Perl 6 - the future is here, just unevenly distributed

IRC log for #git, 2017-09-13

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

All times shown according to UTC.

Time Nick Message
00:00 Noldorin joined #git
00:01 Gsham joined #git
00:01 chrisshattuck joined #git
00:06 durham joined #git
00:10 btree left #git
00:12 Fissure joined #git
00:13 YuGiOhJCJ joined #git
00:13 hash-z3r0 joined #git
00:15 ToBeCloud joined #git
00:16 durham joined #git
00:16 vedu joined #git
00:19 tnecniv joined #git
00:20 lkolstad joined #git
00:26 rcn joined #git
00:33 StephenS joined #git
00:36 netj joined #git
00:39 Gsham joined #git
00:40 tunekey joined #git
00:42 kculpis joined #git
00:43 nd- joined #git
00:48 dreiss joined #git
00:49 justanotheruser joined #git
00:51 peterbecich joined #git
00:55 ahr3n joined #git
00:56 lxnr joined #git
00:57 stutter joined #git
01:00 shdw joined #git
01:03 EricPoe joined #git
01:04 seebyr joined #git
01:05 multi_io joined #git
01:07 dviola joined #git
01:07 dviola joined #git
01:09 str joined #git
01:14 ribasushi joined #git
01:19 QualityAddict joined #git
01:19 finalbeta joined #git
01:21 tarkus joined #git
01:26 Cabanossi joined #git
01:27 Kavothe joined #git
01:28 ahr3n joined #git
01:28 Murr joined #git
01:30 Gsham joined #git
01:31 feodoran joined #git
01:32 f3r70rr35f joined #git
01:32 sunri5e joined #git
01:35 kyan joined #git
01:36 Gsham joined #git
01:37 svm_invictvs joined #git
01:38 texinwien joined #git
01:39 steeze joined #git
01:53 sarri joined #git
01:53 sarri joined #git
01:55 ilbot3 joined #git
01:55 Topic for #git is now Welcome to #git, the place for git help and ><(((('> | Public logs at https://gitirc.eu/log | First visit? Read: https://gitirc.eu | Current stable version: 2.14.1 | Getting "cannot send to channel"? /msg gitinfo .voice | Have you ever seen a git eat its own HEAD?
01:56 xaviergmail joined #git
01:57 peepsalot joined #git
01:57 lxnr joined #git
01:57 tarkus joined #git
02:00 thiago joined #git
02:01 ismarkup joined #git
02:07 noobineer joined #git
02:07 FierceDeityLink joined #git
02:11 Cabanossi joined #git
02:12 jameser joined #git
02:12 systmkor joined #git
02:13 neunon joined #git
02:14 alyptik joined #git
02:25 stutter joined #git
02:28 ujjain- joined #git
02:28 DIDAVISION joined #git
02:30 Gsham joined #git
02:36 rigid joined #git
02:36 rigid joined #git
02:39 mkali joined #git
02:41 chachasmooth_ joined #git
02:42 lagothri1 joined #git
02:43 Giant81 joined #git
02:45 VladGh joined #git
02:45 Vortex34 joined #git
02:46 dmj` Hi, I merged a PR on github that had 3 commits I wish I would have squashed, what’s the best way to squash them and still give the author of the commits attribution
02:46 dmj` should I just rebase
02:47 roelmonnens joined #git
02:50 peterbecich joined #git
02:50 parsnip dmj`: there is author email and committer email. maybe somehow leave the author email intact.
02:51 Gsham joined #git
02:51 AndreasLutro if you already merged, you can't do that without rewriting history of the branch you merged into, which is not recommended
02:51 dmj` Ugh
02:51 dmj` @AndreasLutro so should I undo the PR? There is a revert option
02:51 dmj` which opens a new PR to undo
02:52 AndreasLutro you can do that but the commits you merged will still be in history
02:52 AndreasLutro so up to you if that's worth it
02:52 dmj` yuck
02:53 dmj` @AndreasLutro what about, git rebase -i HEAD~3
02:53 AndreasLutro rebase rewrites history
02:53 dmj` I’m ok with that as long as the original committer retains some attribution
02:55 dmj` AndreasLutro: do you think that’s the case?
02:56 AndreasLutro what?
02:56 dmj` that I can rebase, but retain the original commiters attribution in the history
02:57 AndreasLutro maybe, you can probably figure it out
02:57 AndreasLutro !rewrite
02:57 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
02:58 peterbecich joined #git
02:58 kadoban dmj`: If you rebase -i and do squashes or fixups, I bet it does that automatically, keeps at least the author intact. I'd be surprised if it didn't.
02:58 lxnr joined #git
02:58 dmj` @AndreasLutro I did it :)
02:58 dmj` @kadoban yea, that seems to be the case
02:59 dmj` there were 3 commits, and I squashed 2 of them into the third
02:59 dmj` it says I commited with the original author on github
02:59 kadoban Sounds good
03:00 parsnip now give them the good news ;)
03:00 dmj` heh :P “hey I rebased all your extraneous commits away, thanks!"
03:01 dmj` Oh, you know what
03:01 parsnip what?
03:01 dmj` It actually changes their +/- line count on GitHub I believe
03:01 orbyt_ joined #git
03:02 kadoban It probably should, if multiple changes to the same line get combined.
03:03 chrisshattuck joined #git
03:04 pks joined #git
03:10 cqi joined #git
03:11 justanotheruser joined #git
03:13 dre_ joined #git
03:18 ahr3n joined #git
03:19 igemnace joined #git
03:19 kfoonamalik joined #git
03:23 dwmw2_gone joined #git
03:26 Cabanossi joined #git
03:27 stutter joined #git
03:28 Murr joined #git
03:29 Mattiaslndstrm joined #git
03:33 BUSY joined #git
03:33 xaxxon joined #git
03:40 Cabanoss- joined #git
03:41 chrisshattuck joined #git
03:42 lkolstad joined #git
03:42 watabou joined #git
03:46 ahr3n joined #git
03:52 diqidoq joined #git
03:55 sarbs joined #git
03:59 beerdrop joined #git
03:59 Pseudocrat joined #git
03:59 Pseudocrat joined #git
03:59 lxnr joined #git
04:03 Paker joined #git
04:03 snowkidind joined #git
04:13 atesc joined #git
04:16 robotroll joined #git
04:17 igemnace joined #git
04:18 steeze joined #git
04:20 netj joined #git
04:25 Cabanossi joined #git
04:29 toast99_ joined #git
04:29 BrAsS_mOnKeY joined #git
04:33 toast99_ I have an ssh-agent running and added my ssh private key to the agent, and when I go to "git push origin master" it prompts me for username and password
04:33 toast99_ public key is on github account etc
04:34 toast99_ any help would be appreciated
04:34 _ikke_ toast99_: git remove -v
04:34 _ikke_ Are you using https or ssh?
04:34 _ikke_ (I suspect the former)
04:34 kadoban 'git remote -v' that is (t, not v)
04:35 _ikke_ ah, d'oh
04:35 imack joined #git
04:35 roelmonnens joined #git
04:35 kadoban _ikke_: I don't think it's just you, at some point I just gave up and added an alias for that one in my config.
04:35 _ikke_ haha
04:36 toast99_ trying to skim through the `man` but what does that do
04:36 Kaisyu joined #git
04:36 toast99_ sorry to be a bother
04:36 kadoban It just tells you what the URL is for the remote
04:37 kadoban and some other stuff, but that's what we're interested in. We're trying to determine if it's an HTTPS URL or an SSHy one.
04:37 toast99_ oh its https:
04:37 _ikke_ Then you need to change that
04:37 _ikke_ ssh keys obviously only work over ssh
04:38 kadoban So yeah, that's the issue. You can change it with uhm, should be 'git remote set-url origin git@github.com:something/whatever'  NOTE: make sure you change the last part to match your repo
04:38 kadoban (and possibly the remote name too)
04:39 toast99_ ah
04:39 toast99_ wicked awesome
04:40 toast99_ thanks a bunch =)
04:40 toast99_ lol git remove
04:41 toast99_ one thing I was curious about was this "https://help.github.com/articles/blocking-command-line-pushes-that-expose-your-personal-email-address/"
04:42 _ikke_ What about it?
04:42 toast99_ if your email is set to private. how could your email possibly be exposed through the command line
04:42 toast99_ like why the need for two checkboxes
04:43 kadoban toast99_: In git, you have to set your email address in your config, which goes in each commit. github couldn't/wouldn't be changing the actual commits, so all it could do is prevent you from pushing commits like that.
04:43 renarc joined #git
04:43 kadoban So that's what that setting sounds like it's for.
04:43 _ikke_ right
04:44 toast99_ I looked all over my github repo and couldnt find my email in my commits or anywhere though
04:44 kadoban Though ... don't really understand why. Email addresses are public information at this point.
04:44 toast99_ when I had that second box unchecked
04:44 kadoban toast99_: Which repo? They might not be displayed in github's interface very much (or maybe at all?)
04:45 toast99_ ok thats what I figured
04:45 _ikke_ toast99_: And when people clone your repo, they can see it too
04:45 kadoban But depending on how you have your git config, your actual email address might just not appear. Git doesn't check what you tell it your email address is, it just trusts you.
04:45 toast99_ my dotfiles repo which is just basically to help me learn and backup my files while im at it
04:46 rkta_ joined #git
04:46 toast99_ so like if I was to dig into the local repository i might be able to find that email somehow
04:46 toast99_ my first commit was with my plain email, then changed my git config to use the no reply email by the second
04:46 _ikke_ git log --format=fuller
04:47 toast99_ yes there it is
04:47 toast99_ I mean if one day I want to share this repository and people download it they can see my exposed email probably right?
04:47 _ikke_ yes
04:47 toast99_ crud
04:48 cqi joined #git
04:48 toast99_ only way to fix is delete and recreate?
04:49 _ikke_ toast99_: you can use man git filter-branch to change all commits
04:49 gitinfo toast99_: the git-filter-branch manpage is available at https://gitirc.eu/git-filter-branch.html
04:49 _ikke_ but that means the history will change (all commit hashes)
04:50 toast99_ i mean its just a few files...
04:51 toast99_ if I delete on github it wont affect my local repository right
04:51 toast99_ then I just rm -rf .git and then git add ., push etc?
04:52 _ikke_ toast99_: You can fix your local repository, and then force push
04:52 _ikke_ No need to delete it on github
04:53 howdoi joined #git
04:55 toast99_ "git filter-branch --commit-filter..."
04:57 _ikke_ toast99_: I think you just want --env-filter
04:57 _ikke_ the man page has an example
04:57 _ng joined #git
04:57 toast99_ it seems like I need to write a script first?
04:58 _ikke_ You can use it inline
04:58 _ikke_ And if you are the only one, you can just set the value
04:59 toast99_ this is where I lose you
04:59 toast99_ inline meaning I can enter the first command then it prompts me for the next inputs?
05:00 _ikke_ toast99_: git filter-branch --env-filter='GIT_AUTHOR_EMAIL=".."; GIT_COMMITTER_EMAIL=".."
05:00 _ikke_ toast99_: git filter-branch --env-filter='GIT_AUTHOR_EMAIL=".."; GIT_COMMITTER_EMAIL=".."' -- --all
05:00 Mattiaslndstrm joined #git
05:00 toast99_ okay thanks
05:03 cagomez joined #git
05:06 atesc joined #git
05:06 govg joined #git
05:09 toast99_ yeah still not getting it
05:10 toast99_ $GIT_AUTHOR_EMAIL isnt it
05:12 cagedwisdom joined #git
05:14 _ikke_ What do you mean?
05:17 freimatz joined #git
05:17 Bock joined #git
05:18 kfoonamalik joined #git
05:19 serialoverflow joined #git
05:20 dec0n joined #git
05:23 ShapeShifter499 joined #git
05:25 toast99_ g2g if youre on tmo i might come bug ya
05:25 toast99_ =P nut thanks for the help
05:25 toast99_ but*
05:28 lordjancso joined #git
05:28 Murr joined #git
05:29 ahr3n joined #git
05:29 jakogut_ joined #git
05:30 danieli joined #git
05:30 beerdrop joined #git
05:32 danieli joined #git
05:33 serialoverflow joined #git
05:34 glebihan joined #git
05:34 ThomasLocke joined #git
05:34 ThomasLocke joined #git
05:34 ZyX-I joined #git
05:36 oojacoboo how can I find out if my current branch is out of sync with the remote origin without explicitly specifying the branch name?
05:36 oojacoboo or do I need to determine the branch name first?
05:40 a3Dman joined #git
05:40 Cabanossi joined #git
05:41 moritz oojacoboo: git fetch; git log HEAD..@{u} # or something like that
05:41 igemnace git status -b has that info
05:41 igemnace or maybe just through git status -b --porcelain
05:41 lss8 joined #git
05:41 igemnace after a git fetch of course
05:41 moritz or compare git rev-parse @{u} with git rev-parse HEAD
05:41 oojacoboo hmm
05:42 oojacoboo git status -b --porcelain isn't going to work without some string manipulation
05:42 oojacoboo or grep work
05:42 oojacoboo but it has the data
05:42 igemnace this isn't for interactive use?
05:42 oojacoboo no, it's for a deployment script
05:42 oojacoboo validation
05:43 igemnace what does it need to find out
05:43 igemnace just if it's out of sync?
05:44 igemnace maybe just grab the first line and match it for ahead or behind inside []
05:44 oojacoboo git diff --stat works, but it's attempting to be interactive
05:44 oojacoboo igemnace yea, just want to make sure it's in sync
05:45 akushner___ joined #git
05:46 oojacoboo actually, comparing git rev-parse @{u} and git rev-parse HEAD will work
05:46 oojacoboo moritz git fetch isn't required for that, right?
05:49 beerdrop joined #git
05:50 moritz oojacoboo: you need a git fetch to get current state from the remote
05:50 moritz oojacoboo: unless you know it's already up-to-date
05:50 oojacoboo oh - that doesn't make a call to the origin server?
05:50 moritz that's exactly what "git fetch" does
05:51 oojacoboo well, that actually updates your local tree
05:51 oojacoboo I thought maybe origin/HEAD would check the remote server without a fetch
05:51 oojacoboo or something like that
05:51 oojacoboo but I guess you'd have to include the branch name in that
05:52 moritz slow down a bit
05:52 moritz git fetch doesn't change the working tree
05:52 moritz it's update the state of remote branches, so origin/master, origin/HEAD etc
05:52 oojacoboo it doesn't update the local git repo with remote references?
