?17:33
→ leen^| joined17:33
munificent_ i not sure, any idea what servers i should be able to connect to? i don't use ssh much/ever.17:33
canton7 munificent_, ssh 178.79.161.15 ?17:35
success is a prompt for password17:35
j0llyr0tten FauxFaux: hi again, any idea where i need to put the key?17:35
→ gnufied joined17:35
FauxFaux j0llyr0tten: I thought it Just Worked... although, you did say OS X.17:35
munificent_ result of GIT_TRACE=1 git push: https://gist.github.com/122529617:35
ManDay cbreak: Any idea how it is with the history with filter-branch --index-filter? Will all history associated with a file be garbage-collected if I prune it out of all indices in all commits?17:36
→ p3rror joined17:36
munificent_ also, ssh 178.79.161.15 is just sitting there, so it's probably going to fail too...17:36
cbreak what?17:36
j0llyr0tten FauxFaux: hold on, i may not have specified the signing key correctly17:36
← Dave^|| left17:36
cbreak unless you fetch the new branch into a new repository, the old history will stick around for a long time17:36
→ madprops joined17:36
cbreak kept alive by reflog, other branches, tags, ...17:36
← kadoban left17:37
canton7 munificent_ sounds like the problem is your end, namely that something's blocking your ssh attempts17:37
← ManDay left17:37
munificent_ yeah, looks like it. weird. thanks!17:37
→ ManDay joined17:37
ManDay Any idea how it is with the history with filter-branch --index-filter? Will all history associated with a file be garbage-collected if I prune it out of all indices in all commits?17:37
canton7 munificent_, as a work-around, you can always push over http17:37
thiago ManDay: garbage collection happens at a later time17:38
munificent_ thanks canton7, i'll try that.17:38
thiago it's independent of filter-branch17:38
← munificent_ left17:38
canton7 munificent_, sweet np17:38
ManDay thiago: so what does actually happen when I filter a file out of the index at certain commits with filter-branch and how can I get it to eventually completely disappear?17:39
thiago ManDay: it removes from the commits17:39
cbreak ManDay: read above17:39
thiago ManDay: but filter-branch keeps a backup. You have to delete it.17:39
j0llyr0tten FauxFaux: done! it just works once i tell the config the signing key id correctly17:39
cbreak if you want to make it really clean, do as I told you hours ago, and fetch the new history into a new repository17:40
j0llyr0tten FauxFaux: do i need to git push the tag now?17:40
ManDay cbreak: I got that17:40
cbreak if you do NOT do that, the old history will stick around17:40
ManDay Does that also work for the --index-filter (the counterpart)17:40
cbreak for probably a month or longer17:40
it works for any kind of history rewriting17:40
FauxFaux j0llyr0tten: If you want people to see it, yes.17:40
→ OOPMan joined17:40
ManDay cbreak: In the case of --index-filter i do not really understand what happens. If I remove a file from the index at only specific times, for example, what will happen in the file in the future comits?17:41
j0llyr0tten FauxFaux: but I did not tag a commit :( i tagged nothing it seems :(17:41
cbreak ManDay: they will be there17:41
inex-filter filters each commit individually17:41
so if you remove a file only in the first half of the commits, it would appear as if it was added the start of the second half17:42
← metcalfc left17:42
cbreak this is different from rebase -i of course17:42
ManDay it would be in the future commits unchanged?17:42
wereHamster ManDay: filter-branch works on commits. It gives you each commit in the range you specify and you can modify each commit however you want.17:43
ManDay wereHamster: But arent the second half of commits based upon the first half17:43
So if I mess with the first half it has consequences for the second half?17:43
← w0bni1 left17:43
wereHamster ManDay: it looks at each commit individually17:43
cbreak ManDay: tree/index filter filter the tree/index of a commit17:44
← possibilities left17:44
cbreak ONLY a commit17:44
DireFog so yay, it seems like this repository is ultimately breaking filter-branch.17:44
ManDay wereHamster: cbreak : yes i got that17:45
DireFog https://gist.github.com/122531817:45
just farking ONCE I'd like a tool to just work.17:45
→ beto joined17:46
ManDay i just dont understand how it can go without consquences. if I travel back to the past and change something, that will have consequences for the future no?17:46
cbreak ManDay: no17:46
think of git as a series of snapshots17:46
← diegoviola left17:46
← beto left17:46
cbreak imagine you take a picture of a house that is built17:46
now with filter-branch you go back and filter out the fundament of the first three pictures17:46
this doesn't change any of the latter ones17:47
← subbyyy left17:47
cbreak git works with snapshots17:47
→ subbyyy joined17:47
ManDay i read that. but technically - doesnt git base its commits upon diffs?17:47
→ diegovio1a joined17:47
ManDay if I change something in the past, those diffs wont patch any longer, no?17:48
cbreak no17:49
http://eagain.net/articles/git-for-computer-scientists/17:50
Araxia__ ManDay: your thinking about git is being unduly influenced by your familiarity with other vcs systems.17:50
