Perl 6 - the future is here, just unevenly distributed

IRC log for #git, 2016-12-27

| Channels | #git index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:00 ejb joined #git
00:10 Foxhoundz joined #git
00:10 xaviergmail joined #git
00:11 whatitdo joined #git
00:14 jost joined #git
00:14 Rodya_ joined #git
00:15 rwp satbyy, I can't imagine a situation where would would "git checkout origin/master". It will work though. It will place you on a "detached head". Normally one does "git checkout master".
00:16 whatitdo joined #git
00:17 satbyy rwp: yes, but the question _why_ does it give an error when I try it. It should work right?
00:17 rwp It should work, should say you are on a detached head.
00:18 satbyy but like I mentioned it gives error
00:18 rwp It doesn't for you?  Then your origin/master is not there is the only reason I can think.
00:18 Noodlewitt joined #git
00:18 satbyy > error: pathspec 'origin/master' did not match any file(s) known to git.
00:18 satbyy so why isn't it there?
00:18 rwp If it were me I would "git remote -v" and "git remote show -n origin" and "git remote show origin" in that order and see what it says.
00:21 rwp Since normally origin/master is tracking an upstream repository and hopefully mostly in sync with your local branch master you can fix a messed up origin by removing it and adding it back in the way it should be added and then fetching again.
00:22 satbyy rwp: http://paste.debian.net/hidden/aebda440/
00:23 rwp It looks like your remote brances are not being tacked.
00:23 rwp s/tacked/tracked/
00:23 rwp No "Remote branches" section.
00:23 BenderRodriguez joined #git
00:23 satbyy how do i fix it?
00:24 rwp I would not call myself a git expert on this channel, advice beware, but I would try "git branch -u origin/master master" to repair that.
00:24 thiago joined #git
00:24 YuGiOhJCJ joined #git
00:26 satbyy rwp: well, didn't work. http://paste.debian.net/hidden/cf8739ab/
00:27 rwp Hmm... In a similar way that rebooting is a big global reset I have been known to remove origin and then add it back. A big clean sweep at that time.
00:27 satbyy git fetch --all and git push -u origin master -- didn't change anything
00:27 rizonz Yeah! pull that bottle of champagne! I have comitted something and pushed it and it got pulled again well :D
00:28 rwp Try: "git remote rm origin" "git remote add origin ssh://example.org/path/foo.git" "git fetch" "git branch -u origin/master master"
00:28 thiago that's git remote prune origin, plus breaking any branch dependencies
00:28 rwp rizonz, Congratulations!
00:28 thiago are you sure?
00:29 rwp thiago, Are you commenting upon my suggestion? If so then I will say possibly. I cautioned that I am not a git expert in this channel. But that is from probably five year old notes on doing that operation.
00:30 rwp git has been evolving at a rapid pace. Sometimes the way things are done change rapidly too.
00:30 thiago yes
00:30 satbyy rwp: thiago: 'remote rm' worked, but 'remote add' fails
00:31 thiago fails how?
00:31 Noodlewitt joined #git
00:31 satbyy > git@gitlab.com:satbyy/satbyy.gitlab.io.git
00:31 satbyy > fatal: remote origin already exists.
00:31 satbyy I just 'rm'ed it!
00:31 rwp Honestly I tried those commands in a repository here before I pasted them. Worked for me.
00:31 thiago are you sure that was after the rm and nothing in-between?
00:32 rwp After the "git remote rm origin" then "git remove -v" should show nothing at all.
00:32 rwp Then with no origin one can add one back in with "git remote add origin $URL"
00:32 rwp Be sure to use the URL you have for your own upstream. Basically what was previously shown in your previous "git remote -v" and other output.
00:33 satbyy yes, just check this: http://paste.debian.net/904879/
00:33 satbyy The URL is same as it was in git remote -v
00:34 rizonz rwp: indeed!
00:34 rwp Hmm...  Sorry.  I don't know why things are so unhappy for you.  At that point if it were me I would git clone a new fresh checkout in a differnet directory.
00:34 rwp In that new fresh clone everything should be in a completely happy state.
00:35 rwp It would feel like defeat but it should definitely work.
00:35 nowhere_man joined #git
00:35 rwp Unless the more knowledgeable thiago has a better suggestion?
00:35 rawtaz satbyy: but what does `git remote -v show after you do the git remote rm origin command?
00:36 arescorpio joined #git
00:36 satbyy git remote -v
00:36 satbyy > origin
00:36 satbyy Just prints 'origin' without the URL
00:36 rwp If that shows "origin" as an output them the "git remote rm origin" failed.
00:37 rawtaz satbyy: well cant you remove that origin?
00:37 rawtaz that remote*
00:37 satbyy rawtaz: ... how exactly ? git remote rm ?
00:38 satbyy trying it again throws up: "error: Could not remove config section 'remote.origin'"
00:38 rawtaz ok
00:38 rawtaz so i have no idea but you could try hacking the .git/config file :D
00:38 rawtaz open it and see what might be wrong
00:39 rawtaz here's what one of mine look like:  https://pastebin.mozilla.org/8956642
00:39 rawtaz i dont know if remotes are stored somewhere else too
00:40 satbyy rawtaz: aww thanks. Was missing "fetch" URL. Adding it worked!
00:41 bronson joined #git
00:41 rawtaz nice :)
00:41 satbyy thanks to rwp as well... just gotta remember when it happens next time
00:41 rawtaz i search for it in my restic repo and there were very few hits for "origin" there, probably nothing that is affected by you adding this line
00:41 rwp I don't know why "git remote add $URL" didn't set that up for you. It should.
00:42 rawtaz i guess it was perplexed by finding the origin already existing :>
00:42 rawtaz might have felt betrayed, even
00:42 thiago ok, so stop trying to remove origin
00:42 thiago we've concluded that's not good advice to give
00:43 satbyy haha
00:43 rwp There are several components that are like aircraft flying in close formation. When everything works then everything works. But if things get out of sync sometimes they need to be put back in sync first.
00:43 rawtaz satbyy: what were you trying to do anyway by removing origin?
00:43 tyriondwarf joined #git
00:43 rwp That was my suggstion.
00:43 rawtaz to solve some other fancy problem i take it? :)
00:43 rwp When I have had things get completely out of sync I have found it good to remove origin completely.  Then add it back in so that everything is clean.
00:44 rawtaz out of sync how? just curious
00:45 rwp rawtaz, satbyy had posted http://paste.debian.net/hidden/aebda440 and I commented that it laced a "Remote branches" section.
00:45 thiago it's fairly hard to get things out of sync in such a way that removing origin helps
00:45 rawtaz rwp: ok
00:45 rwp thiago, You would be surprised.
00:46 rawtaz i never had something like that happen. out of sync how?
00:46 rawtaz like, out of sync with what :P
00:46 * thiago is curious too
00:46 rwp Frequently some people will check out from a git:// anonymous read-only repository and then try to push. Realize that they can't. And then hork things up trying to fix it.
00:47 thiago git remote set-url fixes that
00:47 GavinMagnus joined #git
00:47 rwp Except that sometimes it does not.
00:47 thiago sure it does
00:47 justanotheruser joined #git
00:47 thiago if the problem was the URL, changing the URL fixes it
00:47 rwp Note that I have never had this problem myself. This is only other people that seem to have such problems.
00:47 thiago I'm sure they've done something else
00:48 rwp They probably have.
00:48 thiago they've probably compounded the problem, which requires more fixes to remove the problems they introduced while trying to fix the original problem.
00:48 rwp Which is why when I suggested it I likened it to a full system reboot.
00:48 thiago rm -rf && clone again
00:48 rwp I suggested that too.
00:48 satbyy in my case it went like this: (1) Have a local repo already (2) create new repo from gitlab (3) git remote add origin (4) git push -u origin master
00:49 rawtaz im glad im not an advanced git user, cuz it sems that the things i do dont screw up as much :P
00:49 thiago satbyy: I don't see how that would be a problem
00:49 satbyy After that, I get-pull but notice that origin/master doesn't exist
00:49 thiago or would cause a problem
00:50 GavinMagnus left #git
00:51 gozo joined #git
00:51 rwp And the above information is why I suggested making a clean origin of things. Since all objects in the object store will still be there it is a fast trivial operation to remove and re-add it back so that all associations are reset clean again.
00:52 rwp You may not like that. Which is fine. But then please feel free to jump in with a better suggestion.
00:52 Emperor_Earth joined #git
00:53 sevenever joined #git
00:53 thiago removing the origin wouldn't solve any problem I've heard here
00:53 thiago I don't see how satbyy's commands would cause a problem in the first place
00:53 thiago more than likely, the problem is in something he didn't say, which means the solutions we've provided didn't apply
00:54 rwp Great! What would you suggest to debug it and then fix it?
00:54 thiago first, identify the problem
00:55 rwp I think we are going in circles.
00:56 rwp In the meantime, why are you so adamently opposed to "rm origin" "add origin $URL"?  It is fast and trivial and does reset many things.
00:57 rwp Yet you act as if it is actively damaging. What damage is done by it?
00:57 Anja joined #git
00:58 Gsham joined #git
00:59 satbyy Huh, I tracked down the problem
00:59 satbyy It was this line in ~/.gitconfig
01:00 satbyy [remote "origin"]
01:00 satbyy tagOpt = --tags
01:01 satbyy So even when I 'git remote rm' Git implicitly thinks that there is an origin because I had these lines!
01:01 rwp Seems plausible! I am glad you found that.
01:02 osse That is a bug. What is the git version?
01:02 satbyy 2.11.0
01:03 rwp It is probably from https://subgit.com/remote-book.html which mentions it.
01:03 osse Huh
01:03 pbrewczynski joined #git
01:03 tyriondwarf hey guys, new to git, started a new site.  i chose gitlab, free and tons of private repos.  do i need github or can i bypass it
01:03 rwp osse, Huh to which of us?
01:04 tyriondwarf and git the software should work with gitlab correct?
01:04 rwp I searched for satbyy's configuration and found it documented in the URL I pasted.
01:04 phroa tyriondwarf: github is a third-party site, you don't need to use it if you don't want to
01:04 phroa (just like gitlab is)
01:05 rwp tyriondwarf, Both github and gitlab are management wrappers around git as an upstream service. You can use either one, or the other, or both, or neither and use something different entirely.