05:52 parsnip doesn't change "local refs", either
05:52 moritz oojacoboo: tree != repo
05:52 oojacoboo yea, I actually don't know the term for this
05:52 oojacoboo I thought I did and was searching
05:53 parsnip commit graph, let's say
05:53 oojacoboo the local database copy that references the remote server
05:53 oojacoboo sure
05:53 moritz working tree = files on your local disk outside the .git repo
05:53 parsnip git fetch doesn't touch working directory and index
05:53 oojacoboo gotcha
05:53 oojacoboo so it's quite harmless to throw it in there
05:53 parsnip "tree" has several uses in git, unfortunately
05:54 oojacoboo ah :/
05:54 oojacoboo I figured there was a single graph
05:54 oojacoboo just with pointers referencing different locations
05:54 parsnip oh right, further, git fetch doesn't mess with "local" refs
05:54 oojacoboo maybe with additional origin/local metadata
05:54 ArcaTech joined #git
05:55 moritz oojacoboo: there is mostly one graph of commits, with branches and tags pointing to those commits
05:55 parsnip and some of the several uses of the word "tree" might be informal
05:55 moritz oojacoboo: but you can have multiple root commits, and thus multiple graphs
05:55 moritz (it happens rarely, but it does happen)
05:56 moritz anyway, back to the topic at hand
05:56 oojacoboo cool :)
05:56 oojacoboo I swear my git knowledge is the worst of all my dev skills
05:56 moritz git fetch says "hey, sever, tell me what's up", and stores the result in origin/master, origin/HEAD etc.
05:56 oojacoboo it's kind of disappointing
05:57 oojacoboo moritz I see
05:57 moritz and then you can inspect that state with git log, git diff, git rev-parse, git status etc.
05:57 oojacoboo I'm going to consider that harmless for our use case
05:57 jknetl joined #git
05:57 moritz without those commands talking to the server again
05:57 oojacoboo makes sense
05:57 oojacoboo if I explicitly referred to `remote`, would it make a request?
05:57 moritz so, if you want to know "is master on the same state as origin master?"
05:58 igemnace depends on how you refer to it
05:58 oojacoboo moritz not master - current checked out branch
05:58 igemnace there are remote tracking branches in your local repository
05:58 moritz oojacoboo: one step at a time :-)
05:58 igemnace this helped me a bit https://git-scm.com/book/en/v2
05:58 Raging_Hog joined #git
05:58 moritz you could compare git rev-parse master with git rev-parse origin/master
05:58 rscata joined #git
05:58 oojacoboo igemnace holy hell weekend project read
05:58 igemnace make sure to go from start to finish, you might miss something crucial (such as the nature of objects in git)
05:59 moritz now you can replace "master" with "HEAD" if you want to refer to the current branch
05:59 moritz and @{u} to the corresponding upstream branch
05:59 oojacoboo hmm
05:59 moritz so you can run something like
05:59 moritz [ $(git rev-parse HEAD) == $(git rev-parse @{u}) ]; echo $
06:00 moritz which prints 0 if they are on the same state, and 1 if they aren't
06:00 parsnip i had some extra time on my hands recently, so i went through https://gitexercises.fracz.com/exercise/master, and part way through read https://jwiegley.github.io/git-from-the-bottom-up/ for a second time, and other resources such as #git, _really_ improved my understanding, still have a _long_ way to go
06:00 oojacoboo what's the upstream branch?
06:00 moritz oojacoboo: if you are on master, origin/master is your upstream branch
06:00 oojacoboo parsnip thanks for the links
06:00 parsnip :)
06:01 oojacoboo moritz gotcha
06:01 oojacoboo and will that make a request to the remote?
06:01 moritz it's the branch that "git pull" will merge for you, "git push" pushes to etc.
06:01 oojacoboo or do you still need to fetch?
06:01 tg joined #git
06:01 moritz oojacoboo: the git rev-parse doesn't do a request, no
06:01 oojacoboo so, even though you're saying the upstream, it's only what's stored locally?
06:02 moritz oojacoboo: there are only very few commands that talk to the upstream: fetch, pull, push, "remote update" and a few low-level commands
06:02 moritz right
06:02 oojacoboo gotcha
06:02 oojacoboo thanks!
06:04 rigid joined #git
06:04 rigid joined #git
06:04 parsnip i watched a talk that said, "never git pull"
06:05 Noodlewitt joined #git
06:06 oojacoboo parsnip I've read that multiple times
06:06 oojacoboo and I abuse it all the time
06:07 shdw- joined #git
06:07 shawn joined #git
06:07 tarkus joined #git
06:08 d-fence_ joined #git
06:08 JeroenT joined #git
06:10 JeroenT joined #git
06:10 MattMaker joined #git
06:11 MattMaker joined #git
06:12 polyrob joined #git
06:12 MattMaker joined #git
06:13 MattMaker joined #git
06:14 MattMaker joined #git
06:20 dreiss joined #git
06:22 abba joined #git
06:22 diogenese joined #git
06:23 roelmonnens joined #git
06:23 ThomasLocke joined #git
06:24 kadoban parsnip: I try to follow that unless I know exactly what's going to happen if I pull (like, my own repos I'm doing alone, or branches that are pure upstream I just follow)
06:25 parsnip i think the idea with rules is, number one they're meant to be broken, and number two is, the experienced/talented need not adhere.
06:28 _ADN_ joined #git
06:28 parsnip someone said in a video talk, they didn't mind ruby's author didn't adhere to the speaker's best practices, because ruby's author knows what they're doing.
06:29 Noodlewitt joined #git
06:29 phluks joined #git
06:31 ArcaTech joined #git
06:32 theoceaniscool joined #git
06:34 overlord_tm joined #git
06:35 nazarewk joined #git
06:41 tkore joined #git
06:43 energizer joined #git
06:43 zulutango joined #git
06:44 xaxxon joined #git
06:47 roelmonn_ joined #git
06:47 _ADN_ joined #git
06:48 jast I use 'git pull' a lot (often with --rebase), because I prefer having changes combined as quickly as possible. if the merge gets problematic or breaks things, I just undo it. it's all good as long as you don't push until you've got things sorted out...
06:48 jast and typically I know what to expect when pulling
06:51 lxnr joined #git
06:51 sci-fic joined #git
06:51 Chuguniy joined #git
06:52 netj joined #git
06:53 sci-fic Hi, I have a error  "error: cannot lock ref 'refs/remotes/origin/feature/sprint/branchname': is at 8135808350b95fe3ec7ed648e7fbf but expected 3a54913bfdff7cc2851e29e37b2bc1" somethink like that what can ı do? You guys any idea? thanks
06:54 HcsOmot left #git
06:55 parsnip sci-fic: what command provokes that error?
06:56 selckin sci-fic: any chance you have 2 branches with the same name & different case & on non case sensitive filesystem (winows)
06:56 parsnip or two refs?
06:56 sci-fic push, pull, fetch all of them. Ip has change and this error appear
06:56 selckin sci-fic: can try to delete it git update-ref -d refs/remotes/origin/feature/blabla and fetch it again
06:57 parsnip sci-fic: sounds like this one: https://stackoverflow.com/questions/6656619/git-and-nasty-error-cannot-lock-existing-info-refs-fatal
06:57 igemnace joined #git
06:57 parsnip maybe check ./.git/config for urls
06:59 ronskisika joined #git
07:00 parsnip is the issue with locking ref, because it wants to check the hash, but new ip is effecting hash?
07:01 parsnip so it created the remote ref before the ip change?
07:01 jast the stackoverflow link seems unrelated
07:01 selckin find it hard to believe ip would be included in that
07:01 jast IP addresses are irrelevant here
07:01 jceb joined #git
07:01 jast is your local git repo on a networked filesystem? (NFS, CIFS/SMB, ...)
07:02 parsnip i regetted pasting a stackoverflow link after i did it
07:03 jast this seems like some kind of race condition which either means someone else is using the same repo or the filesystem is being wonky
07:04 parsnip maybe too many apps open?
07:05 parsnip er, or one too many, like some IDE doing magic on the repo?
07:06 JeroenT joined #git
07:06 parsnip maybe investigate the two hashes above?
07:06 selckin time to wait if hes still there
07:12 MACscr joined #git
07:13 schleppel joined #git
07:13 Tobbi joined #git
07:14 diqidoq joined #git
07:15 wandering_vagran joined #git
07:15 parsnip hmm, if remote has a new branch, will fetch create a new local branch?
07:16 TomyWork joined #git
07:17 selckin not a local branch, but will show up under remotes/ in git branch -a etc
07:17 * selckin doesn't know all the terminology
07:17 akushner___ joined #git
07:18 JeroenT joined #git
07:18 parsnip okay, so you'd have to `git-checkout` for a local to be made
07:19 cagomez joined #git
07:19 insolfrac joined #git
07:19 insol_fract joined #git
07:19 zefferno joined #git
07:21 NonParity joined #git
07:21 pix9 joined #git
07:23 insolfrac joined #git
07:23 insolfrac joined #git
07:23 jast yep
07:23 jast terminology: !rtb
07:23 gitinfo [!remote_tracking_branch] Remote-tracking branches (branches which start with e.g. 'origin/', listed by 'git branch -r') are read-only mirrors of the branches in another repository. They're updated by 'git fetch'. You can't edit them directly (trying to check them out results in a !detached HEAD), but you can create a local branch based on a remote-tracking branch using e.g. 'git checkout -b <branch> <remote>/<branch>'
07:24 jast personally I prefer calling them mirror branches, though
07:25 insolfrac joined #git
07:28 Murr joined #git
07:35 Solvaring joined #git
07:37 Murii joined #git
07:41 theoceaniscool joined #git
07:42 StephenS joined #git
07:44 insolfrac joined #git
07:44 insolfrac joined #git
07:50 cagomez joined #git
07:54 ertes-w joined #git
07:55 texinwien joined #git
07:56 Cabanossi joined #git
07:57 acidjnk joined #git
07:59 luc4 joined #git
08:04 xaa joined #git
08:07 kurkale6ka joined #git
08:08 oxymoron93 joined #git
08:09 garyserj joined #git
08:11 Persondude joined #git
08:13 insolfrac joined #git
08:13 insolfrac joined #git
08:17 roelmonnens joined #git
08:18 guideline joined #git
08:19 Acerific joined #git
08:19 WrinkledCheese_ joined #git
08:19 Acerific joined #git
08:20 netj joined #git
08:21 WrinkledCheese_ Hey.   I've been pushing to github for the past 11 branches and all of a sudden I'm getting an error: "fatal: 'name_of_branch' does not appear to be a git repository
08:21 lss8 joined #git
08:21 cagomez joined #git
08:21 alkino WrinkledCheese_: what is your command?
08:21 texinwien I ran filter-branch with --index-filter to get rid of a file that shouldn't have been committed and pushed to a pull request.
08:21 WrinkledCheese_ My command I'm using is git push --set-upstream origin name_of_branch
08:21 texinwien In inspecting the results, it looks like way more commits were affected than I expected.
08:22 texinwien I would have expected 4 or 5 commits to have new hashes, but it seems that the number is much higher.
08:22 alkino WrinkledCheese_: remove --set-upstream and do
08:22 alkino git push origin name_of_branch:name_of_branch
08:23 texinwien I haven't counted, but it's probably more than 100.
08:23 alkino texinwien: and ?
08:23 alkino you want to come back?
08:23 jlebrech joined #git
08:23 WrinkledCheese_ Sorry for wasting your time texinwien.  I was omitting the origin part.
08:24 WrinkledCheese_ I mean alkino
08:24 Floflobel joined #git
08:24 texinwien I have a backup. I don't want to push anything that will affect 100+ commits, so I'm trying to figure out what I need to do differently so that only the 4 or 5 relevant commits are touched.
08:24 alkino WrinkledCheese_: nothing
08:24 alkino texinwien: give us commands
08:24 jlebrech I have to comment commits with a JIRA issue number, by cba typing that in for every tweak. should I create a branch with the issue number then at the end rebase or something?
08:24 texinwien You want the verbatim command I ran?
08:25 texinwien This is what I ran: git filter-branch --index-filter 'git rm --cached --ignore-unmatch .idea/watcherTasks.xml' HEAD
08:25 alkino texinwien: yeah
08:25 texinwien taken from the 'Examples' section, here: https://git-scm.com/docs/git-filter-branch
08:26 Cabanossi joined #git
08:26 KeyJoo joined #git
08:26 texinwien Interestingly, it doesn't appear to have affected all 1700 historical commits, just 100 or so.
08:26 alkino If you want to "completely forget" a file, it does not matter when it entered history, so we also add
08:26 alkino --ignore-unmatch:
08:26 alkino texinwien: -^
08:27 alkino otherwise I don't know
08:27 abrcdbr joined #git
08:28 pbandark joined #git
08:28 texinwien right, I read that, but it's not 100% clear what the result would be. I was concerned that it might mean that irrelevant commits might also be touched.
08:29 texinwien I also read the docs on that option to rm, and it's still not 100% clear what the outcome should be.
08:29 oxymoron93 you could back up your project, and try it
08:29 texinwien It's also not clear what would happen if I ran the command without that option.
08:29 texinwien Is there any way to limit 'filter-branch' to a range of commits?
08:30 texinwien oxymoron93 I backed up my project and tried it, and it seems that the results were not what I wanted.
08:31 texinwien So I'm now hoping to get some insight rather than backing up the project again and trying with various combinations of options until I (hopefully) land on the magic combination.
08:31 bcc joined #git
08:31 bcc joined #git
08:32 texinwien With 1700 commits in the repo, 'filter-branch' took around half an hour to execute on my PC last night, so there's also the question of how much time would be involved to blindly try different combinations of options.
08:33 mikecmpbll joined #git
08:33 oxymoron93 I never used filter-branch, but when it comes to "ranging" A..B is used
08:34 jast sure, you can use ranges with filter-branch (specify a ref/branch *and* a commit range)
08:34 oxymoron93 texinwien ^
08:34 jast of course it will have to rewrite descendants of that range, too
08:35 texinwien jast, oxymoron93 I'm fine with it rewriting all descendants. They're all my commits to feature branch on my clone of another repo.
08:35 jast right, just wanted to make sure you're aware
08:36 texinwien I'll do some research on specifying a range for the filter-index operation. Thanks for the tip!
08:36 JeroenT joined #git
08:37 jast search for 'range' in man git-filter-branch. it's just a short remark, though.
08:37 gitinfo the git-filter-branch manpage is available at https://gitirc.eu/git-filter-branch.html
08:37 jast it's really just: git filter branch ... -- HEAD abc123..def456
08:38 jast in case you're not too familiar with ranges, this is *exclusive of* abc123
08:38 jnavila joined #git
08:38 manuelschneid3r joined #git
08:40 parsnip texinwien: why do you expect only 4 or 5? are they near the tips of unmerged branches?