cbreak look at the third picture17:50
syphar|away → syphar17:50
→ haydenmuhl joined17:51
ManDay cbreak: I know17:51
← canton7 left17:51
cbreak then why do you say "based on diffs"?17:51
ManDay But that abstract representation doesnt tell me how this is implemented.17:51
cbreak no17:51
it's not an abstract representation17:51
it tells you how it is implemented17:52
ManDay oh17:52
← diegovio1a left17:52
→ jbrokc joined17:52
← jbrokc left17:52
cbreak git stores files as blobls.17:52
ManDay so you mean that every commit has a full description of the tree associated with it (which can stand alone)?17:52
cbreak it stores tree objects that point to those blobs, giving them file names17:52
trees can contain other trees, making subfolders17:52
a commit points to a tree17:52
ManDay i know17:53
hm17:53
cbreak the commit therefore refers to every single file17:53
not diffs17:53
not patches17:53
ManDay i guess i kind of got it. i need to think that over17:53
→ doug joined17:53
cmyers it is dififcult to wrap your brain around at first, but once you do, very powerful17:53
← brian_g left17:53
→ f0i joined17:53
cbreak as I said, http://eagain.net/articles/git-for-computer-scientists/ :)17:53
doug is there a way to tell what branches exist in a remote repo?17:53
→ toabctl joined17:53
cbreak doug: man git-ls-remote17:53
doug cool. does that allow a hook on the remote side?17:54
cbreak what?17:55
it doesn't matter if the remote has a hook17:55
it won't be run for listing remote refs17:55
doug so, i can't set up a hook for that one on the remote repo?17:56
iband → iband|away17:56
cbreak for doing what?17:56
anyway, food :)17:56
man githooks lists all hooks you can have17:56
syphar → syphar|away17:57
→ canton7 joined17:57
doug hm, seems there no hook for that one.17:58
cmyers doug: a hook doesn't make much sense - what do you want to accomplish?17:58
← mino left17:58
doug ask me in a month or two17:59
cbreak do you know gitolite?17:59
→ vbajpai joined18:00
→ dexter_e joined18:00
doug yup, i'm using that.18:00
iband|away → iband18:01
vbajpai I want to move my public projects from bitbucket to github? how do I export an hg project to git (I have tried the hg-git extension; and I cannot get it to install)18:01
Araxia__ vbajpai: what os?18:01
bremner vbajpai: there is a nice git-hg tool based on fast-export18:01
vbajpai Mac OS X18:02
bremner: are you talking about: http://hg-git.github.com/ ?18:02
bremner http://repo.or.cz/w/fast-export.git18:02
doug cbreak++18:03
thanks18:03
Araxia__ try `sudo easy_install hg-git`18:03
→ guns joined18:03
→ d0k_ joined18:03
bremner but is that even the right tool?18:03
Araxia__ yeah, i used it just a few weeks ago. works quite well.18:03
bremner ok18:03
← d0k left18:04
d0k_ → d0k18:04
← f0i left18:04
→ mino joined18:05
← toabctl left18:06
→ f0i joined18:07
← cyphase left18:08
vbajpai Araxia__: it has a weird dependency on a version of dulwich which is not even out yet18:09
→ possibilities joined18:09
Araxia__ dulwich? hrm. i never had to do anything with that. what version of os x?18:10
→ Tyraziel joined18:10
vbajpai lion18:11
← cjs left18:11
Araxia__ i've got a lion box i haven't installed this on yet. let me see if i can get it working there.18:12
→ jbrokc joined18:12
→ alester joined18:15
ChanServ set mode: +v18:15
Tyraziel Hi all. Just installed GIT and was about to create a repository however I have a few "best practice" questions, if someone could provide some guidance.....18:16
Araxia__ vbajpai: i assume you're using hg 1.9.x, no?18:17
vbajpai yep 1.9.218:17
wereHamster Tyraziel: man gittutorial18:18
Araxia__ okay, looks like there's an incompatibility. but if you grab hggit from the repo, you can get it to work pretty easily. instructions coming...18:18
vbajpai Araxia__: well here was my question: http://stackoverflow.com/questions/7388814/switching-mercurial-projects-bitbucket-to-git-github18:19
→ masonkramer joined18:19
vbajpai I have tried the repo thingy as well18:19
Araxia__ `git clone [email@hidden.address] then edit your .hgrc [extensions] hggit= /path/to/repo/hggit18:20
Araxia__ tried that?18:20
vbajpai yes18:20
and then I get *** failed to import extension hggit from ~/.bin/hg-git/hggit: No module named dulwich.errors18:20
← scarabx left18:20
vbajpai even though I can import duwlich and dulwich.errors in a python interpreter just fine18:21
Araxia__ weird. let me review that page and see what i can come up with.18:21
this was dead simple for me a few weeks ago. :-P18:22
→ drith joined18:22
vbajpai yeah, apparately somebody mistakenly bumped up dulwich requirements a week back, and easy_install thing has got broken since then18:22
← hsingh left18:23
→ ^Mike joined18:23
syphar|away → syphar18:23
→ mjt joined18:23
Tyraziel wereHampster -- my questions deals with handling multiple projects with GIT and the best practices around that.18:24
wereHamster Tyraziel: each project in its own repository18:25
Araxia__ vbajpai: hrm. `which dulwich` for me points to `/usr/local/bin/`, though i don't see it in `brew list`18:25
selckin Tyraziel: it helps to actually ask a question18:25
wereHamster usually, at least18:25