01:05 tyriondwarf ok great.  im trying to figure out how to manage changes to my site
01:06 tyriondwarf how that works.  say i upload my site from repo to server.  then make tweaks to templates and whatnot.  how do i upload those changes without reupping the whole project
01:07 bremner !deploy
01:07 gitinfo Git is not a deployment tool, but you can build one around it (in simple environments) or use it as an object store(for complex ones). Here are some options/ideas to get you started: http://gitolite.com/deploy.html
01:07 bremner tyriondwarf: ^
01:10 satbyy osse: should i report the bug somewhere?
01:12 tyriondwarf interesting. maybe ill just use gitlab to hold backups of my live site and make new ones every 2 weeks
01:19 stick-shift joined #git
01:20 WayToDoor joined #git
01:24 mattcen joined #git
01:35 [log] joined #git
01:36 kyle_ joined #git
01:36 madewokherd` joined #git
01:39 Noodlewitt joined #git
01:41 bronson joined #git
01:41 mizu_no_oto joined #git
01:42 JamesBaxter joined #git
01:43 pothepanda joined #git
01:44 ejb joined #git
01:45 polyrob joined #git
01:49 duderonomy joined #git
01:52 jimi_ joined #git
01:55 snowkidind joined #git
02:04 unreal joined #git
02:06 pks joined #git
02:09 annoymouse joined #git
02:10 finalbeta joined #git
02:15 Noodlewitt joined #git
02:15 diogenese joined #git
02:21 pks joined #git
02:22 Raed joined #git
02:24 andrew710 joined #git
02:25 d^sh_ joined #git
02:26 xall joined #git
02:27 arescorpio joined #git
02:33 Sonderblade joined #git
02:34 looper joined #git
02:35 enleeten_ left #git
02:36 enleeten_ joined #git
02:36 tyriondwarf left #git
02:37 NeverDie joined #git
02:38 jimi_sanchez joined #git
02:38 Goplat joined #git
02:40 cagedwisdom joined #git
02:41 Vampire0_ joined #git
02:41 AlexMax left #git
02:42 bronson joined #git
02:46 NeverDie_ joined #git
02:49 ilbot3 joined #git
02:49 Topic for #git is now Welcome to #git, the place for git help and serious business | Public logs at http://jk.gs/glog | First visit? Read: http://jk.gs/git | Current stable version: 2.11.0 | Getting "cannot send to channel"? /msg gitinfo .voice | The git-jokes project has been suspended for lack of contribution. Help revive it! https://madeitwor.se/git-jokes
02:49 ResidentBiscuit joined #git
02:55 dodobrain joined #git
02:55 menip joined #git
02:56 dodobrain joined #git
02:56 dodobrain joined #git
02:58 aavrug joined #git
02:59 svm_invictvs joined #git
03:03 svm_invictvs joined #git
03:06 Noodlewitt joined #git
03:11 chachasmooth joined #git
03:14 dsantiago joined #git
03:17 nydel07 joined #git
03:21 pks joined #git
03:23 moestevens joined #git
03:26 Crell joined #git
03:31 Vortex34 joined #git
03:33 nd_ joined #git
03:36 Crell Hi folks. When dealing with a bare repository, is this not the correct way to create a new branch: git update-ref refs/head/newbranchname abc123hash ?
03:36 Crell When I run that, I get no output at all but no branch has been created.
03:37 ToBeCloud joined #git
03:38 hexagoxel joined #git
03:42 machinly joined #git
03:42 jshyang joined #git
03:43 bronson joined #git
03:52 sunri5e joined #git
03:54 Noodlewitt joined #git
03:56 leeN joined #git
03:59 svm_invictvs joined #git
03:59 a_thakur joined #git
03:59 a_thakur joined #git
04:00 Noodlewitt joined #git
04:02 Limes_ joined #git
04:03 justanotheruser joined #git
04:08 gwozt joined #git
04:10 svm_invictvs joined #git
04:11 mattcen joined #git
04:11 Guest55 joined #git
04:20 fstd_ joined #git
04:21 APNG joined #git
04:22 APNG should github projects include a gitignore?
04:22 filter #github
04:22 APNG why shouldn't I force all new contributors to setup a .git/exclude?
04:22 filter because you want contributors
04:23 APNG so how am I supposed to argue with these things? https://github.com/FyberOptic/Meddle/issues/1
04:24 APNG altho tbh they hate me so I doubt any amount of arguing is gonna convince them otherwise
04:25 manuelmrtz joined #git
04:26 APNG the MC modding community, the only community which is as painful as shoving a cactus up one's ass :/
04:27 filter I'd just downvote the liach guy
04:27 filter and move on with my day :p
04:29 filter or put a confused reaction
04:30 APNG I wish github supported the finger
04:30 filter heh
04:32 preaction if you know your build process creates files or directories that should never be committed to git, use gitignore
04:34 ayogi joined #git
04:35 aruns joined #git
04:35 Rodya_ joined #git
04:37 BSaboia joined #git
04:44 bronson joined #git
04:45 subhojit777 joined #git
04:50 eizua joined #git
04:50 Hink joined #git
04:51 InfoTest joined #git
04:52 sbulage joined #git
04:56 aruns joined #git
05:02 Mowgli joined #git
05:05 QualityAddict joined #git
05:05 Noodlewitt joined #git
05:12 algun joined #git
05:13 khagler joined #git
05:15 Cabanossi joined #git
05:15 pbrewczynski joined #git
05:16 sevenever joined #git
05:21 Raging_Hog joined #git
05:23 tojo joined #git
05:26 WhiskerBiscuit joined #git
05:29 svm_invictvs joined #git
05:29 dave0x6d joined #git
05:30 subhojit777 joined #git
05:30 jozwior joined #git
05:38 youser joined #git
05:40 NeverDie joined #git
05:45 bronson joined #git
05:47 pks joined #git
05:49 xall joined #git
05:49 ResidentBiscuit joined #git
05:51 GT4066 joined #git
05:53 plos joined #git
05:53 Garen joined #git
05:55 NeverTired joined #git
05:56 aidalgol joined #git
05:57 overlord_tm joined #git
05:59 cqi joined #git
06:02 Sleepy63 joined #git
06:05 cqi joined #git
06:15 pks joined #git
06:16 d0nn1e joined #git
06:19 jozwior joined #git
06:19 cagedwisdom joined #git
06:23 somaReverse joined #git
06:24 _ng joined #git
06:25 cyphase joined #git
06:31 Mowgli joined #git
06:38 Macaveli joined #git
06:40 ayjay_t so if an upstream repo updates a submodule, will pulling my main repo pull the submodule updates too? or do i have update the submodule myself
06:43 navidr joined #git
06:46 bronson joined #git
06:47 cyan__ joined #git
06:49 ThomasLocke joined #git
06:50 [log] joined #git
06:52 dminuoso left #git
07:00 salamanderrake joined #git
07:00 dermoth joined #git
07:01 Anja joined #git
07:01 freimatz joined #git
07:05 JeroenT joined #git
07:05 watabou joined #git
07:15 lb1c joined #git
07:16 cqi joined #git
07:18 jozwior joined #git
07:19 diogenese joined #git
07:19 t-mart joined #git
07:22 [log] joined #git
07:23 robattila256 joined #git
07:23 NimeshNeema joined #git
07:23 xall joined #git
07:25 Kartagis joined #git
07:25 Kartagis ehlo
07:25 sbulage joined #git
07:25 Kartagis remind me please, how do I take back my commit? git reset?
07:26 Macaveli joined #git
07:26 Kartagis I seem to have forgotten to update my credentials
07:31 t0n1c joined #git
07:32 t0n1c joined #git
07:34 Kartagis git format-patch
07:34 Kartagis !git format-patch
07:34 howitdo joined #git
07:36 paranoicsan joined #git
07:38 JeroenT_ joined #git
07:40 zefferno joined #git
07:40 sea-gull joined #git
07:42 varuog joined #git
07:43 vuoto joined #git
07:44 ferr1 joined #git
07:46 tojo joined #git
07:46 Guest49808 left #git
07:47 bronson joined #git
07:47 dikiaap joined #git
07:48 salamanderrake joined #git
07:48 rchavik joined #git
07:52 derk0pf joined #git
07:53 nowhere_man joined #git
07:53 Ratler joined #git
07:53 Ratler joined #git
07:56 cyan__ joined #git
07:59 esc joined #git
07:59 esc joined #git
08:02 Singmyr joined #git
08:07 [0xAA] joined #git
08:10 diogenese joined #git
08:18 Mir4g3 joined #git
08:20 Atrumx joined #git
08:22 danslo joined #git
08:26 anddam hello
08:28 zefferno joined #git
08:28 anddam 'git config credential.helper' returns 0 and prints 'osxkeychain' on my system (Mac OS) but any of --global, --system or --local (I'm in a repo) returns -1 and prints nothing
08:29 anddam from what "file-option" is the former command reading value for credential.helper if not any of those I specified?
08:30 anddam also I see the 1 status means 'The section or key is invalid (ret=1),', why is it invalid?
08:32 zefferno joined #git
08:33 selckin strace -eopen git config credential.helper'
08:33 anddam I see 'git config --list --system' and '--local' fail (I'm not in a repo now) since the relative files don't exist
08:33 anddam does macos have strace?
08:33 selckin probably
08:33 anddam IDK that
08:33 anddam nope
08:34 selckin think its like /etc/gitconfig $XDG_CONFIG/git/config $HOME/.config/git/config ~/.gitconfig $GIT_DIR/config
08:34 anddam yes, I read the FILES section in git-config(1)
08:34 anddam but I don't get where the credential.helper config is coming if not from global, given the other two don't exist
08:35 selckin dtruss -f -t open git config
08:36 anddam nope, the Apple SIP protection thing I guess 'dtrace: failed to execute git: dtrace cannot control executables signed with restricted entitlements'
08:36 selckin try owning your own pc
08:38 zefferno joined #git
08:40 roentgen joined #git
08:40 a_thakur joined #git
08:46 glebihan joined #git
08:46 _rht joined #git
08:46 Balliad joined #git
08:48 bronson joined #git
08:48 xall joined #git
08:48 Panacea joined #git
08:52 davimore_ joined #git
08:56 languitar joined #git
08:57 RxMcDonald joined #git
08:58 RxMcDonald Hello, anyone knows the best way to merge a huge amount of refactoring ? the code hasn't changed it's only the coding style but it's changed 80% of the codebase in a 20k line project
08:58 sea-gull joined #git
08:59 BSaboia joined #git
08:59 RxMcDonald Changes are in a different branch and now I need to merge with master that is like 100 commits ahead and has changed like 20 files (of 200 files)
08:59 RxMcDonald So maybe I can make this easier if I only manually resolve the merge conflicts of those changed files and not the whole thing?