08:41 gxt joined #git
08:42 texinwien parsnip: They're on the tip of a single unmerged branch
08:42 texinwien I cloned a github repo yesterday, created a new feature branch, developed a feature, then committed.
08:43 texinwien Unfortunately, I was careless and committed and pushed an IDE config file along with the source files I'd modified.
08:43 texinwien Then I made a few more commits to that branch.
08:43 parsnip did anyone merge it on origin?
08:43 texinwien 4 or 5, in total.
08:43 texinwien no
08:44 texinwien it has not been merged. I did open a pull request, but it has not been merged.
08:44 parsnip why is rebase not correct here?
08:44 texinwien I'm not sure, perhaps it is.
08:44 oxymoron93 ugh I hate when it comes to deleting files, git so protective
08:45 texinwien I'll read up on it if you think that's the way to go.
08:45 jfr joined #git
08:46 Nizumzen joined #git
08:46 jast vanishing recently added files that didn't change afterwards tends to be simpler with !interactive_rebase
08:46 gitinfo Interactive rebase sounds similar to rebase but has completely different abilities. It can do this to commits: change the order, squash some of them together, remove some, add random existing commits from other branches, split them, and more... it's very powerful. Documentation is in the section "Interactive Rebase" in 'man git-rebase'.
08:47 jast but if you've already composed the right 'filter-branch' command there's no need to switch over to interactive rebase, there's no huge technical advantage or anything
08:48 texinwien the filter-branch command I have composed is not the correct one, so I'll either need to compose the correct one or figure out how to do it via interactive rebase.
08:48 aard_ joined #git
08:49 parsnip i find filter-branch syntax very intimidating :(
08:49 osse who doesn'y
08:49 texinwien ok, my intuition suggests that rebase is probably the more correct solution here, in that it's the one that's likely to be less destructive if I get it wrong.
08:49 parsnip oh sure, i'll "filter... filter... with a `rm -r` command", sure
08:50 King_Hual joined #git
08:51 parsnip yeah, and there's an option to abort before it's complete, and even if you finish it, i'm not worried i can't go back to original sequence of commits by, say, looking in reflog for the un-ref'd tip.
08:52 parsnip git doesn't remove commits quickly, i think 14 days is the lower bound. you only can't see them, but reflog or fsck can help
08:53 parsnip whereas if you run filter-branch across the whole repo, ... maybe similar recoverability, but, okay, sure.
08:54 parsnip oh, and maybe filter-branch would figure out it's only 4 or 5 commits
08:54 texinwien parsnip filter-branch didn't seem to figure it out with the options I used.
08:54 parsnip texinwien: i wonder if it's saying 100 commits cuz someone else made a similar mistake in the past on same repo. maybe you can search the repo for all such commits.
08:55 texinwien I was wondering the same, and went searching, but didn't find the smoking gun.
08:55 parsnip yeah, i still vote that filter-branch has difficult syntax.
08:55 parsnip weird, does filter-branch report on where these 100 are?
08:55 Cabanossi joined #git
08:55 texinwien Interestingly, filter-branch reported that only the 4 or 5 relevant commits had been rewritten.
08:56 texinwien However, when I look in the log view on my IDE, I see more than 100 commits that are doubled on two branches with different hashes.
08:57 texinwien == the the matching commits on each branch have different hashes.
08:57 parsnip check the dates on those? maybe the duplicates are old.
08:58 insolfrac joined #git
08:58 texinwien they're old, but there are no dupes in the backup of the repo.
08:58 parsnip backup?
08:58 texinwien Back to May or so. Not my work. I just cloned this repo and created the feature branch yesterday.
08:58 texinwien I made a backup copy of the repo before I ran filter-branch.
08:59 texinwien cp -a feature-branch feature-branch-bk
09:00 jast filter-branch backs up all changed refs itself, too
09:00 texinwien What I'm planning to do next is run an interactive rebase on a new copy of the backup. Sounds like a good plan?
09:00 texinwien Longer term, I'm planning to be less careless so this doesn't happen again.
09:00 texinwien :D
09:01 texinwien less careless / more careful
09:01 jast sure. the thing with interactive rebase is that you select the right base (the commit before the one where you added the file) and then edit just that one commit, the rest after that is rewritten automatically
09:01 Gurkenglas joined #git
09:01 parsnip i love it, the banner of git, "you can fix any problem, but it may take a while"
09:01 texinwien jast that sounds perfect
09:01 jast by choosing the base yourself, you can make sure that no older commit is touched, not even for a no-op "let's just rewrite the commit timestamp"
09:02 phluks joined #git
09:02 texinwien parsnip yeah, I mentioned last night - gotta love it when fixing careless git errors takes as much time as actually developing a new feature :D
09:02 jast it's a matter of practice ;)
09:02 jast if you make more careless errors you get faster at fixing them
09:02 texinwien OK, fortunately, I think this is a good target for rebase - the actual changes only occurred in one file.
09:02 oxymoron93 practice, not to make mistake
09:02 jast so clearly the way to go is to be *less* careful :D
09:03 belmoussaoui___ joined #git
09:03 oxymoron93 imo
09:03 parsnip maybe that's why i'm here
09:03 texinwien So when I rebase, I should edit that file to match what it was in the errant commit?
09:04 jast you wanted to get rid of the file altogether, didn't you?
09:04 texinwien jast yeah, I also mentioned last night - this happened once before and someone else fixed it for me. In retrospect, I wish they'd taken the time to show me how to fix it...
09:04 parsnip when you rebase -i, you'll mark the commit for "edit", then when you run the rebase, it will stop at that commit waiting for you to make chanegs.
09:04 jast ah, retrospect... my favourite kind of spect...
09:04 texinwien jast two files were committed. I'm asking what state the correct source file should be in.
09:04 texinwien the incorrect file should be removed completely, yes.
09:05 oxymoron93 inspect, least favourite?
09:05 Goyo left #git
09:05 jast introspect, I guess...
09:05 texinwien srcFile -> update to match the state in the errant commit; configFile -> remove from commit.
09:05 jast circumspect is another top candidate for favourite
09:05 jast right
09:06 jast basically, when rebase stops at the problem commit, 'git rm --cached' the config file
09:06 jast and the error in srcFile is in the same commit?
09:06 texinwien no error in srcFile.
09:06 texinwien I'd like to keep everything as-is, just remove the config file.
09:07 jast okay, then what I said is the only thing you need to do
09:07 jast here's the complete outline
09:07 parsnip you need only tell rebase -i what to change
09:07 texinwien errant commit = the commit where this bonehead errantly added an IDE config file.
09:07 vjacob joined #git
09:07 jast 1) rebase -i <errant commit>^ (^ = the commit before that is our base)
09:07 parsnip er, what to change _about_ the change ;)
09:07 texinwien right
09:07 jast 2) in the sequence editor, change the line for the commit from 'pick' to 'edit', save and quit
09:07 jast 3) git commit rm --cached configFile
09:08 jast 4) git commit --amend
09:08 jast 5) git rebase --continue
09:08 parsnip jast: (pick -> edit) on the bad commit, right?
09:08 jast basically this allows you to go back in time and amend the commit
09:08 jast yep
09:08 texinwien perfect. I will dd that to a gist for the next time :D
09:08 jast there's a chapter on interactive rebase in the git book, too
09:08 jast !progit
09:08 gitinfo [!book] There are several good books available about git; 'Pro Git' is probably the best: http://git-scm.com/book but also look at !bottomup !cs !gcs !designers !gitt !vcbe and !parable
09:08 parsnip jast: s/commit/add/?
09:09 texinwien Off to interactive rebase. Will report back. Thanks for being 1,000 times more friendly than #linux :D
09:09 jast nah, 'rm --cached' unstages it and 'commit --amend' bakes that into the commit
09:09 parsnip `git commit rm`?
09:09 jast no such thing :)
09:09 texinwien jast yeah, I need to buy a book. I will order that one.
09:09 jast pro git is available online for free, but I'm sure the author will appreciate you getting a paper copy :)
09:09 parsnip 02:07 <jast> 3) git commit rm --cached configFile
09:09 jast oh, whoops
09:10 jast 3) git rm --cached configFile
09:10 bruce_lee joined #git
09:10 texinwien parsnip jast thanks for the correction :)
09:10 jast git mental-focus --increase
09:10 parsnip jast: why do we need the git commit --amend? i thought if we do continue, it knows to commit?
09:10 parsnip won't that add an extra commit?
09:11 parsnip not that you can't go back and squash it later
09:11 kurkale61 joined #git
09:11 Tee_Pee joined #git
09:12 texinwien In #linux, it seems the default tone of responses is pure derision. Maybe I'm generalizing too much, but that's been my experience. Much more pleasant and helpful crowd here. Itchy neckbeards?
09:12 parsnip irc's channels vary widely. hangout on ##bootstrap for a while for a laugh
09:13 parsnip never a discussion or answers, just people popping in every hour or so asking a question with absolutely no response.
09:14 jlebrech how do you make a feature branch look like it just a single commit on master?
09:14 texinwien parsnip: See #elixir for similar
09:14 oxymoron93 in #doxygen you should wait minimum 12hrs, as banner says
09:14 texinwien although probably fewer questions.
09:14 igemnace jlebrech: as in, when you merge? check out squash commits
09:14 PikalaxALT joined #git
09:14 upperdeck joined #git
09:15 texinwien no responses, although they advertise the channel as a good place to ask questions on the language's official site.
09:15 jlebrech igemnace: yeah when I merge.
09:16 doc_ left #git
09:17 igemnace #emacs is almost always off-topic, but in the interesting way
09:17 igemnace ##javascript in the not-so-good way
09:18 neil_cn joined #git
09:18 parsnip jlebrech: you mean you have a branch `topic` with say 7 commits, and branch `master` has 3 commits from their common ancestor, and you want to squash the `topic` branch from 7 commits to 1 before merging?
09:18 YuGiOhJCJ joined #git
09:18 jlebrech parsnip: yes, squash a feature into one :)
09:18 jlebrech feature branch*
09:18 parsnip okay, i just wanted to make sure there wasn't a misunderstanding.
09:19 jast texinwien: there are people here who try and keep the channel friendly (towards reasonable people at least) so that may be making a difference
09:19 jlebrech because of bloody JIRA i have to put in a issue number before i merge :D
09:19 jast impatience and frustration, left unchecked, has a tendency to make channels a bit less welcoming
09:19 oxymoron93 JIRA is good
09:20 phluks joined #git
09:20 oxymoron93 never used it, but it gives you some flow, from what I read about it
09:20 achadwick joined #git
09:20 jast parsnip: normally that's the way it works in rebase, yes. but in interactive rebase, the 'edit' command stops *after* applying the commit in question
09:20 jast normally during a rebase, conflicts happen *while* applying a commit and so that's where it stops
09:20 ppang joined #git
09:21 jlebrech oxymoron93: it can be good till other team members want to enable all the features at once.
09:21 jast in that case, staging your conflict resolution and continuing simply resumes where it left off, by finishing up that commit being applied
09:21 texinwien so, I have a small issue: I ran 'git rm -r cached configFile' yesterday, thinking that would remove the file from the repo.
09:21 parsnip jast: aaahhh, that explains some recent frustration, thank you!
09:21 texinwien I haven't committed since. when I try to rebase, it complains about that uncommitted change.
09:21 jlebrech git merge --squash?
09:22 kurkale6ka joined #git
09:22 oxymoron93 so jlebrech, you are on your feature branch, use git rebase -i <common_ancestor_sha> and in editor choose what commits to squash, by changing "pick" to "squash"
09:22 jlebrech +1
09:22 oxymoron93 yeah maybe that is good too jlebrech , never used it
09:22 jast texinwien: well, if you're going to rebase that file away, anyway, you can just discard your 'rm --cached'
09:22 parsnip ooh, i'm reading squash can be done with a soft reset and a commit
09:22 jlebrech :P
09:22 oxymoron93 reading manuals is what I do jlebrech
09:22 texinwien I've tried reverting and also checking out the last commit on this branch to discard that operation.
09:22 jast 'git status' tells you at the top of the section how to do that. for convenience: git reset -- <file>
09:23 phluks joined #git
09:23 texinwien great, thanks. I will take a look at 'git status'
09:23 oxymoron93 parsnip is also right jlebrech
09:23 jast 'git status' is full of helpful hints people routinely skim over :)
09:23 cagomez joined #git
09:23 abrcdbr_ joined #git
09:23 texinwien thanks, added to my toolbox
09:24 jlebrech i'm scared of using git :P
09:25 mostlybadfly joined #git
09:25 oxymoron93 with git reset --soft <common_ancestor_sha> && git commit, much easier
09:25 Cabanossi joined #git
09:26 parsnip jlebrech: if you want some extra aids to help make things easier, you could put a ref name at the point of common ancestor, so it's easier to refer to as you go
09:26 jast if you want to squash all descendants of the ancestor, sure
09:26 jlebrech i don't like when non techies change my workflow, as I have to learn scary new git params
09:26 jlebrech oxymoron93: that seems like a nice way
09:27 j08nY joined #git
09:27 texinwien I'm surprised at how long I'm having to wait for something to happen after running the 'git rebase' command.
09:27 parsnip i hate the "off by one" worries in git
09:27 jlebrech can I: git reset --soft master@HEAD ?
09:27 selckin texinwien: stop using cygwin
09:27 texinwien selckin nailed it
09:28 texinwien what's your recommendation?
09:28 Murr joined #git
09:28 selckin linux
09:28 igemnace jlebrech: git reset --soft master should do, iinm
09:29 igemnace i don't think master@HEAD means anything
09:29 texinwien ok, I've been considering it for a while. In the meantime, is git rebase going to work in cygwin? Is it just a matter of having enough patience?
09:30 navidr joined #git
09:31 jast yes, there's no such thing as master@HEAD
09:31 jast !HEAD
09:31 gitinfo HEAD is a 'pointer' to the currently checked out branch (or commit, if HEAD is !detached). In bare repositories it tells clients which branch to checkout initially after cloning. Unlike commonly believed, HEAD is *not* something that exists separately for every branch. It also is *not* necessarily the newest commit in the repo (that's hard to define in a DVCS, anyway...)
09:31 SerusWork joined #git
09:32 oxymoron93 git merge-base feature master will give you the hash of common_ancestor between those two
09:32 oxymoron93 don't know how to include it inside another commands :D
09:32 oxymoron93 not learning bash at all
09:32 jast using shell magic:
09:32 parsnip $(...)