sometimes, we can read minds18:25
but not this case18:25
Tyraziel wereHamster - that actually answers my question18:26
Araxia__ vbajpai: are you using homebrew?18:26
→ herdrick joined18:26
→ tshauck joined18:26
vbajpai yes, but a brew search dulwich doesn't give me anything18:26
← muneeb left18:26
vbajpai Araxia__: ^18:26
→ muneeb joined18:26
Araxia__ yeah, i'm thinking it gets installed as part of something else. i have git-hg and git-extras installed.18:27
→ jceb joined18:27
vbajpai tortoiseHG maybe18:28
syphar → syphar|away18:29
→ brian_g joined18:29
→ warlock_mza joined18:29
← Tyraziel left18:31
vbajpai okay, hg-fast-export worked for me18:34
I am going to use that for now18:34
Araxia__: thanks for the help :-)18:34
bremner: thanks for the suggestion18:34
← vbajpai left18:35
bremner welcome18:35
← f0i left18:35
← towski left18:35
← evelyette left18:36
→ cesc joined18:36
DireFog can't even tree-filter the file that causes the problem18:36
→ Heimidal joined18:37
← sbell left18:38
← Targen left18:38
← kermit left18:39
← ajpiano left18:39
→ metcalfc joined18:40
→ munificent_ joined18:40
→ workmad3 joined18:40
→ ajpiano joined18:41
munificent_ canton7, i fixed my issue with pushing to git (or really, my ssh issue). strangely, all it took was a reboot. <shrug>18:41
← iband left18:42
→ caseymcg joined18:42
→ w0bni joined18:43
ManDay cbreak: wereHamster I filtered one branch and wanted to filter another now - now it speaks about a backup being overridden18:43
cbreak yeah. you have to delete it.18:43
ManDay that backup is only there to restore the branch which i filtered and is superflous if I had branched it as a beackup measure, right?18:43
cbreak your decision.18:44
it's a copy of the branch you filtered18:44
→ Rvl joined18:44
ManDay yes, but that "copy" in my case is already provided by master, right?18:44
darn18:46
→ glennpratt joined18:46
ManDay now I rewrote with git rm something/* and it also removed something18:46
→ hjax joined18:46
ManDay How can i prevent it from removing something and only remove everything in something?18:46
wereHamster git does not track folders18:47
if you delete all files in a folder that folder disapears18:47
→ theoros joined18:47
hjax hello i seem to be having a problem with git (i have never used it before0 when i try to use the command "git push -u origin master" it says that the remote end hung up unexpectedly18:47
syphar|away → syphar18:48
→ stoffus joined18:48
ManDay wereHamster: right18:48
thanks18:48
← munificent_ left18:49
→ Shusshu joined18:50
DireFog cbreak: moment of WTF, it works when I don't specify a temp directory18:51
→ dangerousdave joined18:52
DireFog i.e., no -d option18:52
syphar → syphar|away18:53
← intchanter left18:53
← doug left18:53
← Davey_ left18:53
dangerousdave hi, i have git repository on my laptop which i pulled from my desktop. I accidently deleted my desktop version, so was hoping to recover from my laptop. Problem is, my pull version only seems to have the master branch, and none of the other branches my desktop version had.18:54
→ f0i joined18:55
→ intchanter joined18:55
→ tdebat joined18:55
← stoffus left18:56
jmah dangerousdave: usually it will know about other branches of the remote18:56
even if there isn't a local branch for them18:57
SethRobertson git branch -a18:57
jmah (run "git branch -r")18:57
dangerousdave SethRobertson: they are all there! phew18:57
jmah: also all there18:58
← agile left18:58
dangerousdave so, what state is my repository in now then?18:58
and how do i get it so that "git branch" shows all branches18:59
← Akari` left18:59
jmah on your laptop copy, git checkout each named branch19:00
→ Akari` joined19:00
jmah (if it shows "origin/x", "git checkout x")19:00
(or, to avoid changing your working copy, "git branch x origin/x")19:01
then clone it back.19:01
You need to make local refs for the branches first because remotes only go one level deep19:01
dangerousdave jmah: great, thanks for all the help,all working now19:03
← zamabe left19:04
jmah pleasure19:04
→ HG` joined19:04
magnate Evening all19:04
← f0i left19:05
magnate Does anyone know how to say to git "show me a list of commits which are present in branch 1 but not in branch 2"?19:05
→ Klif1980 joined19:05
wereHamster magnate: git log branch1 ^branch219:05
^ == not19:05
← hjax left19:06
magnate Excellent - thanks19:06
I should have known it would be something simple!19:06
Trying it now19:06
→ AlexC_ joined19:06
AlexC_ g'morning19:06
jmah are they the same commits? or different commits with the same effect? (for that there's "git cherry")19:06
← Prasad left19:07
AlexC_ I have two branches, 3.2.x and master. The master branch has many many different commits that I would never want in 3.2, however there is some commits that I want to merge into 3.2.x. What is the best way of doing this? I know about cherry-pick, but I also know that breaks a lot of the history19:07
→ archis joined19:07
wereHamster AlexC_: why do you think it breaks history?19:07
← archis left19:08
jmah I agree that it's not great on your history, but I think it's still the best way19:08