09:01 selckin this is where you learn to never do that again
09:01 zefferno joined #git
09:01 j416 RxMcDonald: rebase often, not after 100 commits
09:01 RxMcDonald selckin: yes I know, I didn't do it on purpose I realize it was a mistake afterwards
09:02 RxMcDonald j416: too late
09:02 selckin is it only reformatting or actual changes?
09:02 RxMcDonald only reformatting
09:02 RxMcDonald the thing is that master is now like 100 commits ahead with actual functionality changes, but only in 20% of the files
09:02 j416 I would rebase, take the version of master, then re-style that
09:03 j416 there won't be conflicts in files that weren't changed by the other side.
09:03 selckin indee
09:03 selckin +d
09:03 zefferno joined #git
09:04 RxMcDonald so, be a bit more specific? I'm not a git expert
09:04 j416 RxMcDonald: git checkout yourbranch && git rebase master
09:04 selckin don't resolve the conflicts, just take the master version, and do the reformatting again on those files
09:04 marcogmonteiro joined #git
09:04 j416 ^ sounds safest.
09:05 selckin i've tried it a few times, never works out
09:05 j416 what doesn't work out
09:05 selckin trying to merge
09:05 RxMcDonald selckin: so, what exactly? merge ours?
09:06 j416 I never merge when there are conflicts; makes for hard-to-read history
09:06 PCatinean joined #git
09:06 j416 perhaps very tiny conficts, but not otherwise
09:07 jnavila joined #git
09:07 selckin usually you have automated tools for the reformatting & refactoring, its gonna be easier just to start from master and do them again
09:12 RxMcDonald joined #git
09:13 Jackneill joined #git
09:15 jozwior_ joined #git
09:16 zefferno joined #git
09:19 stick-shift joined #git
09:20 [log] joined #git
09:24 sens joined #git
09:26 netj joined #git
09:31 netj joined #git
09:33 mkalfon_ joined #git
09:38 mar77i joined #git
09:41 Jackneill joined #git
09:45 g__k__ joined #git
09:45 a_thakur_ joined #git
09:49 bronson joined #git
09:52 aw1 joined #git
09:53 anli joined #git
09:53 anli I change the url of a submodule, modifying .gitmodules and then git sync, do I have to use git pull as well?
09:54 anli oops, I meant git submodule sync
09:56 Kartagis left #git
09:57 jknetl joined #git
10:02 TomyLobo joined #git
10:03 YuGiOhJCJ joined #git
10:03 rnsanchez joined #git
10:08 shams joined #git
10:09 miczac joined #git
10:09 robotroll joined #git
10:09 GavinMagnus joined #git
10:10 howitdo joined #git
10:11 Suit_Of_Sables left #git
10:22 kexmex joined #git
10:22 pluszak joined #git
10:23 pluszak on branch develop git pull origin merges origin/develop. Why does "git merge origin" tries to merge origin/master?
10:23 pluszak What does "git merge origin" even tries to do?
10:24 selckin git rev-parse origin
10:25 iamtakingiteasy joined #git
10:25 BSAlb joined #git
10:26 selckin 6. otherwise, refs/remotes/<refname>/HEAD if it exists.
10:26 selckin i'm guessing that
10:26 chll_ joined #git
10:31 masuberu joined #git
10:34 cdcdc joined #git
10:34 anli I have checkout out a github branch, but the listing does not resemble the expected one
10:35 selckin what command did you run
10:35 anli git checkout thebranch and git pull .
10:35 anli Maybe I should check the trees and where it becomes different
10:35 anli The clone is a submodule
10:35 selckin thats not how you checkout a remote branch
10:35 tobiasvl what do you mean the listing does not resemble the expected one?
10:35 anli I have change the url of it
10:35 selckin git checkout -b thebranch origin/thebranch
10:36 anli The commits are not the same
10:36 anli Its other commits, with overlapping time from the expected ones
10:36 anli Git branch shows that I am currently into that branch
10:37 tobiasvl what selckin said. you just created a new branch from whatever branch you were on, and merged something in
10:37 anli It also displays master, but the asterisk is on the branch, named tmpsantos-qt_vendor_gcc4.9
10:38 anli The origin url should be correct
10:38 anli git remote shows origin
10:39 anli git remote get-url origin displays the expected url
10:39 selckin !summary
10:39 gitinfo [!dont_summarize] Please do not summarize what you did; !pastebin _exactly_ what commands you ran, what the output was, and why you think that this is wrong. Otherwise you may hide the real issue. See also !obfuscate and !repro
10:39 selckin you somehow think you did what i said, but thats not what you said you ran
10:39 anli No, I do not think I did that
10:40 anli I just did it
10:41 tobiasvl what did you do and what do you think that would result in
10:41 tobiasvl you said it did not resemble "the expected one", but not what you expected
10:41 JeroenT joined #git
10:42 t0by joined #git
10:42 t0by joined #git
10:42 anli That is that the listings are different
10:42 anli Nothing is vague with the explanation.
10:43 anli There is no commit in one of the listings that equals a commit in the other listing
10:43 rasto joined #git
10:43 anli Where one of the listings are the local one, the other the github one
10:43 a_thakur joined #git
10:43 tobiasvl that is to be expected. as selckin said, that's not how you checkout a remote branch
10:43 tobiasvl so you did not checkout the branch you assume you did, then
10:44 anli That must be it
10:44 shiriru joined #git
10:45 mazert joined #git
10:46 anli After I changed the .gitmodule file in the parent dir, changing url, then git submodule sync and cd thediir and git pull ., I got this outcome
10:46 anli git pull . again shows "Already up-to-date"
10:46 Mowgli joined #git
10:47 anli I wish github could show me a tree instead
10:48 anli Also, it seems like search for hashes failed
10:50 bronson joined #git
10:52 nisseni joined #git
10:54 JeroenT joined #git
10:56 languitar joined #git
10:59 yetithefoot joined #git
10:59 yetithefoot joined #git
11:02 languitar joined #git
11:03 languitar joined #git
11:05 willlondon joined #git
11:08 nowhere_man joined #git
11:09 GavinMagnus left #git
11:14 re1 joined #git
11:21 BlessJah joined #git
11:21 thiago joined #git
11:25 Cavallari joined #git
11:28 manuelschneid3r joined #git
11:28 paladinn joined #git
11:32 mar77i joined #git
11:33 pluszak selckin: but why would refs/remote be different for pull and merge?
11:35 selckin pull = fetch + merge
11:35 selckin (or in short, i dunno)
11:36 osse 'git pull .' will pull from the current repo
11:37 thiago and since that will try to merge the current repo's current branch to the current branch, it's a no-op
11:38 thiago but git pull . branchname is exactly the same as git merge branchname
11:38 gareppa joined #git
11:44 JeroenT joined #git
11:46 cyphase joined #git
11:46 selckin its about git pull origin  => git merge origin, which merges refs/remotes/<refname>/HEAD, not sure what pull does with that
11:46 shubhnik joined #git
11:46 thiago not exactly
11:47 selckin well that was my guess ferom the docs
11:47 selckin :)
11:47 thiago the <refname> in this case is origin
11:47 thiago git pull origin will merge origin/HEAD
11:48 selckin feel like its always wrong to pass orirgin/branch to pull command
11:48 thiago right
11:48 selckin at least very advanced usage you normally don't need
11:49 tango_ joined #git
11:50 ertesx joined #git
11:50 bronson joined #git
11:53 _jan joined #git
11:55 Macaveli joined #git
11:57 vicbtmpefczkdnva joined #git
11:58 xall joined #git
11:58 GavinMagnus joined #git
12:00 _jan left #git
12:01 mathematic-alpha joined #git
12:02 GavinMagnus left #git
12:02 _jan joined #git
12:02 _jan left #git
12:07 rivarun joined #git
12:08 rominronin joined #git
12:09 danslo joined #git
12:11 fahadash joined #git
12:11 [Saint] joined #git
12:12 publio joined #git
12:13 jnavila joined #git
12:13 a_thakur joined #git
12:14 MrLawrence joined #git
12:16 fbwnd joined #git
12:17 [Saint] joined #git
12:18 rizonz is there a way to remove al pushes from date X ?
12:18 rizonz or date X until now
12:19 tojo joined #git
12:19 cyphase joined #git
12:22 thiago not exactly
12:22 thiago if you can identify the commit you want to remove, or rather the last commit you want to keep, then you can perform this history !rewrite
12:22 gitinfo Rewriting public history is not recommended. Everyone who has pulled the old history will have to do work (and you'll have to tell them to), so it's infinitely better to just move on. If you must, you can use `git push --force-with-lease <remote> <branch>` to force (and the remote may reject that, anyway). See http://goo.gl/waqum
12:22 arnoldoree joined #git
12:23 nedbat rizonz: do you want to hide that they ever happened, or do you just want master to not have their changes?
12:23 nedbat rizonz: that is, are they secrets, or just not right?
12:24 rizonz nedbat: yeah my privates are in there, I need to have them removed :P
12:24 tobiasvl !sensitive
12:24 gitinfo [!filter_sensitive] You can use filter-branch to remove sensitive data from a repository's history. https://help.github.com/articles/remove-sensitive-data/
12:24 rizonz nedbat: what happened is that my client committed using the wrong username, weird, it uses my windows login name
12:25 rizonz so the changes may be removed
12:29 ddiss joined #git
12:30 mgedmin joined #git
12:31 jknetl joined #git
12:34 rizonz nedbat: ^
12:34 tojo joined #git
12:34 CheckDavid joined #git
12:35 mizu_no_oto joined #git
12:36 ddiss is it possible to use the 'git commit -s' behaviour with a different (e.g. Reviewed-by) tag, based on a environment var or conf?