09:32 jast git show $(git merge-base a b)
09:33 wxg joined #git
09:33 Puffball joined #git
09:38 czart joined #git
09:40 jlebrech I don't need this develop branch now I can git reset :)
09:44 mikecmpbll joined #git
09:45 samlisl_ joined #git
09:49 finalbeta joined #git
09:51 shiva__ joined #git
09:55 phluks joined #git
09:59 rasto joined #git
09:59 tarkus joined #git
10:01 jlebrech gonna stay away from any code being done on master, make feature branches for each jira issue and then git reset before merge back into master, and deploying to production will be of tags.
10:01 lxnr joined #git
10:02 jlebrech kiss develop branch byebye
10:04 ski7777 joined #git
10:04 JeroenT joined #git
10:05 manuelschneid3r joined #git
10:07 kurkale61 joined #git
10:13 kurkale6ka joined #git
10:15 Darcidride joined #git
10:17 wxg joined #git
10:17 h12o joined #git
10:18 bambs joined #git
10:18 doev joined #git
10:20 salamanderrake joined #git
10:21 HannesP joined #git
10:21 bronson joined #git
10:22 watabou_ joined #git
10:25 sansna joined #git
10:25 Cabanossi joined #git
10:25 ShalokShalom_ joined #git
10:26 kurkale61 joined #git
10:27 destiny_ joined #git
10:27 HannesP if a remote repo was once checked out with a username and password, are they (or at least the username) stored anywhere in connection to that repo?
10:28 grawity possibly
10:28 grawity were the credentials specified in the URL, or entered separately during clone?
10:28 grawity assuming by "checked out" you're referring to `git clone`
10:28 HannesP oh sorry, yes, of course i mean cloned
10:29 HannesP honestly i don't know, i've taken over somebody else's project and they didn't leave much documentation... but it's on a server, and the method of updating the software is pulling the last revision from a certain branch in the repo
10:29 tobiasvl if it was cloned via HTTPS, the username is stored in the remote URL itself
10:29 tobiasvl but not the password of course
10:30 grawity do you have any credential helpers (git config -l | grep helper)?
10:31 HannesP hmm no, but it seems to show me the username that was used, and that answers one of the questions!
10:31 grawity unrelated question
10:31 grawity if I `git bisect bad` the same commit twice, will it confuse the tool?
10:33 kurkale61 joined #git
10:34 _ikke_ grawity: from what I understand from the script is that it just updates the same ref twice
10:35 jnavila joined #git
10:35 _ikke_ grawity: it will add 2 entries to the bisect log
10:35 jlebrech joined #git
10:35 grawity so just a visual issue, okay
10:37 grawity alright, 8 more kernels to go
10:37 JeroenT joined #git
10:38 h12o joined #git
10:38 texinwien joined #git
10:38 donnib joined #git
10:39 psyk3d joined #git
10:39 hseg joined #git
10:40 hseg Hi. Just wrote a commit-msg hook and would like to check my old commits (at least, those I haven't pushed yet) are OK. How do I do this? i.e. how do I run a script on each commit_msg?
10:41 cagomez joined #git
10:41 tvw joined #git
10:42 _ikke_ I guess by iterating each commit, write the message to a file, and then run the hook with the filename
10:44 stamina joined #git
10:46 hseg That feels ugly, but OK. One way I can think of to do this is if there's a way to get e.g. the list of shas between master and origin/master, and then a way to dump the commit message for each sha.
10:46 bsanford joined #git
10:46 _ikke_ git rev-list master..origin/master
10:47 _ikke_ hseg: Perhaps git filter-branch also runs the hooks, but I'm not sure of that
10:47 Solvaring joined #git
10:47 hseg ty
10:47 tarkus joined #git
10:47 _ikke_ git show --format "%B"
10:48 hseg OK, so it's a matter of iterating over $(git rev-list...) and running test <(git show...)
10:49 _ikke_ Right
10:51 hseg OK, that worked. Thanks
10:52 _ikke_ Note that, especially in the beginning, a lot of git commands were written in a similar manner
10:54 JeroenT_ joined #git
10:54 abrcdbr joined #git
10:54 hseg Yeah, and I suppose these were too simple and configuration-specific to bake in?
10:55 Cabanossi joined #git
10:55 _ikke_ Yeah
10:56 _ikke_ Nobody bothered to write it
10:56 _ikke_ It's easy to do it like you did
10:56 cagomez joined #git
10:57 hseg BTW, does git protect the .git directory? otherwise, I could make a submodule containing my hooks (since many of my hooks are generic)
10:58 grawity it refuses to add or commit anything named ".git"
10:58 grawity central repo for hooks is not uncommon though
10:58 hseg But .git/hooks/commit-msg is OK?
10:59 grawity no, because it involves a directory named .git
10:59 hseg Ah. So I'd need to ln -s the hooks from the submodule?
10:59 grawity yes
10:59 grawity recent git versions have a config option for central hookdirs
11:00 str joined #git
11:00 hseg Ah, nice. That is in addition to .git/hooks?
11:01 hseg i.e. does it fallback to .git/hooks if core.hooksPath is set?
11:01 mentazoom joined #git
11:02 [worksti] joined #git
11:03 JeroenT joined #git
11:03 theoceaniscool joined #git
11:03 [worksti] my workflow has a lot of DRY violations when it comes to versions, am i doing it wrong ? my workflow: a) change version in pom.xml to 1.2.3, b) commit change, commit message: "Raise version to 1.2.3" (repeat 1) c) tag commit, tag name: "v1.2.3" (repeat 2), tag comment (which i cant leave empty, why not ?) "v1.2.3" (repeat 3)
11:06 moritz you can leave out the tag message if you don't need an annotated tag
11:07 moritz git tag with out -a
11:07 moritz you could also restructure your workflow
11:07 jeffreylevesque joined #git
11:08 moritz pom.xml contains a placeholder. You create a git tag, and then have a tool that puts the version number from the last tag in your pom.xml, and builds an artifact from that
11:08 watabou joined #git
11:09 [worksti] ill consider that
11:09 cstratak joined #git
11:09 [worksti] thanks for the -a tip, that cuts down at least one of the repeats :D
11:11 Cavallari joined #git
11:12 PioneerAxon joined #git
11:15 noidea joined #git
11:15 jknetl joined #git
11:17 JeroenT joined #git
11:17 JeroenT joined #git
11:17 kurkale6ka joined #git
11:20 al-damiri joined #git
11:21 bronson joined #git
11:22 venmx joined #git
11:22 venmx_ joined #git
11:24 kurkale6ka joined #git
11:26 Cabanossi joined #git
11:26 rchavik joined #git
11:28 Murr joined #git
11:28 se1_ joined #git
11:30 frobnic joined #git
11:32 czart_ joined #git
11:32 southernirc joined #git
11:36 theoceaniscool joined #git
11:40 eblip joined #git
11:40 sotoz joined #git
11:41 ShekharReddy joined #git
11:42 tchan joined #git
11:42 LBV_User joined #git
11:44 averell joined #git
11:50 frobnic joined #git
11:51 wxg joined #git
11:51 kurkale61 joined #git
11:51 no_maam joined #git
11:53 tvw joined #git
11:58 cagomez joined #git
12:00 JeroenT joined #git
12:01 JeroenT joined #git
12:02 ankit01ojha joined #git
12:05 texinwien oh man, this rebase is turning into a saga.
12:06 fatihgenc joined #git
12:07 watabou_ joined #git
12:08 texinwien it's just a comedy of errors, one after the other
12:09 immae joined #git
12:10 oxymoron93 interactive rebase?
12:11 savoir-faire left #git
12:11 boombatower joined #git
12:13 cebor joined #git
12:16 texinwien yes, only I haven't made it that far yet.
12:16 synthroid joined #git
12:16 texinwien cygwin on Win 10
12:16 texinwien I had an older version of git that was erroring when I tried to rebase.
12:17 texinwien So I reinstalled via cygwin, and now git status shows that every file in the repository has been updated, when only 2 files have. The rest are shown due to lf / crlf mismatches.
12:17 kurkale61 joined #git
12:17 texinwien I can't start the rebase because git thinks I have unstaged changes. It's about to kill me.
12:18 cqi joined #git
12:18 lkolstad joined #git
12:19 texinwien that describes around 10% of the issues I've had with this today.
12:22 zefferno joined #git
12:22 Tobbi joined #git
12:26 peacememories joined #git
12:30 JeroenT joined #git
12:31 JeroenT_ joined #git
12:33 tcp_j joined #git
12:34 texinwien does anyone know what the default rebase editor is on Cygwin?
12:34 texinwien I'm trying to figure out which editor it's using, because whichever editor it is, it's not acting like one I'm used to (vim, usually)
12:35 plasticle joined #git
12:39 \drew joined #git
12:40 Cabanossi joined #git
12:42 str joined #git
12:42 CussBot joined #git
12:43 peterbecich joined #git
12:43 bb_mac joined #git
12:48 Paker joined #git
12:50 sekmo joined #git
12:51 jeffreylevesque joined #git
12:51 sotoz joined #git
12:52 no_maam joined #git
12:53 perlpilot texinwien: you mean when the $EDITOR environment variable isn't set?
12:53 perlpilot texinwien: probably pico or nano or something like that.
12:54 texinwien looks like it was vi or vim, only without a .*rc file that makes it work like I'm used to.
12:56 elsevero joined #git
12:56 donnib joined #git
12:56 thierryp joined #git
12:57 donnib joined #git
13:00 cdg joined #git
13:01 mkali joined #git
13:04 mentazoom joined #git
13:05 cdg_ joined #git
13:06 stutter joined #git
13:08 TikityTik joined #git
13:08 Gsham joined #git
13:09 fission6 joined #git
13:10 stutter_ joined #git
13:10 jsears joined #git
13:11 grawity hmm, is it just me, or is `git rev-list --bisect` broken...
13:11 crose joined #git
13:11 grawity i'd expect it to return "bad ^good ^good ^good ^good", but all it returns is the 'bad' ref
13:12 stutter joined #git
13:14 copyjosh joined #git
13:14 gugah joined #git
13:18 jeffreylevesque_ joined #git
13:18 _ikke_ grawity: WHy do you expect that?
13:19 wandering_vagran joined #git
13:21 paultreny joined #git
13:21 Gsham joined #git
13:21 gloomy_ joined #git
13:22 bronson joined #git
13:22 dan2k3k4 joined #git
13:22 GingerGeek joined #git
13:22 GingerGeek joined #git
13:23 Jak_o_Shadows joined #git
13:23 tnecniv joined #git
13:28 grawity apparently I'm confusing it with another subcommand then?
13:28 upperdeck joined #git
13:28 JeroenT joined #git
13:28 JeroenT joined #git
13:28 Murr joined #git
13:28 Levex joined #git
13:29 thiago_ joined #git
13:30 venmx joined #git
13:30 venmx_ joined #git
13:31 jimInDevNull joined #git
13:31 donnib hi guys
13:31 donnib I did git subtree split -P Folder/MyFolder -b onlyMyFolder to split a folder into a branch so i can move it to a separete repository
13:31 JeroenT joined #git
13:32 donnib i did above while beeing on master
13:32 donnib now when the operation is done i can see that split did not take last commit into the new branch although the changes in that commit are solely done to that path that i split
13:32 JeroenT_ joined #git
13:32 donnib what is the reason for that ?
13:32 donnib any ideas ?
13:33 Paker joined #git
13:34 SteffanW joined #git
13:40 wxg joined #git
13:40 Cabanossi joined #git
13:40 MineCoins joined #git
13:42 erhandsome joined #git
13:44 EricPoe joined #git
13:45 phinxy joined #git
13:46 phinxy left #git
13:46 rawste joined #git
13:47 kurkale6ka joined #git
13:49 netj joined #git
13:50 ahr3n joined #git
13:50 monomon[m] joined #git
13:50 Dougie187 joined #git
13:52 watabou_ joined #git
13:52 bitmod joined #git
13:53 Es0teric joined #git
13:55 kurkale61 joined #git
13:55 theoceaniscool joined #git
13:56 tnecniv joined #git
13:59 Ladus joined #git
14:00 indistylo joined #git
14:00 [worksti] because we have a legacy SVN system here that i use via subgit, i would like to configure my local clones of git repositories in a way that they have two upstreams: the subgit repo, which will have downstream changes from other devs, to which i only want to push the master branch and no tags, and my gitlab repo, to which i want to push everything, and from which no changes will come down. can i set that up
14:00 [worksti] on a git level ?
14:01 peterbecich joined #git
14:01 kpease joined #git
14:01 _ikke_ [worksti]: Yeah, I think that might work
14:02 Gsham joined #git
14:02 synthroid joined #git
14:03 imrg joined #git
14:04 Fallen0223 joined #git
14:06 texinwien on the road to getting my interactive rebase working, it seems that I borked git in several ways, and since I use it daily, I'm hoping to get some pointers.
14:06 cqi joined #git
14:06 jast here's a pointer: char *foo;
14:06 jast ... I'll see myself out
14:07 texinwien Prior to this debacle, I was able to push to git from my IDE and from Cygwin.
14:07 texinwien Thanks for the pointer ;)
14:07 _ikke_ Use it wisely :P
14:08 texinwien I had to update git in order to get the rebase working - I was on a 1.7 version.
14:08 jast 1.7 should be capable of rebasing, really
14:08 texinwien I first downloaded and ran the Windows installer.
14:08 _ikke_ That's rather old, but even then rebase should be working
14:08 texinwien I think it was an issue specific to cygwin.
14:08 [worksti] _ikke_: and thats not going to make me go to hell ?
14:09 jast right. the Git installer is independent from cygwin, so that cygwin version is, in theory, irrelevant now
14:09 jast [worksti]: for the most part Git doesn't care if you mirror things to another repo
14:10 jast I mean, Git doesn't support multiple remote configs per repo just for the pretty commands :)
14:10 texinwien that's what I would have thought, but after installing git on Windows, it stopped working inside Cygwin completely.
14:11 texinwien I have read that cygwin can get confused about which git to use if more than one is installed on the PC.
14:11 jast why are you using it in cygwin, though, if you're wanting to use the new version you just installed?
14:11 texinwien Next up, I installed git in Cygwin
14:12 texinwien Good question. I was using Cygwin because when I first tried to use git through my Windows IDE after installing the new version, I was getting Cygwin errors.
14:13 texinwien does that even sound possible?
14:13 ashpokemonZ joined #git
14:13 jast I don't know what your IDE does. if it uses cygwin in the background, then... yeah, I guess
14:13 texinwien Yeah, I may have configured it to use Cygwin.