AlexC_ wereHamster: hum, I was under the impression that it did - since they are effectively two different commits19:08
wereHamster AlexC_: can you elaborate on the 'break' aspect?19:08
AlexC_ wereHamster: not entirely; maybe break wasn't the best word19:09
wereHamster maybe19:09
jmah it the cherry-pick applies cleanly, some tools can deal with it nicely (like git-cherry)19:10
but if you need to fix up conflicts, I don't know of anything that considers them equivalent19:10
← alnewkirk left19:10
AlexC_ the commits will apply cleanly (at least, that is the plan)19:10
jmah isn't it always :)19:11
AlexC_ :)19:11
cybersphinx wereHamster: Is "git log branch1 ^branch2" different from "git log branch2..branch1"?19:11
AlexC_ ok, thanks. I'll go cherry picking19:11
← cesc left19:11
wereHamster cybersphinx: no, it's exactly the same19:11
→ bauruine joined19:11
← madprops left19:12
→ macmartine joined19:13
cybersphinx Ok. Never saw the ^ version before.19:13
wereHamster cybersphinx: you've never read man git-rev-parse19:14
← gretch left19:14
→ maestrojed joined19:14
← guns left19:15
cybersphinx Very possible.19:15
← AlexC_ left19:15
→ gbacon joined19:16
← demi` left19:16
← ^Mike left19:16
→ f0i joined19:17
← dexter_e left19:19
← rolfb_ left19:19
→ macmartine_ joined19:19
magnate wereHamster: it worked a treat - many thanks19:19
← macmartine left19:20
macmartine_ → macmartine19:20
← macmartine left19:20
selckin i really wish i could use head instead of HEAD19:20
← robotmay left19:20
selckin typing caps just doesn't flow19:21
← warlock_mza left19:22
← alester left19:22
jmah selckin: any command in particular?19:22
selckin any refspec19:23
← canton7 left19:23
→ demi` joined19:25
→ redii joined19:25
← resmo_ left19:26
→ resmo_ joined19:26
ManDay Do you know how to create a new repository on gitorious?19:28
→ VxJasonxV joined19:28
← CannedCorn left19:28
→ chrisf_ joined19:29
→ pantsman joined19:30
← pantsman left19:30
→ pantsman joined19:30
VxJasonxV is it possible to delete the 'push' only portion of a remote?19:31
← f0i left19:31
← nazgul101 left19:31
← nyuszika7h left19:32
← maestrojed left19:32
→ sylr joined19:33
jmah VxJasonxV: please explain19:34
→ nazgul101 joined19:34
← tshauck left19:34
VxJasonxV when I clone a third party repo, I have a fetch and a push url19:34
I have no permission to push, so I don't wish to enable certain software (github for mac) to try.19:35
→ f0i joined19:35
← adamm left19:35
jmah but you still want to pull from it?19:36
VxJasonxV yeah19:36
← harsh left19:36
← demi` left19:36
VxJasonxV GitHub for Mac has a sync option, but whenever I do I get a push error for a third party repo19:36
→ maestrojed joined19:36
← gnufied left19:36
ManDay How do I list all branches (including remote ones) of a repo?19:36
→ harsh joined19:36
← maestrojed left19:36
jmah git branch -a19:36
VxJasonxV I don't know python, so I have no intention of submitting patches to the repo owner, so forking my own repo seems unnecessary.19:37
ManDay thanks19:37
VxJasonxV I'd rather just maintain a straight clone of their repo.19:37
jmah Well it sounds like the GH client wants to push to somewhere, even if you change it19:37
VxJasonxV but without errors every time I sync :/.19:37
jmah there's Repository > Pull menu item19:37
esc VxJasonxV: 'git remote' has a '--set-url' option, perhaps that helps?19:37
← baburdick left19:37
VxJasonxV just set it to blank?19:37
→ maestrojed joined19:37
VxJasonxV jmah, is there? I think I've missed it...19:38
esc its worth a try, have a look at the manpage first though19:38
VxJasonxV oh, so, jmah you're saying don't sync, just pull?19:38
esc there is apparently a '--push' option too19:38
← workmad3 left19:38
jmah VxJasonxV: yeah probably, if that works19:39
VxJasonxV mmm19:39
← nazgul101 left19:39
VxJasonxV seems like the better answer is to just not use github for mac for this repo19:39
← SethRobertson left19:40
→ markit joined19:41
← maestrojed left19:41
VxJasonxV git remote set-url --delete --push origin 'https://github.com/rg3/youtube-dl.git' doesn't seem to change much. re-running git remote -v lists both the fetch and the push urls19:41
→ SethRobertson joined19:41
markit hi, newbie here. I've a branch "A", and now I create a new one from it "B". I would like to have a "tree" that shows that there is the main code, branch A and from it branch B. Is it possible? $ git branch shows me in a plain list19:42
→ maestrojed joined19:42
jmah git show-branch19:42
wereHamster git log19:42
→ baburdick joined19:42
jmah git log --oneline --graph19:43
markit git show-branch works, even if not very clear19:43
git log --oneline --graph does not show B as being from A19:43
→ nazgul101 joined19:43
cbreak git log --graph --oneline --decorate --all19:44
wereHamster markit: git doesnt' store the fact that B was created form A19:44
markit AH!19:44
so I could merge B from whatever branch?19:44
← nazgul101 left19:44
wereHamster mrege?19:44
markit merge to, sorry19:44
← maestrojed left19:44
wereHamster you can merge any commit with any other commit19:44
jmah um, sure19:44
markit oh...19:45
but this seems an important information for me19:45
← Shusshu left19:45
wereHamster what?19:45
← flavius left19:45