12:36 fbwnd joined #git
12:36 danslo joined #git
12:36 ddiss i.e. GIT_SIGNOFF_TAG="Reviewed-by:" git commit -s --amend
12:37 grawity it always uses the committer name/email for that
12:37 grawity you might find `git interpret-trailers` useful though
12:38 mgedmin I'm in a bit of a pickle: I'm writing a tool that wants to get a list of versioned files from git (a la ls-files --recurse-submodules)
12:38 mgedmin but I'd like to support git versions before 2.11
12:39 mgedmin so what I'm doing is printing all the submodule roots, and then invoking git ls-files on each
12:39 ddiss grawity: thanks, though my question isn't on the use of name/email, just replacing the "Signed-off-by:" string
12:39 mgedmin and this works wonderfully in Linux, or on Windows, but fails badly in Git Bash on Windows
12:39 ddiss my use case is reviewing commits during interactive rebase...
12:39 ddiss currently I use `x GIT_EDITOR='sh -c "echo \"Reviewed-by: $NAME <$EMAIL>\" >> \"$1\""' git commit --amend` - but it's not ideal
12:39 mgedmin because git submodule --quiet foreach --recursive 'printf "%s/%s\n" $toplevel $path' prints msys2 paths /like/this
12:39 grawity ddiss: add one using git-interpret-trailers
12:40 ddiss grawity: okay, will take a look - thanks again
12:40 mgedmin and I don't know how to ask git to print regular windows paths c\like
12:40 mgedmin \this
12:40 mgedmin (same git does both, depending on whether I invoke it from cmd.exe or from bash.exe)
12:40 JeroenT joined #git
12:41 WaReZ joined #git
12:43 mazert joined #git
12:44 Panacea joined #git
12:47 iamtakingiteasy joined #git
12:47 RxMcDonald joined #git
12:47 thiago mgedmin: the problem is not git, it's bash
12:47 thiago mgedmin: git submodule is a bash script. You have to deal with Bash.
12:48 thiago mgedmin: besides, what's the problem with slashes?
12:49 bket joined #git
12:51 bronson joined #git
12:55 xall joined #git
12:55 danslo joined #git
12:57 ojdo joined #git
13:01 trondth joined #git
13:01 coasterz joined #git
13:01 JeroenT joined #git
13:04 [Saint] joined #git
13:05 synthroid joined #git
13:05 mgedmin turns out it was tox clearing MSYSTEM=MINGW64 from the environment, which made git print paths differently
13:05 mgedmin (tox is a Python test meta-runner that tries to keep things reproducible by, among other things, clearing the entire environment.  ouch)
13:07 byte512 joined #git
13:07 Darcidride joined #git
13:10 danslo joined #git
13:11 doener joined #git
13:13 cyphase joined #git
13:17 Oatmeal joined #git
13:17 dimi1947 joined #git
13:20 Juzzika joined #git
13:24 ddiss left #git
13:25 nisseni joined #git
13:26 sdothum joined #git
13:27 [Saint] joined #git
13:27 MikeChernev_ left #git
13:28 MikeChernev joined #git
13:29 ddiss joined #git
13:30 paws_ joined #git
13:30 ddiss left #git
13:31 Schrostfutz joined #git
13:32 mgedmin (the problem with slashes is that /tmp is a msys2 mount and a non-msys2-program has no idea it's actually c:\users\me\Documents And Settings\Temporary Files or whateverr)
13:33 mgedmin (https://github.com/mgedmin/check-manifest/issues/64 has all the gory details)
13:34 khagler joined #git
13:34 mar77i joined #git
13:42 aavrug joined #git
13:44 Akimb joined #git
13:45 rwb joined #git
13:47 thiago mgedmin: there must be a program that translates those paths to actual Windows paths
13:48 mgedmin yeah, cygpath, but it's better to avoid the necessity by not removing essential environment variables ;)
13:48 sentinal8473 joined #git
13:49 siddart joined #git
13:53 bronson joined #git
13:54 theoceaniscool joined #git
13:55 m0viefreak joined #git
13:56 marcogmonteiro joined #git
13:57 sbulage joined #git
13:58 Shock- joined #git
14:01 guest_ joined #git
14:01 pbrewczynski joined #git
14:02 JrCs joined #git
14:03 Schrostfutz joined #git
14:04 jeffreylevesque joined #git
14:05 rnsanchez_ joined #git
14:10 diogenese joined #git
14:10 nickthetait joined #git
14:10 mathematic-alpha left #git
14:11 jimi_ joined #git
14:12 ShekharReddy joined #git
14:13 LiMp^ joined #git
14:15 mario_ joined #git
14:17 cyphase joined #git
14:17 sentinal8473 joined #git
14:19 anli joined #git
14:20 rominronin joined #git
14:21 pmodin joined #git
14:25 JamesBaxter joined #git
14:25 yehowyada joined #git
14:25 ExoUNX joined #git
14:26 [0xAA] joined #git
14:26 nwkj86 joined #git
14:27 dviola joined #git
14:30 shinnya joined #git
14:31 cyphase joined #git
14:31 Gsham joined #git
14:32 JanC_ joined #git
14:36 davimore__ joined #git
14:37 PCatinean joined #git
14:38 p4trix joined #git
14:38 p4trix joined #git
14:42 neilthereildeil joined #git
14:42 mar77i joined #git
14:43 neilthereildeil if theres a repo where someone checked in code in commit 5, and multiple commits after that (6 7 8), and i dont want commit 5, what are my options?
14:44 neilthereildeil branch at 4 and then port changes to the new branch?
14:44 osse the easiest is  "git revert 5"
14:44 selckin create 9 that revert 5 (git revert 5), or rewrite history to remove it, git rebase -i 5^
14:44 neilthereildeil but the code in 6 7 8 is based on 5
14:45 selckin then it'll conflict and you have to fix it
14:45 selckin AI that write code has been invented yet and will put all of us on the street
14:46 neilthereildeil so the easiest is to add commit 9 where we fix all the changes in 5, even though it would have been easier to reject commit 5 and rewrite it?
14:47 selckin you can revert the whole set of commits too
14:47 neilthereildeil 6 7 8 are OK
14:48 neilthereildeil but they are based on 5 which is bad
14:48 cyphase joined #git
14:48 selckin define based on
14:48 neilthereildeil so i was thining about branching at 4, doing 5' correctly, and then porting 6 7 8 to the new branch
14:48 neilthereildeil its source code, so we kept building on top of it
14:48 selckin so reverting 5 678 and starting over
14:49 neilthereildeil nah i cant do that
14:49 neilthereildeil im not gonna redo 6 7 8 if the changes were good
14:49 selckin why not
14:49 kexmex joined #git
14:49 raphael29_ joined #git
14:49 neilthereildeil too expensive
14:49 selckin either you or i is crazy or can't read, i'll assume me and bug out
14:50 guest__ joined #git
14:50 osse neilthereildeil: the big problem with redoing 5 and porting 6 7 8 is that you have to push --force. that'll create problems for others
14:50 neilthereildeil mine is the only downstream version of this tree
14:50 osse neilthereildeil: if that is not a concern:  git rebase -i 4. mark 5 'edit' instead of 'pick' , amend it. done
14:51 osse but others may have local ones that are based on this one
14:53 bronson joined #git
14:54 DrSlony joined #git
14:55 DrSlony Hello, how do I diff file foo/bar.cc from commit 0731975f with foo/bar.cc from branch woof?
14:55 DrSlony commit 0731975f is on branch master
14:56 osse git diff 0731975f woof -- foo/bar.cc
14:56 Aoya joined #git
14:56 osse or  git diff woof 0731975f -- foo/bar.cc
14:56 DrSlony thanks osse !
14:57 ok91 joined #git
14:58 Es0teric joined #git
14:59 zeks left #git
15:00 DrSlony left #git
15:01 cyphase joined #git
15:07 Rapture joined #git
15:08 Es0teric joined #git
15:09 marcogmonteiro joined #git
15:10 aqc joined #git
15:11 synthroid joined #git
15:11 lightstalker joined #git
15:12 treehug88 joined #git
15:13 tsdev joined #git
15:16 rahtgaz joined #git
15:17 WhiskerBiscuit joined #git
15:20 skunkz joined #git
15:21 DrSlony joined #git
15:22 DrSlony left #git
15:27 dsdeiz joined #git
15:28 manuelschneid3r joined #git
15:31 cdown joined #git
15:33 rnsanchez joined #git
15:35 Rapture joined #git
15:39 Salander27_ joined #git
15:44 leeN joined #git
15:48 boombatower joined #git
15:49 cyphase joined #git
15:50 raffael joined #git
15:54 Jackneill joined #git
15:54 bronson joined #git
15:54 raffael left #git
15:55 monsterjamp joined #git
15:56 monsterjamp Hello
15:56 monsterjamp I've been trying to remove a directory from git history to reduce the time it takes for git clone to work.
15:57 monsterjamp However the files are still there everytime I do git clone.
15:57 selckin !sensitive
15:57 gitinfo [!filter_sensitive] You can use filter-branch to remove sensitive data from a repository's history. https://help.github.com/articles/remove-sensitive-data/
15:58 Jackneill joined #git
15:58 selckin !bfg
15:58 gitinfo A tool designed to remove large files, or passwords from history: https://rtyley.github.io/bfg-repo-cleaner/ (!rewrite applies)
15:59 newbie1 joined #git
15:59 Jackneill joined #git
15:59 monsterjamp I tried filter-branch with no success, I didn't know I could remove folder with bfg.
16:00 selckin git doesn't track folders, only files
16:00 WayToDoor joined #git
16:00 theskill_ joined #git
16:00 selckin !paste
16:00 gitinfo Please use a pastebin for snippets longer than one line. https://gist.github.com and https://vomitb.in both work well.
16:01 selckin https://vomitb.in/3rEMOhRyos
16:01 gareppa joined #git
16:03 Rapture joined #git
16:06 monsterjamp It's gonna be a while before I can test that out :P cloning takes a bit.
16:06 idmistir_ joined #git
16:07 Es0teric joined #git
16:07 manuelschneid3r joined #git
16:09 YuGiOhJCJ joined #git
16:10 clemf joined #git
16:11 zartu joined #git
16:13 idmistir joined #git
16:13 skalpin joined #git
16:15 subhojit777 joined #git
16:17 Lyqyd joined #git
16:17 a_thakur joined #git
16:17 monsterjamp selckin: I get this error when running the filter-branch line
16:17 monsterjamp https://vomitb.in/qwCLMrqgtQ
16:18 selckin out of diskspace?