14:13 jast if you can configure it, maybe change it to use the version from the installer instead
14:13 texinwien It seems, if I'm not terribly mistaken, that before all of this, Cygwin was using Windows git and my IDE was using Cygwin.
14:14 jast cygwin has its own git
14:14 texinwien right, but I hadn't installed it prior to all of this.
14:14 jast *normally* cygwin and msys (the basis of the installer thingy) don't get their paths mixed up
14:14 texinwien And I could still use git from Cygwin.
14:14 [worksti] why not just ditch cygwin and use git normally ?
14:15 texinwien I like the idea of having a nominal *nix installation that I can run in Windows.
14:16 texinwien Also for things other than git.
14:18 ResidentBiscuit joined #git
14:18 texinwien strangely, after running the windows and cygwin git installers, my Program Files/Git folder is empty. It's there but empty.
14:18 peterbecich joined #git
14:19 pijiu joined #git
14:20 bergice_ joined #git
14:21 texinwien dunno, maybe I just need to take my time and try to work through it from the beginning. It boggles my mind that getting this all working correctly can be so complicated.
14:22 AaronMT joined #git
14:24 manuelschneid3r joined #git
14:25 knudLeopold joined #git
14:25 nappy[m] joined #git
14:25 reactormonk[m] joined #git
14:25 olbat[m] joined #git
14:25 sanday[m] joined #git
14:25 primeos[m] joined #git
14:25 dikiaap joined #git
14:25 UncleCJ joined #git
14:25 flo[m] joined #git
14:25 aspiers[m] joined #git
14:25 cverna joined #git
14:25 patrickr[m] joined #git
14:25 acme[m] joined #git
14:25 spinningarrow[m] joined #git
14:25 hiq[m] joined #git
14:25 abithvijayan[m] joined #git
14:25 Demos[m] joined #git
14:25 pvn[m] joined #git
14:25 aidalgol joined #git
14:25 unclechu joined #git
14:25 torarne joined #git
14:25 Miguelngel[m] joined #git
14:25 dyce[m] joined #git
14:25 waltervargas[m] joined #git
14:25 Techie[m] joined #git
14:25 dreadful[m] joined #git
14:25 Remramm joined #git
14:25 aviraldg joined #git
14:25 musicmatze[m] joined #git
14:25 yanbbyan joined #git
14:25 swalladge[m] joined #git
14:25 KevinMGranger joined #git
14:25 am2on joined #git
14:25 jstimm joined #git
14:25 ushaNg6i[m] joined #git
14:25 colonelkrud[m] joined #git
14:25 JerkyMcJerkface[ joined #git
14:25 Cabanossi joined #git
14:27 bitmod joined #git
14:29 menip joined #git
14:30 dreiss joined #git
14:30 sleepingforest joined #git
14:31 peterbecich joined #git
14:34 texinwien aight, I think I got it sorted. I uninstalled git from Cygwin and installed it again on Windows. the path to Windows git was already set in Cygwin (I probably set it when I installed Cygwin, but just forgot in the meantime).
14:35 texinwien Initial tests look good. Thanks, everyone, for helping me with my rebase. It went swimmingly once everything was in place.
14:35 texinwien The rebase took about 5 minutes. Getting everything else right probably cost me 12 hours.
14:36 lpapp joined #git
14:36 foodSurprise joined #git
14:36 lpapp hi, what is the use case of git pull --ff-only
14:36 Es0teric joined #git
14:37 foodSurprise yo. I cloned a repo and made changes. I want to change that folder to my own git repo / branch. If I just throw the folder in my repo's folder I get issues about a submodule that won't commit
14:37 foodSurprise how do I switch one .git/ to my own repo/branch
14:39 foodSurprise i don't care about preserving the branch histoty of the repo I want to switch from
14:42 jast foodSurprise: the folder contains a .git subdir with the commit/branch metadata and such, remove that and it turns into a normal folder with files that you can add to a surrounding git repo
14:42 jast since you don't want to keep the history of this stuff, that should work out just fine
14:43 jast just don't remove the .git in your own repio :)
14:44 tcorneli joined #git
14:44 YuGiOhJCJ joined #git
14:45 Murii|osx joined #git
14:46 imack joined #git
14:46 tinanoo joined #git
14:46 foodSurprise ok thanks
14:47 Es0teric joined #git
14:48 synthroid joined #git
14:49 tnecniv joined #git
14:49 SwiftMatt joined #git
14:49 crose joined #git
14:50 achadwick joined #git
14:51 noecc joined #git
14:53 StuartMI joined #git
14:54 nicolai86 joined #git
14:55 Cabanossi joined #git
14:56 courrier______ joined #git
14:59 neunon joined #git
15:00 FrostCandy joined #git
15:01 Acerific joined #git
15:01 Acerific joined #git
15:01 ForeignBiscuit joined #git
15:02 ForeignBiscuit joined #git
15:02 ForeignBiscuit joined #git
15:02 ForeignBiscuit joined #git
15:07 roelmonn_ joined #git
15:07 SwiftMatt joined #git
15:11 cverna joined #git
15:13 Gustavo6046 joined #git
15:15 tisquib joined #git
15:15 tisquib sorry if off topic,
15:15 tisquib I'm looking for the graph view of my branches and commit timeline
15:15 tisquib on github
15:15 steeze joined #git
15:15 tisquib anyone using that?
15:15 Es0teric joined #git
15:17 grawity afaik that's under Insights → Graphs → Network
15:17 tango_ tisquib: #github is probabl a better place to ask
15:17 tango_ grawity: isn't that just about the forks of your repo?
15:17 grawity no
15:18 xall joined #git
15:18 lkolstad joined #git
15:23 bronson joined #git
15:24 watabou_ joined #git
15:25 roelmonnens joined #git
15:26 cagomez joined #git
15:29 Murr joined #git
15:31 peterbecich joined #git
15:32 synthroid joined #git
15:34 tribly joined #git
15:34 tribly joined #git
15:34 Lagrik joined #git
15:34 Lagrik left #git
15:37 gloomy joined #git
15:38 tchan joined #git
15:38 venmx joined #git
15:38 venmx_ joined #git
15:39 imrg joined #git
15:42 xall joined #git
15:43 Borg- joined #git
15:43 raffo joined #git
15:44 Es0teric joined #git
15:44 lpapp left #git
15:45 sebboh` joined #git
15:45 se1 joined #git
15:47 freimatz joined #git
15:47 radial joined #git
15:48 pijiu2 joined #git
15:48 madewokherd joined #git
15:48 jknetl joined #git
15:50 Guest67164 left #git
15:50 jknetl joined #git
15:51 levex joined #git
15:51 radial i've made a commit, and now i've changed my mind.. and i want to go back to the commit before that.. but i don't want to lose history.. i just want to continue on as if i hadn't made that last commit, but keep that commit in history should i ever want to reference it in the future
15:51 radial how do i do that?
15:51 glad joined #git
15:52 svm_invictvs joined #git
15:52 zmt00 joined #git
15:52 kyan joined #git
15:53 Mattiaslndstrm joined #git
15:53 tnecniv joined #git
15:54 overlord_tm joined #git
15:54 f3r70rr35f joined #git
15:56 radial my normal way to hack around this is to check out the last good commit i want, then manually copy over the changed file(s) to some temporary location, check out master, and manually replace the originals with the copied files and do a commit
15:56 radial there's got to be a better, more idiomatic way to do this in git
15:56 Kiruwa You're going to need a second branchname. Do you want it to refer to your current head? or the new attempt?
15:56 MacWinner joined #git
15:57 radial i guess to my current head
15:57 j08nY joined #git
15:57 leeN joined #git
15:57 radial actually, i'm not sure what you're asking
15:57 grawity the simplest method would be 1) branch off (git branch wip-fancy-new-thing), 2) rewind the current branch to commit-before-that (git reset HEAD^)
15:57 Kiruwa Then make a new branch at your current location. just with git branch, then git reset HEAD^ whichever one you want to backup
15:58 grawity or, alternatively, keep it in the current branch, but start the *new* branch at commit-before-that – then it's just one command (git checkout -b wip-whatever HEAD^)
15:59 durham joined #git
16:00 dreiss joined #git
16:00 Kiruwa radial: You're currently working on a branch with a name.  If you wanted to throw away the current work all you'd need to do is move that name to an earlier location (git reset HEAD^), but since you want to keep the current work, you need a new name for it.
16:01 Kiruwa ^ clarification of my earlier questions
16:01 radial ok, so then i'd have a new branch with its HEAD at the commit i want
16:01 radial so next i'd have to merge it back in to master?
16:01 thiago joined #git
16:01 Kiruwa hmm? probably not... unless you decide to
16:02 grawity *with its head, and no, in this situation there's no merging involved
16:02 roelmonnens joined #git
16:02 radial well, if not, then i'd have to keep working in the new branch from here on out?
16:02 Kiruwa you can use `git checkout X` to bounce between branches in git
16:02 grawity branches are free to create and free to switch
16:02 courrier_______ joined #git
16:03 tisquib Im trying to get a directory from a nother branch
16:03 radial but what happens if i need to do it again in the future?  i'd create another branch?  and then yet another branch?
16:03 radial why not just merge back in to master?
16:03 tisquib I read its enough to do   git checkout otherbranchname -- directoryname
16:03 tisquib but nothin seems to happen, no error nothing
16:03 grawity radial: what would merging to master achieve?
16:03 igemnace radial: what sort of history do you want to end up with anyway
16:03 grawity radial: you did say you only want to keep those misc commits for historical reference
16:04 igemnace A-B-good-bad, you want it to be A-B-good-bad-good ?
16:04 Borg- tisquib: just tag them
16:04 igemnace you can always git revert
16:04 Kiruwa tisquib: hmm... does it work if you try directoryname/*?  directories don't really exist in git :/
16:04 grawity radial: you could alternatively *undo* the commit using `git revert`, yes
16:04 Borg- no need to create branches.. if you want to keep them for history only
16:04 mkali left #git
16:04 igemnace otherwise branching off one of the two is the best solution
16:04 radial grawity: a cleaner history, without all these extra branches that seem to only exist in order to fix screwups like the one i originally described
16:04 grawity git-revert is practically a direct translation of "meh I changed I mind, undo that"
16:05 grawity radial: but, uh, merging would achieve the exact opposite of "a cleaner history"
16:05 radial grawity: but i don't want to lose the commit.. i might need it later
16:05 Borg- radial: TAG it!
16:05 grawity radial: I'm not sure what sort of screwups you're describing
16:05 radial grawity: well, after the merge i could delete that branch i created
16:05 radial then i'd have just one branch: master
16:05 grawity radial: but that'd make the history messier, not cleaner.
16:06 tisquib Kiruwa, no that also does nothing
16:06 radial grawity: the screwup was that i'd made a commit that i don't want (except for historical reference)
16:06 grawity you can keep miscellaneous branches locally, and they don't automatically become public history
16:06 grawity however, merging them to master *does* put those "screwups" in history
16:06 tisquib Borg-, sorry how do I tag something in?
16:06 taki joined #git
16:06 igemnace radial: what do you want your history to look like? do you want your master branch to retain that bad commit? or do you want it floating around somewhere, but *not* garbage collected?
16:06 Borg- tisquib: huh? git tag <tagname> [<sha1>]
16:07 Borg- for example...
16:07 Kiruwa tisquib: I suspect the conversations got crossed :)
16:07 tisquib hm,
16:07 grawity radial: perhaps you're thinking of hg/mercurial, where branches are a permanent thing
16:07 radial grawity: well, the screwup is already history.. it's the commit i don't want.. but the fix of creating a branch need not be history if i merge the branch and then delete it, right?
16:07 grawity radial: not right at all
16:07 tisquib maybe there is a git copy from branch?
16:07 tisquib surprised I cant checkout a directory recursively
16:07 tisquib I have to remember all the file names
16:07 Borg- tisquib: you can
16:08 grawity radial: in Git, "history" is per-branch, not per-repo
16:08 grawity radial: you can have dozens of private branches and just don't publish them
16:08 Borg- tisquib: git checkout <something> -- dir/ <- this does not do the job?
16:08 grawity radial: it won't affect the "main" history (i.e. the master branch)
16:09 radial igemnace: i want the commit to stay in master the way it is.. i just want the newest commit in master to reflect the commit before the bad commit i don't want, and i want to continue development on master from then on
16:09 Borg- tisquib: if not.. wrait custom dir-checkout script.. easy to do
16:09 Borg- w/wrait/write/
16:09 peterbecich joined #git
16:09 igemnace radial: that's git revert
16:09 grawity radial: well then revert it
16:09 torshido joined #git
16:09 Borg- radial: you cannot DO IT!!! its not how git works
16:09 radial so revert will still keep the bad commit?
16:09 Borg- yes
16:09 igemnace radial: yes
16:10 radial Borg-: well, i've done it in the past by working around git by simply checking out the last good commit, copying the affected file(s) to a temporary location, checking out master again, and then replacing the file(s) with the copied ones, and doing a new commit
16:11 Cabanossi joined #git
16:11 \drew joined #git
16:11 Borg- radial: but, isnt that STUPID?
16:11 Kiruwa that's WAY too much work for what git can do for you in a command or two
16:11 radial so it is definitely possible.. but i was just looking for a way to do it in git instead of manually
16:11 igemnace radial: git revert does the same thing
16:11 grawity radial: you've been told about it quite a few times
16:11 radial Kiruwa: yeah, exactly.. that's why i'm asking here
16:11 Borg- radial seems to be lunatic.. who doesnt understand VCS...
16:11 Borg- EOF from my side...
16:11 grawity radial: it's called `git revert`
16:11 radial grawity: well, now i've been told at least 3 different things... including "you cannot DO IT!!! its not how git works"
16:11 grawity radial: it does exactly what you've described
16:12 upperdeck joined #git
16:12 InfoTest joined #git
16:12 radial ok, i'll read up on git revert
16:12 radial thank you
16:12 Borg- good.. finaly
16:12 Borg- :)
16:12 Borg- because I want to write a rant...
16:12 tisquib hm, how do I list last commit message?
16:12 Kiruwa I really didn't understand that comment, tbh.  Although, what you've described you wanted and what you're doing with either revert or your manual method have nothing to do with each other
16:12 radial Borg-: please be civil
16:12 tisquib git log is a bit verbal
16:12 grawity tisquib: git log -1
16:12 tisquib ah ok
16:13 Borg- radial: its hard to be.. these days :(
16:13 radial Kiruwa: ok, so revert is not what i want?
16:13 Jackneilll joined #git
16:13 grawity based on your *last* description, it's definitely what you want
16:13 Kiruwa What you've described and what you do (manually) are different things
16:13 Borg- radial: you want revert...