markit that a certain branch is "derived" from another one19:46
wereHamster well, other people disagree19:46
cbreak that's not really the case19:46
it's like copying an mp319:46
the result are two MP3s that are equivalent19:46
jmah don't think of it as one derived from another19:46
cbreak you could say one is the original and one is the copy19:46
jmah but rather that they used to be the same at a point in history19:46
cbreak but data wise, both are exactly the same19:47
markit I have a rails project at version 2, I create a branch for upgrade to version 3, then I create from it a branch for a fix I need, but that branch can reasonably be merged only to verion 3, not 219:47
→ macmartine joined19:47
wereHamster markit: well, git won't prevent you from trying to merge it into 219:47
← macmartine left19:47
wereHamster you can try if you really want to. It's up to you not to do it19:47
markit if someone else sees branches 2, 3 and "fixed", could be tempted to merge "fixed" to 2 and not to 319:47
← DireFog left19:48
wereHamster markit: name it fixed-only-for-319:48
markit I see, good workaround19:48
thanks for the clarification :)19:48
→ macmartine joined19:48
herdrick question: it looks like git revert to an older commit will try to only un-apply that commit, leaving behind any changes since then19:48
which is ccol19:49
← metcalfc left19:49
herdrick but what if I just want to get the codebase as it existing after that commit?19:49
← field_it left19:49
jmah git checkout -b as-it-was <commit>19:49
herdrick I just want to go back to an earlier point in time19:50
jmah, ah, ok, thanks19:50
jmah (or you can use "reset" to destructively update the current branch)19:50
cbreak herdrick: revert creates undo commits19:50
reset kills history19:50
markit btw, $ git branch shows a "delete" branch in the list... I've no memory of having created it, is it automatic?19:50
cbreak pick your tool.19:50
← redii left19:51
herdrick jmah: cbreak: ok, that's what i had read about reset - definitely don't want that19:51
wereHamster markit: no, you created it19:51
→ shedammit joined19:52
← f0i left19:52
→ davzie joined19:54
herdrick ok, so in the docs, ex. http://linux.die.net/man/1/git-checkout , <tree-ish> could be anything that could be a tree of code?19:54
i.e. "commit, tag or tree"19:54
?19:54
← OOPMan left19:54
→ f0i joined19:55
→ n8o-mba joined19:55
→ codebeaker joined19:55
← Rvl left19:55
markit wereHamster: maybe I wanted to delete a branch, and issued the wrong command that created a "deleted" one? sigh19:55
ManDay cbreak: You said that changes only affect the current branch19:56
is fit filter-branch -- --all an exception to that?19:56
markit wereHamster: how can I tell if it's code has been merged?19:56
wereHamster ManDay: the --all means you want to apply the changes to all branches19:56
so yes, it's an 'exception' to that19:57
ManDay wereHamster: Hence my question19:57
wereHamster markit: git branch --merged19:57
ManDay: it's not really an exception. You told git to do that.19:57
→ CannedCorn joined19:57
cbreak ManDay: I said "unless told otherwise" :)19:57
--all affects all19:57
← codebeaker left19:58
markit wereHamster: says "* delete" and the following line "master", I guess means it has been merged to master, correct?19:58
→ zamabe joined19:58
← ajpiano left19:58
→ ajpiano joined19:58
wereHamster markit: no. it means your current branch is delete19:58
markit: go read a git tutorial19:58
markit ok, I feel lost with git19:59
← muneeb left19:59
markit better read the basics19:59
wereHamster hnece why I said to go read a tutorial19:59
← dcarper left19:59
markit yes, good suggestion, I thought was very simple but I've discovered I'm messing it a bit too much, lol20:00
← Klif1980 left20:00
← harsh left20:00
→ dcarper joined20:01
→ harsh joined20:01
← dcarper left20:02
ManDay cbreak: How do I fetch only one specific branch from the repository at ../mydir/myrepo ?20:02
cbreak read man page20:03
← n8o-mba left20:03
cbreak it'll end up in FETCH_HEAD20:03
→ xpot-mobile joined20:03
ManDay i dont understand20:03
← localhost left20:04
jmah git fetch ../mydir/myrepo thebranch20:05
← mattdipasquale left20:05
ManDay oh, so a branch name is a valid refspec, i see20:05
→ nyuszika7h joined20:05
→ archis joined20:06
← archis left20:06
→ localhost joined20:06
← ehsan left20:07
← brian_g left20:08
← masando left20:08
→ EricInBNE joined20:09
→ OOPMan joined20:09
ManDay why does the fetched branch not show up in git branch?20:09
cbreak it'll end up in FETCH_HEAD20:09
→ brian_g joined20:09
cbreak you have to make yourself a branch yourself20:09
ManDay and then? how do I get the branch which i fetched checked out?20:10
→ belkiss joined20:10
cbreak man git checkout20:10
← f0i left20:11
ManDay i dont see how. it says just what i already know: how to check out an existing branch20:11
wereHamster ManDay: see the -b option20:11
or use git brnach to create the branch before checking it out20:11
ManDay wereHamster: It just says that with -b I specify a start point20:12
wereHamster it probably does20:12
ManDay I still have no clue what happend when the branch got put into "fetch head"20:12