16:18 monsterjamp Um.. no?
16:19 Oie joined #git
16:19 selckin um check?
16:19 Jackneill joined #git
16:19 monsterjamp I'm not, far from it.
16:19 WaReZ joined #git
16:24 vrlx joined #git
16:24 cdown_ joined #git
16:24 monsterjamp Hm, looks like it was a permission issue. It's fixed now
16:31 cdown joined #git
16:31 monsterjamp I ran the commands, now I'm just waiting for 'git push master --force' to finish.
16:32 selckin have to push all tags and all branches
16:32 looper joined #git
16:33 selckin you can check the size of .git
16:35 aavrug_ joined #git
16:37 cyphase joined #git
16:37 youser joined #git
16:37 nydel07 joined #git
16:38 neilthereildeil left #git
16:41 monsterjamp This is gonna take a while  even though it's not that big of a project:P
16:42 a_thakur joined #git
16:43 courrier joined #git
16:45 jost joined #git
16:47 a_thakur joined #git
16:49 monsterjamp selckin: No luck, I tried to clone it again after pushing to all branches/tags but the size is still about the same (70+MB).
16:49 nisseni joined #git
16:50 LBV_User joined #git
16:51 crayon joined #git
16:52 Akimb joined #git
16:53 relipse joined #git
16:53 relipse i have a list of Changes not staged for commit of modified files and deleted files, how can I stage all of them at once
16:54 selckin monsterjamp: https://stubbisms.wordpress.com/2009/07/10/git-script-to-show-largest-pack-objects-and-trim-your-waist-line/
16:54 selckin relipse: git add -u
16:55 kadoban joined #git
16:55 bronson joined #git
16:56 relipse thank you selckin
16:56 hashpuppy joined #git
16:57 tmsmith joined #git
17:01 xall joined #git
17:04 Mowgli joined #git
17:05 ispn joined #git
17:07 xaviergmail joined #git
17:07 synthroid joined #git
17:20 jnavila joined #git
17:20 monsterjamp selckin: Thanks, apparently there were more binaries that had to be deleted.
17:21 Macaveli joined #git
17:21 monsterjamp joined #git
17:21 monsterjamp left #git
17:21 NeverDie joined #git
17:23 a_thakur joined #git
17:25 aavrug joined #git
17:26 looper joined #git
17:27 algun joined #git
17:28 cagedwisdom joined #git
17:32 Vaelatern joined #git
17:32 LiMp^ joined #git
17:35 sbeller joined #git
17:36 duderonomy joined #git
17:36 overlord_tm joined #git
17:38 WhiskerBiscuit joined #git
17:38 looper joined #git
17:38 a_thakur joined #git
17:40 aavrug joined #git
17:50 PHPanos joined #git
17:54 GavinMagnus joined #git
17:56 bronson joined #git
17:57 GavinMagnus left #git
17:58 a_thakur joined #git
17:58 Starky joined #git
17:59 Starky joined #git
18:00 robotroll joined #git
18:01 kulelu88 joined #git
18:02 jnavila joined #git
18:07 raphael29_ left #git
18:07 aw1 joined #git
18:08 rath joined #git
18:10 freimatz joined #git
18:10 pcre joined #git
18:12 JrCs joined #git
18:15 Juanito[m] joined #git
18:20 [0xAA] joined #git
18:21 g105b joined #git
18:25 jimi_ joined #git
18:26 zeroed joined #git
18:26 zeroed joined #git
18:27 Pursuit joined #git
18:28 _xor joined #git
18:28 bl0m1_ joined #git
18:29 dar123 joined #git
18:31 g105b Hey folks, please could you offer me some guidance? I have a repository that contains a lot of custom rules for a coding styleguide. There is already an automatic style fixer project out there in another well-known repository, so I would like to include it along with my repository, but with some changes made to it. I've noticed that I can't add a submodule
18:31 g105b with changed contents, so I wonder what the recommended approach here was?
18:31 shiriru joined #git
18:32 moritz g105b: maintain a fork of that well-known reository, and add that fork as a submodule?
18:32 sbeller commit in the submodule, then add the commit to the superproject
18:33 sbeller well yeah, you'd want to have a fork such that you can write to it. Or if these are just style fixes, think if it is really needed
18:33 g105b moritz: ah I see, yes that is an acceptable solution.
18:33 g105b sbeller: the project is not just style fixes, it is style fixers.
18:34 jnavila joined #git
18:35 stick-shift joined #git
18:39 little3_14 joined #git
18:41 gigq joined #git
18:46 dreiss joined #git
18:46 shreyasminocha joined #git
18:47 sentinal8473 joined #git
18:48 misha left #git
18:48 tyreld joined #git
18:48 lvns joined #git
18:48 Es0teric joined #git
18:50 shiriru joined #git
18:55 troulouliou_div2 joined #git
18:56 sentinal8473 joined #git
18:57 theskillwithin joined #git
18:57 bronson joined #git
18:58 netj joined #git
18:58 jeffreylevesque joined #git
19:00 kristofferR joined #git
19:01 nidr0x joined #git
19:01 Es0teric joined #git
19:02 sea-gull_ joined #git
19:05 nowhere_man joined #git
19:06 miczac joined #git
19:07 miczac joined #git
19:07 shreyasminocha joined #git
19:09 howitdo joined #git
19:11 miczac joined #git
19:15 jimi_ joined #git
19:18 lacour joined #git
19:18 joshszep joined #git
19:24 PHPanos joined #git
19:25 JelmerD joined #git
19:28 cyrus_mc joined #git
19:29 cyrus_mc Got a question regarding https://git-scm.com/docs/githooks and the core.hooksPath. In my testing this is set in the .git/config file of the repo, but that configuration file isn't preserved across clones
19:29 osse cyrus_mc: there is no way to pass hooks across clones
19:29 osse there is no way to pass a local git config either
19:30 cyrus_mc osse: ah, I thought this option might solve that
19:30 shreyasminocha left #git
19:30 osse cyrus_mc: you can commit hooks as files, and provide a script to install them
19:30 cyrus_mc osse: automatically or manually?
19:31 osse manually
19:32 lightstalker joined #git
19:33 sentinal8473 joined #git
19:34 arnoldoree joined #git
19:35 Sasazuka joined #git
19:36 Sasazuka joined #git
19:36 ResidentBiscuit joined #git
19:39 jstein_ joined #git
19:40 sdtr443w joined #git
19:40 sdtr443w I have a personal repository that has become complete screwed up since 3 days ago. Tons of objects are corrupt and the remote indicates the same.
19:41 sdtr443w I get a lot of data stream errors (unknown compression) and one loose object
19:41 sdtr443w I have a copy of the repo where I'm just whacking them. They're all recent objects, fairly small, but very numerous.
19:41 sdtr443w I think it's getting rid of everything I've done since I switched computers a few weeks ago.
19:41 siddart_ joined #git
19:41 sentinal8473 joined #git
19:45 sdtr443w Hmm in particular it looks like every object dated 12/21 at 1:23PM is corrupt
19:47 xall joined #git
19:49 bmahe joined #git
19:51 Sasazuka joined #git
19:51 satifant joined #git
19:53 hahuang65 joined #git
19:55 thiago satifant: my guess is that you introduced those errors while copying the files for your computer switchover.
19:55 satifant what errors? :)
19:56 sdtr443w The switchover was 3 weeks ago and I had been committing since then
19:56 Eizv joined #git
19:57 sdtr443w Pushing, even
19:57 sdtr443w A few days ago I switched motherboards and had to reactivate Windows.
19:57 sdtr443w Christmas Eve. I had last committed in the 21st as far as I can tell, and today is the first day I tried to commit again
19:57 sdtr443w At this point, everything's call screwed up
19:58 bronson joined #git
19:58 sdtr443w The error I was getting was a whole bunch of data stream errors (unknown compression method)
19:58 sdtr443w unable to unpack [insert checksum here] header
19:59 sdtr443w And then it would tell me that loose object was corrupt
19:59 sdtr443w It would only feed me one at a time.
19:59 sdtr443w I decided to take a copy of smoke all of them. There were a ton of them, but they happened on 12/21 as far as I can see. There were ultimately two timestamps I saw.
20:00 sdtr443w It looks like I can reset my HEAD to the master, which goes back to November.
20:00 sdtr443w And I can just shove all my deltas onto a branch off of that.
20:01 sdtr443w I lose all intermediate progress.
20:01 auq joined #git
20:01 sdtr443w Basically it looks like this branch I was using over the month dropped dead
20:02 coasterz joined #git
20:03 tobiasvl looks like data corruption. either the disk you're on is dying or the "switching of computers" (whatever that means) involved data transfer that failed?
20:03 tobiasvl oh wait hang on
20:03 tobiasvl sdtr443w: what branch are you on?
20:03 tobiasvl it's not master. but is it another branch?
20:03 sdtr443w Yes it's another branch I created off of master in late November
20:03 tobiasvl or are you on a detached HEAD or something weird that's being garbage collected?
20:04 tobiasvl oh ok. not that then.
20:04 sdtr443w Nope not a detached head
20:04 sdtr443w Life would be so easy if it were just a detached head
20:04 sdtr443w The sequence was
20:04 sdtr443w November: created branch on old computer. Worked on it, pushed it as a remote branch.
20:04 sdtr443w Early December: Got new computer parts and migrated. New hard disk
20:04 sdtr443w Mid December: Made more commits and pushes. All is good.
20:05 sdtr443w Christmas Eve: New motherboard since the old one couldn't handle do video like I needed.
20:05 sdtr443w Today: Problems
20:05 sdtr443w The corruption seems to stem from 12/21
20:06 thiago well, then my theory is wrong
20:06 thiago but you have a corrupt repository
20:06 thiago restore from untainted backups
20:07 cdown joined #git
20:07 sdtr443w I don't have a backup from 12/21 so I've basically lost that history.
20:07 sdtr443w It's a personal repo so I'm not about to get reamed by corporate for my poor data retention policies.
20:08 sdtr443w But it'll rekindle my search for some hosting where I could put a repo like this. It's a personal repo but I still don't want to be sharing it so I don't want to GitHub it.