16:13 grawity since it does exactly what you manually do
16:14 grawity but, frankly some of your descriptions *did* sound a lot like hg/mercurial workflow, where things work differently from git
16:14 Kiruwa If you want to emulate your manual process, you want revert
16:14 Borg- radial: but, you also have misconceptions about wcopy (snapshot) and history.. I think
16:14 vjacob joined #git
16:14 cdg joined #git
16:14 radial Borg-: how can i clear up those misconceptions?
16:14 Borg- radial: okey, time for ascii drawing..
16:14 Kiruwa Just try revert... see if it does what you want
16:14 Borg- lets say you have this history: a-b-M
16:14 Borg- M is mistake commit...
16:15 Borg- yeas, you can do link like.. checking out b.. and commiting it.. so it will looks like this
16:15 Borg- a-b-M-B
16:15 Borg- where b!=B (its different commits)
16:16 Borg- so you are back to b state.. having extra commit..
16:16 radial yep
16:16 Borg- what revert will do..
16:16 radial and that's ok by me
16:16 Borg- a-b-M-M'
16:16 Borg- with this history.. you are basicaly again at b.. but much more logical way..
16:16 Borg- you reverted M commit... (commit log will say so)
16:17 radial Borg-: i understand a-b-M-B  but what is a-b-M-M' ?
16:17 Borg- M' is revert commit of M
16:17 grawity same general idea, different drawings
16:17 Borg- grawity: yeah, but different workflow...
16:17 grawity .digraph a<-b<-c<-bad<-"revert of bad"
16:17 gitinfo grawity: .digraph error processing the definition: Warning: non-terminated HTML string starting line 2, file <stdin> | Error: <stdin>:3: syntax error near line 3
16:17 radial grawity: so the state of my files at M' == the state of my files at B ?
16:17 grawity oh come on
16:17 grawity radial: yes
16:17 radial ok
16:18 Borg- radial: yes!! but only in that particual case!
16:18 Borg- radial: imagine this history: a-b-M-c-d
16:18 Borg- you want to revert M...
16:18 Kiruwa tisquib: Hmm... I just tested it.   `git checkout hasdir_branch -- adir` worked fine for me
16:18 Borg- how would you do that now? :) with your checkout :D
16:18 radial Borg-: to "revert M" means to have a new commit of B?
16:18 Borg- radial: no!
16:19 tisquib Kiruwa, bot branches have this dir for me,
16:19 Borg- radial: to revert M you will have to commit invert changes of M.. having commit M'
16:19 tisquib I want all the changes from the other,
16:19 tisquib maybe I had to remove local dir first?
16:19 Borg- bascialy.. looking at unidiff.. its - -> + and + -> i.. or patch -R
16:19 Borg- and thats correct thinking about reverting something
16:19 Kiruwa yeah... git won't overwrite when updating
16:19 Borg- not bringing back snapshot of past commits...
16:20 Borg- radial: now, it sounds clear to you?
16:20 radial Borg-: i don't know what unidiff is or what "- -> + and + -> i" or "patch -R" refer to
16:20 SwiftMatt joined #git
16:20 Borg- radial: hmm how come? did you ever done git diff ? :)
16:20 radial Borg-: i have
16:20 Borg- git displays diffs by default in unidiff format..
16:20 chrisshattuck joined #git
16:21 Borg- unified diff..
16:21 Gurkenglas joined #git
16:21 Borg- radial: IDE user?
16:21 grawity radial: in short: if you do it immediately after M, then it's the same
16:21 ToBeCloud joined #git
16:22 Borg- radial: yeah.. as grawity said.. . but its not always that simple case..
16:22 radial Borg-: ok, so you're saying that in the case of reverting the M commit in a-b-M-c-d  i'd get a-b-M-c-d-M' where M' is d with the changes that M made to b undone?
16:22 grawity sounds about right
16:22 Borg- yep.
16:22 radial ok
16:22 radial that's cool
16:23 chatose joined #git
16:23 upperdeck joined #git
16:23 igemnace radial: this helped me with all the necessary concepts and terminology https://git-scm.com/book/en/v2
16:23 h12o joined #git
16:24 radial igemnace: thanks
16:24 igemnace radial: it'll make reading the git help <command> pages easier
16:24 igemnace so you can consult git help revert in the future
16:24 radial great
16:25 Borg- NOOOOOO.. compilation failed..
16:25 Borg- *headashe*
16:27 plasticle joined #git
16:27 akushner___ joined #git
16:28 knudLeopold joined #git
16:29 jstein_ joined #git
16:29 lafleurdubien joined #git
16:29 Tobbi joined #git
16:29 orbyt_ joined #git
16:29 plasticle joined #git
16:30 jimmyrcom_ joined #git
16:32 cdg_ joined #git
16:33 maryo_ joined #git
16:34 _ADN_ joined #git
16:35 xall joined #git
16:36 courrier________ joined #git
16:37 kurkale6ka joined #git
16:38 radial in magit, when i run the revert command, it prompts me to "Revert commit(s)" or "Revert changes", which of those do i want?
16:39 radial and if i try to "Revert commit(s)" it asks me to revert "master~1", "HEAD", or "master"
16:39 radial which of those should i choose?
16:39 * radial is a bit confused...
16:39 bronson joined #git
16:40 Kiruwa I doubt too many people here use an interface like that
16:40 JanC_ joined #git
16:41 Borg- ugh.. git porcelain inside emacs ;)
16:42 igemnace radial: revert commit, HEAD should be fine
16:42 igemnace your last commit is what you want to revert, yes?
16:42 radial yes
16:42 radial igemnace: what's the difference between all those?
16:42 orbyt_ joined #git
16:43 freimatz joined #git
16:43 watabou joined #git
16:43 igemnace Revert changes vs Revert commit sounds silly to me
16:44 Kiruwa HEAD is the most recent commit.  master is the most recent commit on branch master. master~1 is one commit back from the tip of master IIRC
16:44 igemnace master~1 means the 1st parent commit of the commit referenced by master
16:44 radial Kiruwa: so if i'm already on master, then reverting "master" and "HEAD" would be equivlanet?
16:44 igemnace HEAD is a special marker in git, it's always the last commit you checked out
16:44 radial equivalent, even
16:46 radial so if my commit history was a-b-c-M and i reverted HEAD or "master" it would now be a-b-c-M-c  and if i reverted "master~1" it would be a-b-c-M-b ?
16:46 Borg- radial: no
16:47 igemnace not exactly. reverting master~1 would give you a-b-c-M-x
16:47 igemnace where x is a commit that has M and c changes but not b changes
16:47 Borg- where X is M without b changes
16:47 igemnace what? wait a second, no
16:47 Borg- igemnace: stop it!
16:47 Borg- ionelmc: you failing into his trap :D
16:47 svm_invictvs joined #git
16:47 igemnace x that has M changes but not c changes
16:48 igemnace my bad
16:48 ionelmc left #git
16:48 igemnace counted 1 too many commits away from tip...
16:48 Borg- oh right.. ;)
16:48 Borg- radial: please... why you think akward...
16:48 radial i'm just trying to understand
16:49 Borg- radial: its simple: git revert <sha1_commit> will revert changes that done that commit.. nothing else
16:49 radial ok
16:49 radial makes sense
16:49 Borg- and... it will do no matter where that commit is located in history
16:49 radial i see
16:50 radial thanks, that's nice and simple and clear
16:50 Borg- radial: why you use that weird emacs stuff? what was called?
16:51 radial magit
16:51 texinwien joined #git
16:51 Borg- yeah... itsnt better and more clear to use git from itself?
16:51 h12o joined #git
16:51 radial because it's a lot less keystrokes, and it has an nice menu system and nice visualizations
16:51 radial and allows me to pick what to commit easily
16:51 radial it's all nice and integrated in to emacs
16:51 Borg- well... it might be true.. but also.. it hides some git internals from you.. and can lead to confusion
16:51 radial a lot less typing and less hassle
16:51 radial lots of people like it.. not just me
16:52 radial yeah, it hides git internals, but so do git commands
16:52 Borg- yeah.. lots of ppl like *clickty* *clickty* gui shit...
16:52 r00twrh joined #git
16:52 Borg- does it mean they are more productive?... NO!
16:52 Borg- ;)
16:52 Borg- but thats just my 2 cents.. biased
16:52 radial it's all typing, no clicking
16:52 radial but it is a gui, i admit
16:53 radial and it's faster and more efficient, imo
16:53 Borg- radial: well, GIT can be used with nice bash autocompletion and etc stuff
16:53 radial still too much typing.. and that means i have to switch to bash
16:53 Borg- I personaly use plain bash... its ok.. you can still you emacs like stuff in bash.. to bring and edit history.. etc..
16:53 igemnace magit is as faithful to git internals as i've seen in *any* git porcelain
16:53 radial i'd prefer to stay in emacs
16:53 igemnace but learning magit before git could still teach you the wrong lessons
16:53 Borg- radial: haha :D can I cite good old IRC stuff from tao-of-irc ? :D
16:53 radial well, that's why i'm asking here :)
16:54 igemnace the git-scm book is fine reading material to accompany you using magit
16:54 radial yeah, that looks like a good book
16:54 radial have to schedule some one-on-one time with it...
16:55 radial but in the mean time... since you all are so helpful... :)
16:55 radial i could just ask here :)
16:55 Borg- yeah.. I noticed that this channel is very polite and helpfull
16:55 Borg- would NOT happen in my times... ;)
16:55 Cabanossi joined #git
16:56 MineCoins joined #git
16:56 chll_ joined #git
16:58 d10n-work joined #git
16:58 \drew joined #git
16:59 Mattiaslndstrm joined #git
17:00 kadoban joined #git
17:01 gxt joined #git
17:03 dave0x6d joined #git
17:05 j08nY joined #git
17:06 aFeijo joined #git
17:06 hahuang65 joined #git
17:06 heym0e joined #git
17:09 Cuken joined #git
17:09 Noldorin joined #git
17:10 radial i need some help in using git bisect..  i'm trying to find a bad commit, but i want git bisect to skip a whole bunch of commits early on in my commit history because i know all those are good
17:10 radial how can i tell git bisect that all commits up to a certain one are good?
17:10 Cuken Hello all, I need help understanding the easiest way of accomplishing something (or even if it makes sense to do this in git). I currently have a repo setup that has a Master and Development branch. I have vendor specific changes I need to make to a large portion of the package that are unique to each vendor. I was thinking of making them seperate repo's entirely, but I was wondering if there was something clever I could do with rebasing
17:11 sonOfRa radial: git bisect good <ref>
17:11 sonOfRa where ref can be a sha1, a tag, a branch, whatever you please
17:11 armyriad joined #git
17:11 grawity isn't that something you're required to specify when starting the bisect...
17:11 Cuken I'd like to make a commit to dev and then choose via rebase?? which vendor gets the update, but keep the packages in sync that way. So in the event the file that gets updated is not vendor specific, I'm guranteed their all using the same version
17:11 radial sonOfRa: thanks
17:12 sonOfRa And of course "git bisect bad" to mark the current revision as bad
17:12 radial yep
17:12 Kiruwa Cuken: I'd think you'd want to use merge instead... and perhaps merge without accepting changes for the "wrong" vendor.  With a release branch for each vendor?
17:13 Cuken Kiruma: So have a base branch (dev / master) then merge that into each vendor specific release, rejecting changes that are vendor specific?
17:14 Kiruwa We do something like that for internal release branches
17:14 Cuken Do you guys also have a dev variant for each release branch that track the release specific changes, or do you commit directly to the release?
17:15 mizu_no_oto joined #git
17:18 crose joined #git
17:19 upperdeck_ joined #git
17:21 qswz joined #git
17:21 qswz I have 2 remote branches, and only one of them locally
17:21 qswz I can't find a way to get the other
17:21 qswz tried: git pull --all
17:22 grawity 1) git fetch
17:22 grawity 2) look at `git branch -r`
17:22 Kiruwa Cuken: Our dev process is actually fairly complicated... but to simplify significantly, we only have the single developmental trunk.
17:22 grawity 3) if it doesn't show up, `git config -e` and check the fetch refspec
17:22 qswz origin/HEAD -> origin/gh-pages
17:22 qswz origin/gh-pages
17:22 qswz origin/master
17:23 grawity if the refspec is for a specific branch only, change it to +refs/heads/*:refs/remotes/origin/*, then GOTO 1
17:23 Cuken Kiruwa: Thanks man, this helped
17:23 grawity 4) check out the desired branch
17:23 qswz I'd like master
17:23 grawity git checkout master
17:23 Gsham joined #git
17:23 qswz oh thanks
17:23 Kiruwa Cuken: If at all possible, figure out if you can make your build system distinguish between customers though
17:24 qswz it was simple
17:29 mlindner joined #git
17:29 thierryp joined #git
17:30 mlindner hi
17:30 mlindner I'm trying to install git, but I need a pre-built binary, does anyone know where I can acquire that?
17:31 grawity for Windows, Linux, osX, FreeBSD, or TempleOS?
17:31 qswz apt install git
17:31 mlindner qswz: its an ancient CentOS release, current version of git is 1.7.1
17:31 mlindner thus why I need binary
17:32 qswz hmm 2.11.0 here
17:32 qswz bit old yea
17:32 mlindner yes it was released 2010
17:32 grawity you can make binaries with `git clone git://github.com/git/git && cd git && make`, can't you
17:32 qswz you could update the apt-sources
17:32 mlindner grawity: well I also don't have root privledges
17:32 mlindner and the documentation won't build
17:32 mlindner as its missing many things
17:33 grawity you don't need root privileges, nor I guess the docs
17:33 grawity qswz: centos doesn't use apt
17:33 mlindner so this is why I want a pre-built binary
17:33 grawity mlindner: there are sites which index .rpm's from many distros
17:33 grawity download a more recent one manually, unpack it, and pray that all the libraries are still there
17:36 mr_sm1th joined #git
17:37 Kavothe joined #git
17:40 zhodge joined #git
17:40 upperdeck joined #git
17:41 upperdeck_ joined #git
17:41 Es0teric joined #git
17:41 osse at my previous job I downloaded RPMs and rpm -fi them all the time
17:42 osse good times
17:42 osse as for the docs, that's a separate target
17:43 Plouescat joined #git
17:45 zhodge I'm working on a simple deployment workflow on Windows to copy files from a bare Bonobo repository to a target directory (on the same server) based on the post-receive hook
17:45 TomyLobo joined #git
17:45 zhodge using the hook detailed here https://gist.github.com/noelboss/3fe13927025b89757f8fb12e9066f2fa#file-post-receive
17:46 zhodge going smooth except for "error: unable to unlink old {file} (Permission denied)" messages
17:46 upperdeck joined #git
17:46 zhodge I've tried what appears to be the fix for that: `chmod -R g+w` over my "$TARGET" dir
17:46 bongjovi joined #git
17:47 zhodge still no dice. any Windows users experience anything similar?