markit wereHamster: I'm reading a "git pro" tutorial about merging and branching... I've found the command "git branch --no-merged" that should show not merged brances.. but here shows nothing, while sure I've not merged ror31 to master. Do I have to study this stuff deeply, or is something you can easely enlight me on?20:13
jmah git checkout -b other-branch FETCH_HEAD20:14
wereHamster markit: checkout master first20:14
markit oh, thaks a lot20:14
← chrisf_ left20:14
ManDay fatal: Not a valid object name: 'master'.20:14
When I try to create a branch20:14
wereHamster ManDay: the complete command please20:15
← ReekenX left20:15
ManDay git branch master20:15
wereHamster that doesn't make sense20:16
ManDay i guess it doesnt20:16
wereHamster you're saying: create a new branch with name 'master' and make it point to the same commit as HEAD20:16
markit wereHamster: I've the feeling that I've messed things up a lot, sigh... but I'm not sure if "deleted" is the brench I used to create ror30 and then, from it, ror31. If is so, I could easely merge ror31,otherwise I'm in troubles. That's why I thougth were important know the "merge tree" (what is merged from what)20:16
or, probably, I'm not aproaching the stuff properly20:16
ManDay wereHamster: I still have no clue how to set up a repository based upon only that remote branch as cbreak suggested20:16
(ps: this is a bare repo, if that matters)20:17
syphar|away → syphar20:17
cbreak in bare you don't even need to check out anything20:17
just fetch20:17
← __iron left20:17
cbreak then git branch somename FETCH_HEAD20:17
ManDay ah!20:17
cbreak but I told you that like three times already20:17
wereHamster step 1: create the repository. step 2: fetch commits into it. step 3: create branches to your liking20:17
ManDay so FETCH_NAME is also a valid refspec?!20:17
wereHamster step 4: ???. step 5: profit20:17
jmah markit: if there isn't too much going on, you can just do a big graph and try to untangle it20:18
git log --decorate --oneline --graph --all20:18
← Jon47 left20:18
← Raging_Hog left20:19
→ TeckniX joined20:19
← leen^| left20:20
markit jmah: mmm can't decode the output... may I post in private chat the colored lines to you?20:20
← Tommy[D] left20:20
mjt Hello here. I've a project in git which was based on upstream tarball releases imported as git commits into "upstream" branch. Now I want to switch to real upstream git branch instead. If i just merge -X theirs into my branch, it will tell me "no commits in common" and import whole upstream history between previous and current version. Is there a way to work around that, maybe by just removing whole upstream branch?20:21
→ srji joined20:21
→ Tommy[D] joined20:21
syphar → syphar|away20:22
→ demi` joined20:22
← unreal left20:24
mjt it looks like the best way is to recreate whole my branch using their git tree.20:24
→ f0i joined20:25
ManDay whats teh possibly advantage of --tree-filter vs --index-filter?20:25
When would I use the former?20:25
← harsh left20:25
wereHamster ManDay: I believe it's described in the man page20:26
ManDay wereHamster: I believe the manual only unconditionally states that index-filter is faster20:26
wereHamster it also states why20:27
ManDay it does not say when not to use it in place of tree filter20:27
yes. but thats not my questions20:27
wereHamster the tree filter checks out the tree20:27
ManDay I know20:27
My question is when I would have to use the tree filter?20:27
wereHamster so if you need the tree, use the tree filter. If you can operate only on the index, use the index filter20:27
ManDay ah, ok, thats plausbile20:27
sorry im a little tired20:27
a lot actually20:28
wereHamster how was that not obvious?20:28
jmah if you want to change file content, it's easier to use tree-filter20:28
ManDay yes yes i see20:28
wereHamster if you're tired, get some sleep20:28
ManDay wereHamster: I thought that if you wanted to perform 'rm' you would never need it20:28
jmah if you want to delete some files, it's just as easy to use index-filter, and much faster20:28
ManDay but of course the content could matter20:28
wereHamster ManDay: never need what?20:28
ManDay tree filter20:29
← davzie left20:29
wereHamster you can perform a rm only with the index (git update-index --remove ...)20:29
or git rm --cached ..20:29
→ unreal joined20:29
← unreal left20:30
→ unreal joined20:30
→ davzie joined20:30
← jds left20:33
← f0i left20:34
→ nadavoid joined20:34
→ metcalfc joined20:34
→ maestrojed joined20:34
→ apocalyptiq joined20:35
→ robotmay joined20:36
→ hsingh joined20:37
→ jds joined20:37
← jceb left20:37
→ ISF joined20:37
← bburhans left20:39
→ bburhans joined20:39
← bburhans left20:40
→ Anchakor joined20:40
Anchakor hi, is there a way to change the description note of the last commit?20:41
← dSebastien left20:42
jmah git commit --amend20:42
(that will re-write it as a new commit, so it's not good if it's been pushed or merged)20:42
Anchakor ah thanks, this will do20:43
→ HG`` joined20:43
← tdebat left20:44
infogulch yay! i'm starting to use the command line with git more. but i was lazy, i copied/pasted from one branch into another instead of cherry-picking xD20:45
→ Ben1980 joined20:45
→ bburhans joined20:45