20:08 tobiasvl but hopefully you'll have learned to take backups :) git makes backup easy, at least.
20:09 sdtr443w Every once in a bit I do back it up but I don't really have a solution for having a real remote. Honestly that's what I want.
20:09 synthroi_ joined #git
20:09 thiago any other repository of the same contents are a backup
20:09 sdtr443w Yeah they're on the same machine
20:09 sdtr443w The remote has the same corruption according to this.
20:09 sdtr443w So both the local and remotes have the same checksums for the objects it couldn't read.
20:10 sdtr443w okay I didn't check all 30+ or whatever objects but the first 5 were identical.
20:11 tobiasvl they're both corrupted?
20:11 sdtr443w Yeah I cloned the remote to a new folder and it had exactly the same problems.
20:12 sdtr443w According to fsck anyways. When I cloned it, I was on master there instead of this messed up branch.
20:12 sdtr443w It put my log back in November before the branch either way.
20:12 diogenese joined #git
20:12 sdtr443w I have this all saved out in different folders so I can refer back to it even after this crude repair.
20:14 sdtr443w I honestly though that maybe some checksum based on my Windows account got all messed up. When I switched the motherboard, I had to reactivate Windows with Microsoft.
20:14 sdtr443w And I reinstalled Git Extensions and friends.
20:14 sdtr443w I have a Linux VM and was pondering seeing what it thought of the mess.
20:16 ResidentBiscuit joined #git
20:16 Sasazuka_ joined #git
20:21 nckpz joined #git
20:22 sdtr443w Oh grand I try to force push and get an SHA1 collision
20:22 tobiasvl huh?
20:22 aarobc joined #git
20:22 tobiasvl never seen that before. !repro
20:22 gitinfo [!transcript] Please paste (using https://gist.github.com/ or similar) a transcript ( https://git.io/viMGr ) of your terminal session so we can see exactly what you see
20:24 sdtr443w https://gist.github.com/rockobonaparte/2f7a8b47fe38de921bcf90b3ebc1189a
20:24 sdtr443w I'm not surprised really due to the shenanigans I'm pulling.
20:24 sdtr443w In my case--since it's a personal repo--I may just init a new bare repo and push into there.
20:24 nickabbey joined #git
20:25 robattila256 joined #git
20:27 sdtr443w I just pushed it all into a new bare repo so I have some record of it.
20:27 sdtr443w I noticed in all this a dangling commit pointing to work going back to December 17th.
20:27 sdtr443w It's possible I could have rolled back to just there and kept some more history, but whatever.
20:27 sdtr443w I gotta do some cleaning IRL.
20:28 troppie joined #git
20:28 cdown_ joined #git
20:32 bsswork joined #git
20:33 raspado joined #git
20:36 bsswork I have run (git mergetool) and resolved all conflicts.  I have not yet committed.  I want to view a combined diff (index/staging + both parents).
20:36 bsswork I'm pretty sure this is possible, but I'm having trouble finding it by just doing a internet search.
20:36 bsswork I expected (git diff --cached --cc) to work, but no such luck.
20:37 d0nn1e joined #git
20:38 jnavila joined #git
20:39 Es0teric joined #git
20:39 jimi_ joined #git
20:39 snowkidind joined #git
20:39 fatalhalt joined #git
20:40 Sasazuka joined #git
20:40 anli joined #git
20:41 anli Is there a way to do a partial commit interactively with the git command?
20:41 ResidentBiscuit like `git add -p`?
20:42 anli Never tried, does it list changes sorted by modification time desc?
20:42 causasui joined #git
20:42 anli I am forced into using linux now so there is not tortoisegit
20:42 anli no
20:42 causasui joined #git
20:42 osse .
20:43 ResidentBiscuit I dont know what a partial commit is. git add -p lets you add just chunks of a file to be staged
20:43 ResidentBiscuit You can build your commit however you like
20:43 anli I want to do a commit by first listing changes, then ticking those files I want to commit, then go into the diff of each file and possibly remove some lines or add some
20:43 osse git add -p is the closest you're gonna get
20:43 osse but it's pretty damn close
20:43 anli Would it let you go into a diff viewer?
20:44 osse not directly. put you can edit diffs directly
20:44 anli I want the left hand and right hand view where I can navigate to the next different part with a keystroke
20:45 osse you're not gonna get it from git
20:45 anli I presumed that, but maybe you know if there is some tool that provides that
20:46 rewt git difftool should be able to diff staged vs head
20:46 anli Would be great if it was ncurses based actually
20:46 osse maybe tig can
20:46 rewt uses whatever editor you want
20:46 osse or a vim plugin, possibly combined with git difftool
20:46 anli tig you say
20:46 anli I do not use vim if I do not have to
20:46 osse but give git add -p a try before you go hunting
20:46 anli Must check
20:46 Anja joined #git
20:46 anli Thanks for the tip
20:48 dsdeiz joined #git
20:49 cdown joined #git
20:50 anli when pressing e in git add -p, it would be nice if I could "restore after commit"
20:50 osse what do you mean by that? it won't change the files themselves
20:50 anli Lets assume I have changed two lines, but only one should go into the commit
20:51 anli If I had "restore after commit", I can remove the one I do not want and have the possibility to commit the other line in subsequent commits
20:51 rewt that's what partial commit does
20:51 osse ah s, you do do want the edits to go into the file itself. i don't think that's possible
20:52 rewt it leaves the uncommited part in your working tree
20:52 osse oooh
20:52 osse i mis-mismisunderstood. yes as I said the files themselves don't change
20:52 p4trix joined #git
20:52 nisseni joined #git
20:52 idlemind using the git for windows command line how do i make sure i told it to use openssh and then how do i set it to use a particular key for a particular host?
20:52 anli Partial commit to me could be either, commit two out of three changed files, or commit two out of three changed lines in the file A
20:53 osse git add -p can do both
20:53 rewt git commits are always the whole tree (directory + subdirs + files etc)
20:53 anli Could it also let me delete a change so it actually dissapears?
20:53 osse idlemind: i think all it does is set a value in the global git config. (git config --system --edit).  for the latter, write ~/.ssh/config properly
20:53 anli Lets assume I have some debug nonsense that I want to remove
20:53 osse anli: no
20:53 anli Ah, ok
20:53 idlemind osse nice i'll try it
20:53 idlemind thx
20:54 anli But can git add -p let me skip one file?
20:54 osse yes
20:54 anli Maybe it is what I want after all
20:54 osse when you're prompted, press ?. there's an option to skip directly to the next file. i think it's d
20:55 osse you can also filter it at invocation:  git add -p -- justthisdir/ thatfile.cpp
20:55 anli One drawback is that I will edit diffs, I will not see the whole code mass
20:55 rewt when you git add a file, it adds it only to staging; this is a special area where you can have the files contain whatever you want, and when you commit, the files as they are in staging are what gets committed to the repo
20:56 idlemind osse what should i set in git config --system
20:56 idlemind search for ssh shows nothing currently
20:56 rewt this area is completely separate from you working tree (normal dir where you normally edit files)
20:56 osse idlemind: then i think it will already use openssh
20:56 anli rewt: Thats cool actually
20:57 ResidentBiscuit Yeah the staging area is nice
20:57 anli I will not have a great context editing diffs, but maybe I can get used to it
20:57 anli The rest of the code could be added as # lines
20:58 anli But then I would have to seek for the different lines
20:58 osse it's not very friendly to edit the code directly, no. if I were you I would skip that hunk, finish, then edit the file and run git add -p again
20:58 anli It opened with nano btw, I like that
20:58 ResidentBiscuit You can avoid this problem entirely if you just commit more often
20:58 kadoban Ya it uses whatever editor you tell it to
20:58 anli Thats true
20:58 osse there is a config option to adjust the size of the context, but git add -p currently doesn't respect it. that should be fixed
20:59 anli On the other hand, on my job, a lot of changes will be kept for months, some are not even meant to ever be committed
20:59 ResidentBiscuit Commits are cheap and can be rearranged/squashed/edited/etc. No reason to not commit small chunks at a time to keep track of individual changes
20:59 bronson joined #git
20:59 hoshi_ joined #git
20:59 anli git add -p uses a certain file order?
20:59 rewt anli, if you use git difftool between staging (index) vs head, it will diff the 2 in your favourite editor, and you'll have the whole file for context
21:00 osse anli: i presume it's whatever order git diff prints in. that can be controlled
21:00 kadoban anli: Keeping changes around for months without them being commited sounds bad, unless they're ignored files.
21:00 ok91 joined #git
21:00 idlemind osse it worked just had to remove and re-add the git remote part to reflect the name i used in my .ssh/config thanks
21:00 hoshi_ left #git
21:00 anli kadoban: In my job, we use svn, so there is no decent branch ability
21:00 hoshi joined #git
21:01 anli kadoban: So projects running in parallel will be commits of changes that are old
21:01 anli I am not convinced about git add -p, maybe the world needs another tool
21:01 rewt you can configure the default editor that opens for git difftool with git config ... diff.tool
21:02 anli ok
21:02 kadoban I wonder if there's a way to use a difftool thing directly from add -p or something? I dunno
21:03 osse I don't know of any, but that's why I suggested a git plugin for the editor that is your difftool
21:03 kadoban Ah, yes
21:03 rewt that might be nice... but that would pretty much be a 3-way diff between working tree, staging, and HEAD
21:03 rewt should also be quite possible with git difftool
21:05 osse anli: i think also either gitk or git gui can do this, but maybe that's not your cup of tea
21:05 osse as for the uncommitted stuff, I would make it into a commit that I would always keep on top
21:05 idlemind do you guys use a single public key for development on a project and put that key on your various devices like desktop and laptop or do you use and add a separate key for each device? if so does that make each commit look different or does git only look at it's config for name/email?