17:47 zhodge I executed chmod from the Git Bash executable, not sure if that's where I should be making the permissions modifications from
17:50 osse zhodge: is the file open in another prgoram?
17:50 raynold joined #git
17:50 kfunk as I'm basically setting this on every machine I get access to: has there been discussion about setting the rebase.stat config to true by default? (cf. https://git-scm.com/docs/git-rebase)
17:50 kfunk can someone elaborate why it isn't true by default? /me can't think of reasons
17:51 Marvel joined #git
17:51 osse kfunk: maybe because rebase didn't support --stat for a while
17:51 akushner___ joined #git
17:52 energizer joined #git
17:52 kfunk should I report a bug/wishlist item upstream maybe? what'cha thinking?
17:52 osse default settings have been changed before, so why not
17:53 Plouescat joined #git
17:54 kfunk ok. ah, and as I see I need to write to the mailing list instead.
17:54 sbeyer joined #git
17:54 MattMaker joined #git
17:54 osse kfunk: git hella old school. ain't no issue tracker up in here, kiddo
17:54 naf joined #git
17:54 orbyt_ joined #git
17:55 zhodge osse: after checking, I don't have anything open directly. IIS is serving a few sites whose paths are the parent directory to my deployment target however
17:55 naf joined #git
17:56 skankhunt41 joined #git
17:58 texinwien joined #git
18:00 kfunk osse: haha :)
18:01 Zialus joined #git
18:01 osse kfunk: maybe even git-rebase is so old it's from the days where git didn't even do its own diffing, but used diff and diffstat
18:01 Sasazuka joined #git
18:02 zhodge interesting. granting Modify & Write permissions from the Windows Security panel in File Explorer to my target dir allowed both file/directory creation & unlinking...
18:03 osse zhodge: maybe it's easier to put the files somewhere else, then rename the directories and such
18:03 osse for atomicsnicity
18:04 salamanderrake joined #git
18:04 mlindner osse: how do you use rpm -fi, I get "package x is not relocatable"
18:04 zhodge osse: by files do you mean my deployment target and its contents?
18:05 osse zhodge: I guess... I mean the things in your repo you want to put somewhere :P
18:05 cyr5234 joined #git
18:05 osse mlindner: maybe I was wrong about the -f. I just meant force install
18:06 mlindner osse: it won't
18:07 indieross joined #git
18:07 courrier________ joined #git
18:07 cyr5234 can I ignore a remote file?
18:08 wizardedit joined #git
18:08 wizardedit joined #git
18:08 elsevero joined #git
18:08 kadoban cyr5234: You can't ignore files that are tracked
18:08 kadoban (so probably no, making some assumptions on what you mean)
18:08 JanC_ joined #git
18:08 zhodge osse: ah so my repo contents are coming from the bare Bonobo repository
18:08 cyr5234 kadoban, I have a submodule, but I only want to pull one folder from that submodule, rather than the entire submodule.
18:09 zhodge and then being transferred to the IIS-served location via `git --work-tree=$TARGET --git-dir=$GIT_DIR checkout -f`
18:09 zhodge where TARGET={IIS dir} and GIT_DIR={Bonobo bare repo}
18:10 zhodge I'll see how it plays out. happy at least it's "working for now TM" :)
18:10 zhodge definitely not too familiar with running git (especially servers) on Windows...
18:10 cdown joined #git
18:10 kadoban cyr5234: Yeah, you can't really do that, unless they added that. You'd probably have to do something like create your own copy of the submodule repo and nuke everything else. You can do that in a sane-ish way with the git subtree contrib command, or you can just do it manually with filter-branch fun.
18:11 Cabanossi joined #git
18:11 upperdeck joined #git
18:11 cyr5234 (I want to include a submodule twice, at two different folder paths. I want to exclude half the contents from one path, and the other half from the other path. This saves me from having two remote submodule repos.)
18:13 \drew joined #git
18:14 Gustavo6046 joined #git
18:15 cyr5234 kadoban, hmm, I'd prefer not to introduce subtrees for this... the other option I was thinking was pointing the second submodule to a second "master" branch, but that will effectively hide that second branch on github, which is not ideal. How would filter-branch work for this?
18:16 mizu_no_oto joined #git
18:20 zumba_ad_ joined #git
18:20 zumba_ad_ i went too much on git reset --soft HEAD~3, how do I forward to newer
18:24 zumba_ad_ or how do I undo all the git reset that I execute few minutes ago?
18:24 osse zumba_ad_: git reset branchname
18:24 osse No
18:24 zumba_ad_ ok
18:24 osse git reset HEAD@{n}
18:24 osse Where n is 1 or 2 or...
18:25 osse Use git reflog to find out
18:25 zumba_ad_ i'll try it again
18:25 Ricaz joined #git
18:25 zumba_ad_ the last commit in gitlab says "changes" on the title
18:25 zumba_ad_ but I don't see it with reflog
18:26 zumba_ad_ but i saw it before I ran git reset earlier
18:26 osse then you can do git reset origin/master (for example)
18:26 zumba_ad_ got it
18:27 zumba_ad_ i just want to go back to the original master state just like after a git pull
18:27 courrier________ joined #git
18:27 zumba_ad_ yes!
18:27 zumba_ad_ :D
18:27 zumba_ad_ thanks
18:27 cdg joined #git
18:28 Nizumzen joined #git
18:30 upperdeck joined #git
18:30 wootehfoot joined #git
18:33 belmoussaoui joined #git
18:34 cdg joined #git
18:35 ToBeCloud joined #git
18:36 roelmonnens joined #git
18:42 Noldorin joined #git
18:43 nixjdm joined #git
18:46 thierryp joined #git
18:48 netj joined #git
18:50 hdqn joined #git
18:50 paul424 joined #git
18:51 upperdeck joined #git
18:51 hdqn left #git
18:51 hdqn joined #git
18:51 hdqn left #git
18:51 misosoup joined #git
18:51 hdqn joined #git
18:52 hdqn how can I view the git diff in an external editor?
18:52 cyr5234 hdqn, what os?
18:52 hdqn cyr5234: Arch
18:53 hdqn bah, Linux
18:54 cyr5234 hdqn, download a gui client and point it at your git dir?
18:54 freimatz3 joined #git
18:55 bb_mac left #git
18:55 halfa joined #git
18:56 bb_mac joined #git
18:58 orbyt_ joined #git
18:58 bb_mac Hello everyone. I have some general and specific questions...I have inherited a Source Tree + Git setup from my predecessors and am lost.
18:58 peculiaryak joined #git
18:58 \drew joined #git
18:58 radial left #git
19:00 selckin !starting_point
19:00 gitinfo Good starting points for getting to know Git are: official website http://git-scm.com/ (esp. "Documentation"), http://stackoverflow.com/tags/git/info, http://try.github.com, https://git.wiki.kernel.org/index.php/Main_Page,  http://rogerdudler.github.io/git-guide/ and  https://www.atlassian.com/git/tutorial/git-basics
19:00 Es0teric joined #git
19:01 bb_mac One of the base issues I have is that SourceTree does not recognize the git repository as such
19:01 bronson joined #git
19:02 doyle_ joined #git
19:02 selckin then don't use it
19:02 bb_mac From what I can tell, there is a directory, and it has the .git folder, it has been initialized and performed a commit, but SourceTree tells me "This is not a valid source path / url"
19:02 bb_mac It was set up prior to me and is in place for production already
19:03 bb_mac I need it for a dev version
19:03 hahuang65 joined #git
19:04 bb_mac i cannot tell if this is a git issue or source tree issue, i spent a couple days with google and yesterday with some people on #atlassian
19:04 bb_mac they suggested coming here
19:04 osse Doesn't git on the command line work?
19:04 quenode joined #git
19:05 bb_mac I think it does, but I am not positive
19:05 bb_mac The thing is, I can point to 2 other directories from Source Tree and it recognizes those no problem
19:06 bb_mac So I wonder if the directory I am trying to set up, is wrong
19:06 MattMaker joined #git
19:06 bb_mac what can I look at to determine if it is set up properly?
19:07 osse I don't know how sourcetree works, but check if git itself can make sense of it
19:07 LeBlaaanc joined #git
19:08 mikecmpbll joined #git
19:10 MattMaker joined #git
19:10 MattMaker joined #git
19:10 bb_mac running 'git status' from the directory I am attempting to connect to gives me  https://thepasteb.in/p/JZhpEXOyAMAsg
19:10 Cabanossi joined #git
19:12 robotroll joined #git
19:13 ewitz joined #git
19:13 sz0 joined #git
19:18 chriys_ joined #git
19:18 chriys_ joined #git
19:22 hahuang65 joined #git
19:22 safe joined #git
19:25 aard_ joined #git
19:28 Murr joined #git
19:28 preyalone joined #git
19:29 osse bb_mac: things look good
19:29 osse to git at least
19:31 bb_mac strange
19:33 noecc left #git
19:37 m1dnight_ joined #git
19:37 synthroid joined #git
19:39 voker57 joined #git
19:39 voker57 joined #git
19:41 cdown joined #git
19:42 EvilPenguin joined #git
19:42 osse bb_mac: Hate to send you back, but seems like there's nothing wrong in git's own eyes
19:43 osse bb_mac: You can try running 'git init' in the repo to "reinitialize" it
19:43 kerrick joined #git
19:44 btree joined #git
19:44 kerrick I ran git checkout -p HEAD because I made two changes to a file and wanted to revert one of them (changes are only on working directory, not in index or a commit)
19:44 kerrick On the prompt "Discard this hunk from index and worktree" I chose "n" for the chunk I wanted to keep and "y" for the chunk I wanted to discard
19:45 kerrick but then it printed: error: patch failed: app/scripts/my_script.py:534 \n error: app/scripts/my_script.py: patch does not apply \n The selected hunks do not apply to the index!
19:45 dviola joined #git
19:45 _ikke_ kerrick: hmm
19:45 osse kerrick: Did you also git add the things?
19:45 kerrick no I didn't git add anything
19:45 osse Or some of it?
19:46 kerrick let me upgrade my git to make sure that's not the issue
19:46 kerrick I have 2.14.1 and Homebrew is telling me that's latest
19:46 kerrick yes that is the latest per the channel title
19:46 osse kerrick: Does it also say "apply anyway?".
19:46 orbyt_ joined #git
19:46 _ikke_ 2.14.1 is indeed the latest
19:47 kerrick Hmm so if I hit 'y' at the prompt "Apply them to the worktree anyway?" it does what I want
19:47 kerrick so in spite of the scary error message it seems like it works
19:47 kerrick but this is definitely a bug right?
19:47 osse No
19:47 doyle joined #git
19:47 kerrick like isn't this what checkout -p is supposed to do?
19:47 osse Because the index doesn't contain the stuff you want to remove (because you didn't add it to the index in the first place)
19:47 kerrick it shouldn't be printing extraneous error messages and prompting the user throughout the course of normal operation
19:48 kerrick so I should git add everything and then checkout -p?
19:48 kerrick is there something like git reset --hard -p?
19:48 jnewt joined #git
19:48 osse "should" is a big word... It would get rid of the extra prompt, but that's it
19:50 osse I agree the message is annoying, but it technically makes sense. But I also think git should be able to avoid it
19:50 chriys_ joined #git
19:50 osse It could check that the index is "empty" and then skip it
19:51 aboccia joined #git
19:51 akushner___ joined #git
19:54 cagedwisdom joined #git
19:54 ZyX-I joined #git
19:59 knudLeopold joined #git
20:00 nixjdm joined #git
20:02 bronson joined #git
20:02 Gsham_ joined #git
20:03 steeze joined #git
20:03 MrMagic joined #git
20:06 matoro_ joined #git
20:06 _ADN_ joined #git
20:08 matsaman joined #git
20:08 cdg joined #git
20:08 Solvaring joined #git
20:10 finalbeta joined #git
20:14 kerrick joined #git
20:15 lxnr joined #git
20:16 kerrick joined #git
20:16 a3Dman joined #git
20:17 cinnapie1 joined #git
20:18 ertes joined #git
20:18 ArcaTech joined #git
20:19 troulouliou_div2 joined #git
20:19 troulouliou_div2 joined #git
20:21 hahuang65 joined #git
20:23 rawste joined #git
20:23 Leboeuf joined #git
20:24 cdg_ joined #git
20:24 surfist joined #git
20:26 AfroThundr joined #git
20:27 cdown joined #git
20:27 h12o joined #git
20:27 watabou_ joined #git
20:30 thierryp joined #git
20:32 hash-z3r0 joined #git
20:34 bb_mac although I never got the original issue resolved (being able to clone or add existing repo to source tree)
20:34 bb_mac I was able to clone on the source linux machine, and then add that cloned repo to Source Tree.
20:37 m0viefreak joined #git
20:38 jnavila joined #git
20:39 dege joined #git
20:40 Cabanossi joined #git
20:44 rkta joined #git
20:44 d5sx43 joined #git
20:45 mar77i_ joined #git
20:45 dre_ joined #git
20:46 raijin joined #git
20:48 bb_mac joined #git
20:52 jstimm joined #git
20:53 maxorator joined #git
20:55 theoceaniscool joined #git
20:57 cbd_ joined #git
20:58 cbd_ is there an easy way to clone an individual branch of a git repository without bringing along the .git folder?
20:59 nixjdm joined #git
21:00 osse cbd_: No. But several popular hosting services offer a zip download or similar from their web ui
21:00 _ikke_ or git archive
21:01 peterbecich joined #git
21:01 osse I've yet to encounter a server where git archive --remote actually works :P
21:01 courrier________ joined #git
21:02 akushner___ joined #git
21:02 _ikke_ any plain ssh remote
21:02 kerrick joined #git
21:02 osse Oh?
21:04 nixdork joined #git
21:05 StuartMI joined #git
21:05 jnewt joined #git
21:05 kfoonamalik joined #git
21:06 _ikke_ git itself supports it
21:06 _ikke_ git daemon, you can enable it
21:06 _ikke_ it's only the 'hosted' sollutions that do not offer it
21:07 raijin joined #git
21:07 durham joined #git
21:08 Paraxial joined #git
21:08 kerrick joined #git
21:09 Gsham joined #git
21:10 durham joined #git
21:10 jaggz can kdiff3 be setup so it can do some git management, like so I can click a file to add it to .gitignore...