infogulch luckily it's only 50 or so lines >_>'20:45
← Ben1980 left20:46
← HG` left20:46
→ madprops joined20:47
→ corburn joined20:48
← unreal left20:49
cbreak ManDay: every file inside .git that contains a valid hash or symbolic ref can be used as ref20:50
well20:50
man git rev-parse20:50
← Anchakor left20:50
← johnkpaul left20:51
→ unreal joined20:51
→ psoo joined20:51
← corburn left20:51
→ johnkpaul joined20:51
← Drakonite left20:51
ManDay thanks20:52
→ f0i joined20:53
← m4dc0d3r left20:53
ManDay i need to know: Will the filter-branch -d directory be of the size of the repository?20:53
can you guess what size it will have to be if the repo is 6 gigs?20:53
cbreak hmm?20:53
filter-branch does not shirnk the repo in any way20:53
ManDay cbreak: The suposedly tmpfs dir20:53
wereHamster ManDay: about as big as each tree20:54
ManDay cbreak: I don't know how filter-branch works internally so i dont know what will have to fit into that directory20:54
wereHamster: thanks20:54
wereHamster because each tree will be checked out into that temp dir20:54
that is, assuming you use the tree filter20:54
cbreak oh, you mean the temp dir20:54
← davzie left20:54
cbreak size of the biggest checkout20:54
ManDay yep20:55
← helgikrs left20:55
← Deesl left20:55
← jbrokc left20:58
← akosikeno left20:58
← Daniel0108 left21:00
← Sajaki left21:02
← f0i left21:02
→ mike007 joined21:02
mike007 I have some development and testing branches that I am going to merge into master soon. How do I merge where there is only one commit on master when I do this?21:03
← subbyyy left21:03
thiago mike007: you mean, "how do I discard all the development history of these branches" ?21:03
mike007 ya21:04
i just want one clean commit on master, ie: v1.0 is all you see on master21:04
thiago git merge --squash21:05
← srji left21:05
← gbacon left21:05
← johnkpaul left21:07
→ workmad3 joined21:07
mike007 is this normal to do thiago? or am I looking at it/doing it wrong?21:07
← ManDay left21:07
thiago no, it's not normal21:07
← vmil86 left21:07
thiago discarding development history is not good21:08
mike007 well there will still be the dev branches with all that data21:08
→ f0i joined21:08
→ kermit joined21:08
← bitkiller left21:09
thiago why don't you want to keep the history where people look for it?21:10
→ runa joined21:10
runa heyas. in git flow hotfix, what should I use as "version" the last rev in master?21:11
mike007 because there are a lot of commits, branching, etc21:11
runa ah. forget it21:11
← socketwiz left21:12
→ johnkpaul joined21:12
← w0bni left21:12
← RobertLaptop left21:14
← whitman left21:15
← HG`` left21:15
← apocalyptiq left21:16
← metcalfc left21:16
← workmad3 left21:16
→ workmad3 joined21:17
thiago mike007: clean it up with git rebase -i, if you can21:17
← loic_m left21:17
→ RobertLaptop joined21:18
mike007 i just leave it as is. full dev and testing history21:20
→ jimubao joined21:21
→ bitkiller joined21:21
→ Cromulent joined21:22
← sythe left21:24
← runa left21:26
← hsingh left21:27
→ hsingh joined21:29
→ muneeb joined21:29
← muneeb left21:30
→ muneeb joined21:30
→ mandric joined21:30
← nicoulaj left21:31
← MUILTFN left21:31
← kumbayo left21:32
→ scarabx joined21:32
← Heimidal left21:33
→ madsy joined21:33
← LongBeach left21:34
← bitkiller left21:34
← f0i left21:35
← ajt left21:36
→ airborn joined21:36
→ ajt joined21:36
→ f0i joined21:38
→ Wooga joined21:40
← dangerousdave left21:41
← johnkpaul left21:43
→ johnkpaul joined21:46
← f0i left21:46
→ f0i joined21:50
← haydenmuhl left21:51
← flexd left21:52
← markit left21:54
→ Heimidal joined21:54
→ bitkiller joined21:56
→ flexd joined21:58
→ mocramis joined21:58
→ sythe joined21:58
← sythe left21:58
→ sythe joined21:58
mocramis hello. does git rm actually delete the file ?21:58
← airborn left21:59
← tewecske left21:59
sie How do I fix this? error: error in sideband demultiplexer21:59
kevlarman mocramis: yes22:00
sie mocramis - Yes, unless you make it a dry run with -n22:00
man git-rm22:00
kevlarman mocramis: git rm --cached if you only want to delete it from the index and not the filesystem22:00
→ oriba joined22:00
mocramis kevlarman thanks22:01
← mardum left22:02
← Heimidal left22:02
mocramis and how coul i remove added but not commited files ?22:03
→ mardum joined22:04
sythe → sythe222:04
← bitkiller left22:06
← LouisJB left22:07
← awallin left22:08
→ sbell joined22:08
→ Chaser joined22:09
cbreak mocramis: git status should tell you22:11
← robotmay left22:11
cbreak ... about the power of git reset in file mode22:11
← normanrichards left22:12
→ bitkiller joined22:12
← sirdancealot left22:12
→ jbrokc joined22:13
← johnkpaul left22:14
→ cyphase joined22:14
← ajt left22:15
← f0i left22:15
← macmartine left22:15
→ bigkm joined22:16
← p3rror left22:20
← sorin left22:21
→ Heimidal joined22:24
→ f0i joined22:24
← oriba left22:25
→ adamm joined22:25
← Heimidal left22:28
← churp left22:29
→ normanrichards joined22:29
← belkiss left22:32
→ javaanse_jongens joined22:32
← thansen left22:33
← Amorphous left22:33
← workmad3 left22:34
← pantsman left22:35
→ corburn joined22:35
← jbrokc left22:37
← Textmode left22:39
← TeckniX left22:39
← bitkiller left22:40