21:06 anli osse: hm, maybe
21:06 osse idlemind: SSH keys have nothing to do with Git itself. Git is just an "end user" of whatever is your SSH configuration
21:06 rewt yeah, i always use gitk to preview staging to verify what will be committed
21:07 synthroid joined #git
21:07 ResidentBiscuit I have never used gitk. I should check it out
21:07 unreal_ joined #git
21:08 kadoban I usually use git diff --cached, which I have an alias for
21:08 anli I think gitk must be it
21:08 rewt it's not an all-in git tool, but it's really good for what it does do
21:08 kadoban But I like gitk for seeing branches
21:08 anli I love the idea of using the terminal however
21:09 anli Its a shame a graphical tool like gitk shows me diff output
21:09 ResidentBiscuit I havent found a great git gui. Some are decent, they can be nice for viewing repos with lots of branches, but I just prefer the cli
21:09 idlemind osse so it would read my git config for name / email regardless of machine so in theory it would be better to have an ssh key for each system
21:09 JanC_ joined #git
21:10 anli gitk does not implement a diff viewer, like the one in tortoisegit apparently
21:10 osse idlemind: yep
21:10 idlemind also is it common to use a fake email in your git config? the interwebs seems split on the matter from my googling
21:10 osse afaik it's not common
21:10 anli gitk is very much like a windowed git command
21:11 phroa idlemind: if you don't want to use your own, make a git@ address or use a tag like phroa+git@
21:11 osse e.g. Linux and Git demands you use a full name and email
21:11 kadoban Email addresess are pretty much public at this point. Either you don't use it for anything ever, or you end up on 50 lists within a day.
21:11 howitdo joined #git
21:11 idlemind kadoban osse thanks ya that was kind of my thought ... spamassassin is better than assuming my email would go undiscovered
21:11 kadoban Ya
21:11 anli I installed git-gui, but there is no git-gui command
21:11 ExoUNX joined #git
21:12 martinjlowm joined #git
21:12 kadoban Plus if you use github or anything, you kind of have to use at least *a* real email, AFAIK, for it to work correctly.
21:12 ratdeptrai joined #git
21:12 ResidentBiscuit You can have a fake email, but it uses your email to wire up the commits to your github account
21:12 phroa as long as the email is associated with your account, doesn't have to be the primary, github will know who made the commit
21:12 ResidentBiscuit So if the email is fake, the commits wont get connected to your account
21:13 anli You can always use a mailinator email address
21:13 phroa obviously you can still host a repo with fake emails, but
21:13 anli Disposable
21:13 Guest55 joined #git
21:13 kadoban Can you add an associated email to github without them verifying it? I guess I assumed you couldn't.
21:13 kadoban Ya, could use a disposable one ... but ya w/e. I wouldn't.
21:13 anli mailinator emails can be used for verification, but anyone can read whats mailed to such a mailbox for a couple of hours
21:13 r00twrh joined #git
21:13 idlemind thanks for the input for now i'm just using the ole simple ssh server locally to keep track of some homegrown scripts ... we'll see what it grows into
21:13 phroa I like being able to send patches if I have to
21:14 cbreak git doesn't need a server
21:14 anli Create a gmail account for the single purpose :)
21:14 anli Isnt the git-gui command git-gui?
21:14 phroa it's 'git gui'
21:14 anli ah
21:14 phroa though one named git-gui should exist...
21:15 anli oh, also diff output
21:15 kadoban I hate when OS packages don't make it clear how you actually run the thing after you install it.
21:15 idlemind cbreak i'm aware just a great place to put stuff out off of my single desktop that i can also target from things like a raspberry pi w/o worrying that my pc might be off or my laptop is in my backpack :)
21:15 cbreak it's probably somewhiere in $PREFIX/libexec
21:15 kadoban idlemind: If you get bored sometime, look into gitolite. It's a small step up from just an SSH server in terms of complication, but it offers a lot of cool features
21:16 phroa !host
21:16 gitinfo [!gitolite] Gitolite is a tool to host git repos on a server. It features fine-grained access control, custom hooks, and can be installed without root. Download: https://github.com/sitaramc/gitolite Docs: http://gitolite.com/gitolite/
21:16 anli git gui does not let me edit the commit manually
21:17 idlemind thanks kadoban i'd been surfing some of the different add-ons that was one that had come up
21:17 idlemind brb
21:17 phroa cgit is a nice web interface if you have a program like gitolite to do the repo hosting
21:17 phroa or leave
21:17 kadoban haha
21:17 anli cgit you say?
21:18 m1dnight_ joined #git
21:18 anli I thought about web interface before, but maybe is a weird idea, dunno
21:19 phroa it provides a http repo endpoint as well, though I use it mainly as a way to give people links to code
21:19 anli Is there any tool in linux that has a diff viewer?
21:19 anli Which is two editor windows beside each other for the ones not knowing
21:20 kadoban There's like vimdiff, meld, a few others, if I understand the question correctly.
21:20 kadoban emacs has one I'm sure, no idea what it's called.
21:20 osse diff itself has one! diff -y
21:20 phroa probably diff-mode, which Magit surely extends to infinity
21:20 osse can't edit there though
21:20 anli ok
21:21 anli Is there some way of running tortoisegit in linux? :)
21:21 kadoban Oh ya, diff -y, good call.
21:21 osse probably a "gnuism" for those concered with such things
21:21 anli Hm, bad idea
21:22 martinjlowm Hi! Can someone confirm that the --recursive handle for the submodule foreach command is broken? I have a simple example with a nested submodule and doing 'echo "test"', recursively, doesn't proceed to nested modules. I'm using git 2.11.
21:23 anli Maybe the file order can be specified with git add -p
21:23 anli So files modified later will go first
21:24 a_thakur joined #git
21:24 sbeller martinjlowm: how so?
21:24 martinjlowm The shell command is only executed for first level submodules
21:25 Sasazuka joined #git
21:25 osse anli: make a script in PATH called "git-anli" which finds all modified files (git diff --name-only [-z]), sorts them, then calls git add -p -- {sorted list}
21:25 osse then you run ''git anli'' from your shell
21:25 anli Hm
21:25 anli Do I have to do that? Maybe I should then
21:25 anli But the order could not be specified using switches then?
21:26 osse you can specify an order for git diff, but only in form of glob patterns
21:26 martinjlowm In my case I have two submodules, where one of them has one nested submodule. Doing 'git submodule foreach --recursive 'echo "test"'' only executes twice
21:26 sbeller martinjlowm: reading the source code, this *ought* to work ;)  maybe the output is broken?
21:27 sbeller could you test again with e.g. "touch"
21:27 sbeller to create a file inside each submodule?
21:27 martinjlowm sure, give me s ec
21:27 sbeller Looking at the code (https://github.com/git/git/blob/master/git-submodule.sh#L286)
21:28 sbeller you do not have whitespace in the paths? (line 332 needs quoting)
21:28 nickabbey joined #git
21:28 jstein_ joined #git
21:29 nickabbey joined #git
21:29 anli Can anyone tell me why there is still something like "mailing lists"?
21:29 nickabbey joined #git
21:29 anli How did that even get popular in the first place
21:30 osse because people are too old to rock'n'roll but too young to die
21:30 anli hahahaha
21:30 martinjlowm sbeller: oh, nevermind. I realize now, that the nested submodule wasn't initialized
21:30 anli Thats the answer I was seeking for
21:30 ratdeptrai joined #git
21:30 nickabbey joined #git
21:30 martinjlowm sbeller: thanks for the help, though :)
21:30 sbeller martinjlowm: thanks for asking, I think I found a bug nevertheless :)
21:31 nickabbey joined #git
21:34 osse anli: The real answer I think is that it works for the people in charge, and they have their set ways. Do transition to something else someone would need to implement it and maintain it.
21:35 osse *To transition.
21:35 osse Search the git list for "bug tracker" and you'll probably find a thousand threads on the issue
21:36 osse And by "threads" I mean "bunch of emails where they got luck with the in-reply-to header"
21:37 osse 5
21:38 murr4y joined #git
21:39 nyuszika7h joined #git
21:39 Sasazuka joined #git
21:40 ResidentBiscuit joined #git
21:42 martinjlowm sbeller: maybe I have too! I would think the expected behavior of $path is to hold the value of the relative path from the main project to the entered submodule (using --recursive). That is not the case, however. It's only relative to its first parent.
21:43 sbeller martinjlowm: yeah that is the other issue. (I was mostly concerned about a white space in a path)
21:43 sbeller Reading the man page it is quite unclear what it actually means. When reading it literally the man page and the behavior is in sync
21:43 mozzarella joined #git
21:43 sbeller So I am not sure if we want to change it and call it a bug fixed or rather introduce a new variable ($spath maybe?) that is the relative path from the superproject, where the command was invoked down to the actual submodule
21:44 dweelie joined #git
21:45 relipse how can i list all the modifed files with date modified with git status
21:45 dweelie joined #git
21:45 martinjlowm yeah, it depends on the meaning of superproject
21:46 jagob joined #git
21:47 martinjlowm for now I guess $toplevel and pwd will do
21:47 sbeller martinjlowm: in case you want to get involved (I assume you do :P), you could write this as a bug report to the git mailing list git@vger.kernel.org or even make a patch for it.
21:48 Atm0spher1c joined #git
21:49 martinjlowm sbeller: hehe, I started the day off by wanting to perform `git archive' recursively, which didn't seem to be supported. I suppose I can write up a bug report and send it to the mailing list. I guess a discussion around the issue would be best to start with
21:50 sbeller yeah archive also needs some submodule love
21:51 dweelie joined #git
21:52 glebihan_ joined #git
21:54 davideb joined #git
21:55 thiago can't you just use a script, with git submodule foreach?
21:55 sbeller thiago: I assume you'd want the submodule contents inside the one archive for the superproject?
21:58 thiago yes
21:58 thiago I was thinking you could concatenate the tarballs
21:59 thiago but it looks like that doesn't work
21:59 thiago git submodule --quiet foreach 'git archive --prefix=$path/ --format=tar HEAD'
22:00 bronson joined #git
22:00 sdothum joined #git
22:00 cyan__ joined #git
22:00 sbeller so you'd have to untar and retar it to another big tar ball... I'd guess soem people rather have it in one handy command (archive --recurse-submodules)
22:01 thiago if you could concatenate tarballs like that, the command above would have been enough
22:01 dweelie joined #git
22:02 Vaelatern joined #git
22:04 TomyLobo joined #git
22:07 robotroll joined #git
22:08 fstd_ joined #git
22:11 pks joined #git
22:13 martinjlowm true, but I would expect a recursive behavior to be the default
22:13 tvw joined #git
22:13 martinjlowm .gitattributes can be used to ignore submodules that shouldn't be archived
22:14 martinjlowm submodules are often dependency projects and are probably required for most projects
22:15 martinjlowm I think it's silly to leave them out by default
22:17 davideb joined #git
22:17 [log] joined #git
22:17 mozzarel1 joined #git
22:18 smithbone_ joined #git
22:18 JanC_ joined #git
22:19 grawity joined #git
22:19 jnavila_ joined #git
22:19 sbeller martinjlowm: the default arose from lazyness ;) and then history cemented the default.