21:10 jaggz or to update-index --skip-worktree
21:11 _ikke_ no
21:11 _ikke_ kdiff3 has no knowledge of git, and it has no way to specify custom actions
21:11 jaggz thanks _ikke_
21:11 pbandark joined #git
21:11 jaggz if I have compiled binaries, should I .gitignore them or skip-worktree?
21:12 fatalhalt joined #git
21:12 Gsham joined #git
21:12 _ikke_ .gitignore
21:12 _ikke_ you do not want to track them at all
21:12 Nizumzen joined #git
21:13 jaggz I did skip-worktree once, on one file.. I should untrack that
21:13 jaggz now I need to figure out how to find where that was
21:14 parsnip !log
21:14 gitinfo [!irclog] Public logs of #git are kept at: https://itvends.com/irc/git/ or see stats for the last 30 days: http://itvends.com/irc/git.html or all logs: http://itvends.com/irc/git-all.html
21:14 parsnip oops
21:14 parsnip !git-log
21:15 jaggz no #kdiff3 channel.. hrm.. I want to be able to copy one of the directory paths of files I'm examining (so I can chdir there and do things like add to .gitignore)
21:15 kerrick joined #git
21:15 jaggz parsnip, hrm.. doesn't show what I'm looking for..
21:16 parsnip jaggz: what doesn't show what you're looking for?
21:16 jaggz git log .. looking into its options to see what would list it
21:16 parsnip jaggz: i would check `man git-log` to look for arguments that might help you find it.
21:16 gitinfo jaggz: the git-log manpage is available at https://gitirc.eu/git-log.html
21:17 jaggz https://stackoverflow.com/questions/42363881/how-to-list-files-ignored-with-skip-worktree
21:18 jaggz thanks
21:18 jaggz man google
21:19 parsnip where are file flags stored? the index?
21:20 jaggz cdf () { cd ${1%/*} }
21:20 jaggz parsnip, no clue (if you're asking me).
21:20 parsnip what is skip-worktree?
21:22 jaggz I don't understand it
21:22 jaggz https://git-scm.com/docs/git-update-index
21:22 puff joined #git
21:22 jaggz https://git-scm.com/docs/git-update-index#_skip_worktree_bit
21:23 belmoussaoui_ joined #git
21:27 parsnip okay, sounds like status of skip-worktree is data stored in index
21:28 parsnip maybe `show` or `cat-file` on the commit hash will show the skip-worktree marker.
21:28 Murr joined #git
21:28 kerrick joined #git
21:34 denimsoft joined #git
21:34 leeN joined #git
21:36 raijin joined #git
21:36 \drew joined #git
21:37 King_Hual joined #git
21:38 Oatmeal joined #git
21:39 jaggz parsnip, git ls-files -v . | grep '^S'
21:40 Cabanossi joined #git
21:42 Exagone313 joined #git
21:43 akushner___ joined #git
21:45 inovas joined #git
21:46 courrier________ joined #git
21:47 Gustavo6046 joined #git
21:49 cyberz joined #git
21:52 btree left #git
21:53 kerrick joined #git
21:54 steeze joined #git
21:57 M6HZ joined #git
21:59 nixjdm joined #git
22:00 watabou joined #git
22:01 mr_sm1th joined #git
22:01 Sasazuka__ joined #git
22:02 bronson joined #git
22:04 Guest19798 joined #git
22:06 kerrick joined #git
22:10 diogenese joined #git
22:13 overlord_tm joined #git
22:14 tisquib did linus metnion that git allows him to have a "onion of trust"
22:15 tisquib Im trying to remember what he said about his git workflow
22:15 kerrick joined #git
22:15 DaGoaty joined #git
22:16 Bhootrk_ joined #git
22:16 mar77i iirc there's only one project that really uses octopus merge
22:18 azerus joined #git
22:19 nahtnam joined #git
22:21 tisquib oh?
22:21 tisquib is that call octopus merge?
22:21 tisquib what linux and kernel.org uses
22:21 tisquib layers / onion of trust?
22:23 mar77i nope, that's just the tree-shaped way code makes its way to Linus.
22:23 oojacoboo joined #git
22:23 tisquib ah, its more like a tree of trust
22:23 tisquib nat, octopus
22:24 tisquib or octopus on top of octopuses
22:24 tisquib :)
22:24 DaGoaty joined #git
22:24 cbd_ joined #git
22:24 mar77i yup. octopus merge is a feature in git where it's possible to merge many non-conflicting branches from many sources in a single go.
22:25 mattcen joined #git
22:25 hash-z3r_ joined #git
22:25 tisquib yes, indeed, I can get code review and merge from DevA... who has branch A... and he can then request PR to master from DevMaster guy/gal
22:25 tisquib so, multiple branching and merging
22:25 tisquib keeps everyone working and not waiting
22:26 LeBlaaanc joined #git
22:26 tisquib mar77i,  well, ok thats not octopus branching
22:26 tisquib I simply meant the multiple levels of branching and merging in Git
22:27 tisquib everyone even keeping his local branches if they so wish
22:27 Plouescat joined #git
22:27 tisquib I remember linus giving a lecture talking about how he manages to sort out all the garbage when merging with the kernel
22:28 tisquib he talkes about himself having a few "trusted" guys... with their own branches, and they themselves have their layer of trust... etc.
22:28 tisquib so he only had to look at what passed this filter of his layer of truste
22:29 DIDAVISION joined #git
22:31 bitmod guys how big is too big for a git repo?
22:32 bseyer joined #git
22:33 Gurkenglas joined #git
22:33 sbeyer joined #git
22:34 bluepixel joined #git
22:35 PaulCapestany joined #git
22:36 cdg joined #git
22:37 chriys joined #git
22:38 mar77i if you have time for sword fights and/or a coffee break while cloning, I guess your repo can be considered too big. that all depends on your equipment and uplink. also consider that git is made for text files and will not be able to efficiently store progressive changes in binary files.
22:38 steven joined #git
22:41 bitmod mar77i: is 2.8GB to large?
22:42 mar77i they used to have that saying in these parts "tias". try it and see™
22:43 bitmod mar77i: it's not my repo, but cloning it took about 15 minutes. I have no idea what this guy was on, but it was strong
22:43 bitmod (at 1Mb/sec)
22:43 courrier________ joined #git
22:43 traveler joined #git
22:44 bitmod mar77i: also if i have a branch called 'origin/testing', how do i access it?
22:45 xaxxon joined #git
22:45 multi_io on Github, can I create a pull request for a project that I've forked manually?
22:45 chriys joined #git
22:45 multi_io i.e. I have a fork of it, but I've created it on GitHub directly via git rather than via GitHub's "fork" button.
22:46 bitmod lol this guy had some serious issues, looking through the repo this 'testing' master branch or whatever it's called has 7000 different branches
22:46 bitmod please tell me that's not normal
22:46 mostlybadfly joined #git
22:46 mar77i bitmod: git fetch origin testing # I guess?
22:47 mar77i bitmod: well we're not laughing about the other guy downloading said buffoonery.
22:48 sgrover joined #git
22:48 bitmod mar77i: mate if it was up to me i would rewrite this application with a stack that makes sense, but it's legacy code and it's not my job to rewrite it
22:48 kadoban multi_io: At a guess, no. Can you just recreate it as a fork?
22:48 mar77i bitmod: good luck I hope you know when to give up.
22:48 bitmod mar77i: also can't i just "git checkout origin/testing/subbranch"?
22:49 EricPoe joined #git
22:49 jeffreylevesque joined #git
22:49 mar77i *shrug* checkout doesn't usually initiate a network transfer
22:49 sgrover Merging from release back into develop branch.  git checkout develop; git pull origin develop; git merge release-myrelease;  Getting a conflict on a file that is not involved in the release and a diff of that file - we'll call it fileA.txt shows changes from a different file - fileB.txt.
22:49 bitmod and it's giving me a "you are 'detached HEAD' state", what does that mean?
22:50 sgrover is this file corruption kicking in?  Or is this indicative of another issue?
22:51 mar77i bitmod: try to work on a leaf commit. if you're not on a leaf, you'll have to start a new branch - and attach your own side-towers to the mess.
22:52 multi_io kadoban: yeah, I'd probably have to :\
22:52 TomyLobo joined #git
22:52 mar77i I love git. you can always lose your HEAD :)
22:52 mar77i (and grow a new one: git checkout -b my_new_branch_over_here)
22:53 a3Dman joined #git
22:54 ciscobandit joined #git
22:54 bitmod mar77i: can i run "git checkout -b origin/testing/mynewbranch"?
22:55 bitmod normally i just do "git checkout -b some_new_feature", not used to working on anything other than master
22:55 cagedwisdom joined #git
22:55 bitmod sorry, wrong term
22:55 bitmod whatever the parent branch is called
22:55 multi_io kadoban: there are cases where you couldn't do that though, like if you want to send pull requests to multiple upstream forks of the same original project
22:56 multi_io bitmod: if you have a remote branch origin/testing, you can check it out via git checkout testing
22:57 ahr3n joined #git
22:57 multi_io or git checkout -b testing --track origin/testing, which should do the same thing I reckon
22:58 bitmod multi_io: ah ok, but where do the sub branches (or whatever they're called) come into it? e.g. origin/testing/somename
22:58 multi_io just "git checkout origin/testing" will leave you in "detached head" state, because you cannot checkout such a remote branch pointer directly
22:59 bitmod i.e. what's the difference between origin/master and origin/testing?
22:59 watabou_ joined #git
22:59 Solvaring joined #git
22:59 kadoban multi_io: *shrug* can't say I've ever run into that. It might not matter if it's a fork of one of them.
23:01 kerrick joined #git
23:01 multi_io bitmod: I've never seen "origin/testing/somename" (although it's probably possible to create something like that)
23:02 gjash_342 joined #git
23:02 kadoban Git doesn't care what you call branches, they can have / in them for categorizing.
23:02 kadoban It doesn't really mean anything as far as git is concerned. The origin/ part usually means it's a remote tracking branch, but that's only if origin is a remote.
23:03 multi_io remote branches are usually named "remotename/branchname", e.g. origin/master (branch "master" on remote "origin") or "origin/testing" (branch "testing" on remote "origin")
23:03 hahuang65 joined #git
23:03 bitmod multi_io: this guy who was working on this project has (for some bizarre reason) used git as a datastore and has around 8000 origin/* sub branches
23:04 multi_io they basically tell you where that branch pointed to on that remote when you last fetched/pulled from it.
23:04 sgrover joined #git
23:04 orbyt_ joined #git
23:05 thiago joined #git
23:05 bitmod multi_io: ah so is origin/user/some_email_address a commit then?
23:07 multi_io kadoban: can remote names also have / in them for categorizing? In that case origin/foo/bar would be ambiguous (branch bar on remote origin/foo vs. branch foo/bar on remote origin)
23:07 multi_io I think I recently created a remote like that
23:07 kadoban Lots of things can be ambiguous in git. You can create tags with the same names as branches, etc. I don't know if that one is possible.
23:08 kadoban There's usually some way to disambiguate
23:08 Codaraxis joined #git
23:08 nuked0ne joined #git
23:08 bitmod the thing is there is web interface for this application that allows you to create a testing/branch_name branch, and i'm wondering how i can access this branch using git
23:08 jstimm joined #git
23:08 kadoban Same way you access any branch.
23:09 bitmod kadoban: my git knowledge is pretty limited, but normally i access a branch with "git branch my_branch_name", but this origin/testing, origin/master, origin/** stuff is confusing me
23:10 multi_io bitmod: a commit is denoted using a SHA1 checksum only, e.g. 99520f2d204b03761a7591c370ee953808cb2f0b
23:10 bitmod well git checkout soome_branch
23:10 feodoran_ joined #git
23:10 polyrob joined #git
23:10 polyrob joined #git
23:10 kadoban bitmod: That's not really how you access a branch. You should look up remote tracking branch though. It should be in pro git, the book on the website.
23:10 Cabanossi joined #git
23:12 LeBlaaanc joined #git
23:13 bitmod kadoban: ah i just realized 'testing/feature' is actually a branch, i thought 'feature' was a branch and 'testing' was something else
23:13 multi_io bitmod: so this application creates branches named foo/bar/baz (with two / in the name)?
23:13 jameser joined #git
23:14 elsevero_ joined #git
23:15 bitmod multi_io: correct, as well as branches with only one /, e.g. origin/testing/some_branch
23:15 bitmod but i'm new to git so maybe this is normal for large projects?
23:16 kadoban Having branches organized by directory like that, with / in the branch name, is pretty normal.
23:16 thiago what part are you asking?
23:16 thiago using / in a branch name is a project choice
23:16 thiago style
23:17 thiago the / separation between remotename/branchname is something different
23:17 multi_io ok yeah
23:18 multi_io I've always avoided branch names with / in them, but they're possible
23:18 steven joined #git
23:18 bluepixel joined #git
23:18 multi_io and useful
23:18 bitmod is it normal to have 8000 branches in a project?
23:19 kadoban No
23:20 bitmod ok well i'm trying to checkout this branch, origin/preview/mybranch, and i've just cloned the remote repo
23:20 bitmod i don't have to fetch anything do i?
23:20 thinkpad joined #git
23:20 multi_io if you cloned the whole thing, no.
23:21 bitmod multi_io: ok, so this 'detached HEAD' warning, does that mean any changes i make won't be committed?
23:22 multi_io they'll be committed in a new detached HEAD too... anyway, just check out a branch
23:22 akushner___ joined #git
23:22 multi_io_ joined #git
23:23 bitmod multi_io: er i'm not sure i understand. if i make changes and commit them, can i then push them to origin/testing?
23:24 vjacob joined #git
23:24 bitmod ah ok i think i've figured it out
23:24 bitmod thanks for the help!
23:25 menip joined #git
23:27 mattcen joined #git
23:28 multi_io_ left #git
23:28 Murr joined #git
23:28 multi_io joined #git
23:32 renarc joined #git
23:33 ciscobandit joined #git
23:33 renarc joined #git
23:33 ahr3n joined #git
23:35 kerrick joined #git
23:36 renarc joined #git
23:36 renarcii joined #git
23:37 orbyt_ joined #git
23:37 renarc joined #git
23:39 madewokherd joined #git
23:41 Cabanossi joined #git
23:41 renarc joined #git
23:42 renarc joined #git
23:45 polyrob joined #git
23:45 polyrob joined #git
23:47 Kavothe joined #git
23:50 ahr3n joined #git
23:53 himanshijain joined #git
23:53 jorj joined #git
23:54 kadoban_ joined #git
23:55 Mattiaslndstrm joined #git
23:57 Gustavo6046 joined #git
23:59 stutter joined #git

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