→ storrgie joined22:40
→ tdebat joined22:40
→ utu joined22:42
→ ehsan joined22:46
← bigkm left22:47
→ Amorphous joined22:48
← sylr left22:48
← utu left22:50
→ Targen joined22:51
→ bitkiller joined22:52
→ CoinOpeBoy joined22:52
← gusnan left22:52
→ angelsl joined22:52
→ G00053 joined22:53
→ scarabx_ joined22:53
G00053 I did a repo init in my home directory , how do I undo this, haven't sync'd yet22:54
← scarabx left22:54
← psoo left22:55
Araxia__ G00053: just remove the .git directory that was created.22:56
← bitkiller left22:56
→ subbyyy joined22:57
→ MUILTFN joined22:58
→ ymasory joined22:58
cbreak what's repo?22:58
DrNick Google's git tool for Android22:59
FauxFaux The android set of wrapper scripts for git that nobody understands or cares about.22:59
← the_metalgamer left23:00
→ surikator joined23:00
→ fatline joined23:02
→ sirdancealot joined23:03
← CannedCorn left23:04
→ kennethreitz joined23:05
→ helgikrs joined23:07
← ludde left23:07
← toxico left23:07
← kennethreitz left23:08
← nadavoid left23:09
→ nadavoid joined23:10
→ w09x joined23:10
→ bigkm joined23:12
← f0i left23:13
ouah imagine two commits in my history X and Y, where X is the ancestor of Y. Where in .git/ dir can I have the info that the ancestor of Y is X?23:13
FauxFaux ouah: git cat-file commit Y23:14
→ FylGood joined23:14
→ bitkiller joined23:14
fatline I'm trying to revert to the HEAD revision. As I found in the git book, I type "git revert HEAD"23:15
but, excluded the vim session that opens to change the commit message, nothing really happens23:15
FauxFaux fatline: What git book tells you to type "git revert HEAD"?23:15
→ ludde joined23:15
fatline it says something like: 3 files changed, 23 insertions(+), 24 deletions(-) but the files are not reverted23:16
sythe2 → sythe23:16
ouah FauxFaux: I mean, where in .git/ is this info stored?23:16
FauxFaux The command you have written makes no sense.23:16
fatline FauxFaux, http://book.git-scm.com/4_undoing_in_git_-_reset,_checkout_and_revert.html23:16
ah fantastic :D23:16
FauxFaux ouah: In the file called "X", in .git/objects/X's first two characters/the rest of it's characters23:16
fatline ah no wait, I've looked at the wrong section23:16
FauxFaux fatline: Mmm.. I guess it does make sense. I'm just insane, ignore me. :s23:17
fatline so, how would you revert to the HEAD?23:17
→ f0i joined23:17
FauxFaux That statement is the one that makes no sense. But, instead of making a fool of myself trying to answer it, I'm going to suggest you re-read that page, which is correct. ¬_¬23:17
fatline yep, sorry23:18
found it23:18
git reset --hard HEAD23:18
:P23:18
I was looking at "fixing _committed_ mistakes"23:18
thank you23:18
→ gusnan joined23:19
CareBear\ with or without changing history?23:19
git revert does not change history23:19
sometimes this is much prefered23:19
fatline I don't care about history. I just want to test one thing on the HEAD revision23:19
ouah FauxFaux: ok thank, and is this file compressed or something, I'm doing an hexdump and I can find the commit number in it?:p23:20
← bitkiller left23:20
corburn fatline: I think revert is for fixing mistakes on repo's that have already been pushed.23:20
FauxFaux ouah: Yes, it's compressed. The command I originally suggested will uncompress it for you. http://git.goeswhere.com/?p=git-sh.git;a=blob;f=git-sh.sh;h=dd1e0c8c33eafdcf58930bdefb9a8d452f62883d;hb=HEAD is the only other way I know how to do it. ¬_¬23:20
CareBear\ fatline : then you can make a new branch, or even use git stash if you just want a very temporary thing23:20
ouah FauxFaux: ok, it was just to understand the internal of how git works23:21
→ tshauck joined23:22
FauxFaux Read one of the fine manuals.23:22
fatline FauxFaux, link?23:22
← multi_io left23:22
→ multi_io joined23:22
ouah :)23:22
FauxFaux http://progit.org/book/ch9-2.html23:22
← sitaktif left23:22
FauxFaux Just ignore what it says about pack files, as it's all bullshit. </angry> :P23:23
← Phylock left23:23
ouah thank you, I think I miss this chapter;)23:24
← mlncn left23:25
← surikator left23:25
fatline :)23:26
← notbrent left23:30
→ bitkiller joined23:30
← f0i left23:32
← ludde left23:34
→ f0i joined23:34
→ Maxdaman1us joined23:36
← tshauck left23:38
← cyphase left23:39
← xpot-mobile left23:39
← malumalu left23:39
→ CannedCorn joined23:40
← RobertLaptop left23:40
← LekeFly left23:41
→ dangerousdave joined23:41
→ cyphase joined23:42
← MUILTFN left23:42
→ Yuuhi` joined23:44
← Yuuhi left23:44
← helgikrs left23:44
← dangerousdave left23:45
← syphar|away left23:45
← Lemon|mbp_ left23:46
← lacrymology left23:46
→ helgikrs joined23:46
→ syphar joined23:46
← rendar left23:47
← Cromulent left23:49
→ safinaskar joined23:50
→ jbalint joined23:54
→ mlncn joined23:54
← madewokherd left23:55
← multi_io left23:55
← rudi_s left23:55
→ rudi_s joined23:55
← soulcake left23:56
← kukks left23:56
← Yuuhi` left23:57
→ Yuuhi` joined23:57
→ kukks joined23:57
jbalint is there a mirror of the docs23:57
FauxFaux http://faux.uwcs.co.uk/gitdoc/23:57
jbalint thanks a lot23:58
← nadavoid left23:58
→ MatCat joined23:58
→ ajt joined23:59

Logs Search ←Prev date Next date→ Channels Documentation