22:19 aidalgol joined #git
22:20 Akimbas joined #git
22:20 yanome joined #git
22:20 nettoweb joined #git
22:21 martinjlowm hehe ;)
22:21 nydel07 joined #git
22:24 gigq joined #git
22:27 prg3 joined #git
22:31 relipse how can i list the date last modified for each modified file in git status (not untracked files)
22:31 rizonz something I don't get. I have removed files from my local repo which I comitted locally but want to revert back to the master repo for these files, what do I do ?
22:31 rizonz pulling or fetching doesn't work
22:32 martinjlowm git checkout origin/master ?
22:32 unreal joined #git
22:33 martinjlowm assuming your master repo is called origin
22:33 Sasazuka joined #git
22:34 rizonz yap
22:34 Sasazuka joined #git
22:36 j416 relipse: ls -l $(git diff --name-only)
22:37 dsdeiz joined #git
22:37 dsdeiz joined #git
22:38 nivag joined #git
22:39 nisseni joined #git
22:40 nettoweb1 joined #git
22:40 rizonz martinjlowm: I keep getting the modified local files back
22:43 nwkj86 joined #git
22:44 martinjlowm isn't that what you expect? Or are they modified with your local changes (that you had committed)?
22:47 martinjlowm Was this your workflow: git add modified_file; git commit -m 'committed modified_file'; git checkout origin/master? can you confirm that you committed on your local master branch and not origin/master?
22:49 rizonz martinjlowm: workflow was, local edit, commit -> push and on the remote there has been placed back a backup which I want to get back to my local repo, so actually I want to get rid off my old local commits
22:50 bakers joined #git
22:51 bakers Is there a way to do merge in a branch... but keep all my local changes. This merge is going to make a TON of conflicts... so I don't want to actually merge any code
22:51 bakers just make future merges notice that those pieces are already included
22:52 bakers Something like git merge, git checkout --ours on all the files, and then commit
22:52 thiago git merge -s ours
22:52 thiago it discards all incoming changes, but records a merge
22:52 prawnzy joined #git
22:52 bakers thiago, Perfect
22:53 bakers Now that my staging area is all full of merge conflicts... how do I say "throw away everything here, and reset it to whatever the "next" branch is"
22:53 bakers git reset hard or something
22:53 sdothum joined #git
22:54 PHPanos joined #git
22:55 thiago git checkout next .
22:55 thiago don't forget the .
22:55 thiago note that -s ours produces ZERO conflicts
22:56 bakers What's the difference between "git checkout next ." and without the .
22:56 bakers "-s ours" worked perfectly btw
22:56 bakers thank you
22:57 thiago with the dot: checks out the content of next but keep the current branch
22:58 thiago without the dot: tries to switch to the neext branch and keep any modified files
22:58 bakers ah ok
22:59 guest joined #git
23:00 metachr0n joined #git
23:00 BlaXpirit joined #git
23:01 bronson joined #git
23:01 fmeerkoetter joined #git
23:02 metachr0n joined #git
23:03 martinjlowm rizonz: if I understand correct, you want to reset the remote repo to a commit before the changes you pushed?
23:10 guest joined #git
23:10 martinjlowm sbeller: regarding the $path behavior, is it something that has been discussed earlier? I found nothing on the mailing list archive
23:11 NeverDie joined #git
23:11 sbeller martinjlowm: I do not remember a discussion on that during the last 3 years (since I am part of the Git community). However there was some submodule discussion very early on, so either the discussion you're trying to find  is old (and maybe outdated?)
23:11 sbeller or it wasn't there before
23:12 fbwnd joined #git
23:13 sbeller martinjlowm: I guess the latter as submodules are sort of a niche feature
23:13 martinjlowm $displaypath has the expected value, maybe changing the manpage and mentioning that is a better approach.
23:14 martinjlowm in situations where some users depend on the $path behavior as it is at the moment
23:14 GavinMagnus joined #git
23:15 sbeller not quite. The displaypath takes the pwd into account, i.e. cd superproject/side/dir && git submodule foreach echo $displaypath would give you paths that start with ../../submodule/path
23:15 sbeller (Well I do not know what your expected value is, but my expectation was to to have a variable available that is from the root of the calling superproject)
23:16 martinjlowm I see
23:16 martinjlowm makes sense
23:17 cdown joined #git
23:17 sbeller Maybe displaypath is enough and we only want to add a comment into the documentation that when --recursive is given, $displaypath is easier to use than $path ?
23:17 sbeller good* enough
23:17 jagob joined #git
23:20 martinjlowm In my case $displaypath will do. The other path that you propose can be found by the user I suppose. I think it's fine as it is, $displaypath works, but the path may be ugly
23:20 sbeller yeah that makes sense
23:21 martinjlowm in fact, $displaypath is better in my case and is independent on what is my current directory
23:24 rizonz martinjlowm: nah I did it different but I need to read up about stashes and such
23:25 rizonz martinjlowm: I come from SVN so I normally commited quite often for some modules I needed to test remote, this works different
23:25 martinjlowm haha.. actually no. It wasn't enough in my situation. Bash to the rescue (I hope)!
23:29 aarobc joined #git
23:31 Rodya_ joined #git
23:31 martinjlowm rizonz: the stash is neat! Okay :) I am not at all experienced in SVN (although I was introduced to it briefly at my university a couple of years ago), as I've always stuck to git. Bazaar was the first system I was introduced to. I am completely blank about that one though
23:32 ertes joined #git
23:34 rizonz martinjlowm: heh I used SVN because it had an nice admin and git had some issues for my solution about 3 years ago
23:34 martinjlowm rizonz: what do you use to interface with git? the command line?
23:34 rizonz martinjlowm: no gitlab
23:35 metachr0n joined #git
23:35 martinjlowm I mean locally :) (unless GitLab has a GUI client?)
23:35 rizonz locally, NetBeans
23:36 metachr0n joined #git
23:38 martinjlowm okay, I'm not familiar with NetBeans, but I hope it has support for stash management.
23:39 fbwnd joined #git
23:40 rizonz martinjlowm: there is a menu item stash changes :)
23:40 martinjlowm okay ;) it should be fairly simple then.
23:41 coasterz joined #git
23:42 diogenese joined #git
23:43 engineer-pearl joined #git
23:43 systemovich joined #git
23:44 rizonz martinjlowm: yeah need to real a little but more
23:44 sword` joined #git
23:44 rizonz in my opinion pushing is too easy, there is no question asked like when you commit
23:47 YuGiOhJCJ joined #git
23:49 metachr0n joined #git
23:50 dshap joined #git
23:51 dshap Hey all, is anyone here familiar with the BFG repo cleaning tool? I'm using it to delete a certain file (with sensitive data in it) from the entire history of my repo, but I understand that Github uses read-only refs for pull requests that can't be updated with the cleaned state of my local clone after I run BFG, is that correct? Just curious, if I clean with BFG and then push to Github, but the pull requests remain, and someone else clones the
23:51 dshap from Github, how would they go about finding a copy of the sensitive file using the pull request refs? Can you check out a Github pull request ref somehow and browse a snapshot of the files?
23:52 rizonz martinjlowm: when you pushed something to your master can you indom that simply ?
23:52 kadoban dshap: Yes, you can fetch and then checkout pull requests from github. If that's what you're asking.
23:52 martinjlowm rizonz: you mean undo?
23:53 rizonz martinjlowm: yap
23:53 dshap kadoban: How would I do that? I've been reading the git internals page on references (https://git-scm.com/book/en/v2/Git-Internals-Git-References) and I'm actually a little confused about it...shouldn't those "pull/X" refs be somewhere in my .git directory?
23:53 martinjlowm only for the remote and not the local?
23:54 rizonz martinjlowm: remote
23:54 dshap kadoban: I ran "git fetch origin"
23:54 rizonz martinjlowm: I mean I can push again
23:54 martinjlowm rizonz: you can checkout the previous commit, before the one you pushed and force a new push to reset the remote head to that previous commit
23:55 kadoban dshap: They're not in the usual fetchspec, you have to do something special to get them actually fetched. https://help.github.com/articles/checking-out-pull-requests-locally/ has some of it.
23:55 rizonz martinjlowm: ok, that will undo it, but it will be undone by the new push so you can see it ?
23:55 kadoban dshap: https://gist.github.com/piscisaureus/3342247 is better long-term, when you want to be able to fetch pull requests for a repo over and over.
23:55 * rizonz doesn't want to show his stupidity :D
23:56 rizonz I also need to see where I push to and if master will be bleedin edgre every time, I think so
23:56 shgysk8zer0 joined #git
23:57 martinjlowm rizonz: assuming you have just committed, pushed and not changed branch: git checkout HEAD^ -> git push --force origin/master -> git checkout master. Normally, you would avoid forcing like that. You can just commit changes that undo the older commit, but it will not look good in the history
23:57 dshap kadoban: Ah okay, I'll look at that, thanks! Just curious, I have a bunch of branches in my repo (e.g. master, develop, etc...). Shouldn't I have refs for those in my .git directory? The only thing in .git/refs/heads is /origin, and .git/refs/heads/origin is blank -- where are they stored?
23:58 Schrostfutz joined #git
23:58 rizonz martinjlowm: indeed, but better a timeline and have it right then no timeline and not able to see what is going on
23:58 rizonz I read something about it, that is there the timeline is for
23:58 dshap kadoban: actually it looks like all the branches are in my .git/config file....but I thought they'd be in the refs for some reason
23:59 martinjlowm rizonz: you can always squash the commits at a later point though
23:59 rizonz martinjlowm: ah writedoen, need to check what that is
23:59 rizonz :D
23:59 kadoban dshap: They're usually in .git/refs/heads if they're local branches. I believe they can also be "packed", which I forget where those are offhand.
23:59 dshap kadoban: ah, they must be packed then. I don't have anything in .git/refs/heads other than origin/

| Channels | #git index | Today | | Search | Google Search | Plain-Text | summary