Perl 6 - the future is here, just unevenly distributed

IRC log for #git, 2017-02-07

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

All times shown according to UTC.

Time Nick Message
00:00 dmod_ joined #git
00:01 joshszep joined #git
00:04 matoro joined #git
00:05 Gsham joined #git
00:06 solenodic joined #git
00:07 JelmerD joined #git
00:08 acetakwas joined #git
00:10 johnny56_ joined #git
00:12 Guest55 joined #git
00:14 Milos_ joined #git
00:24 cdown joined #git
00:24 spriz joined #git
00:34 cagedwisdom joined #git
00:36 mizu_no_oto joined #git
00:38 SwiftMatt joined #git
00:41 xlegoman joined #git
00:41 sochi left #git
00:44 pur3eval joined #git
00:46 d5sx43 joined #git
00:46 Guest55 joined #git
00:47 mikecmpbll joined #git
00:51 mostlybadfly joined #git
00:52 ocbtec joined #git
00:57 a_thakur joined #git
00:57 ISmithers You know how people sometimes use the term 'pythonic' to describe Python code that really leverages the language features, is there a term like that for Git? To say "This is a very 'git' way of doing things"?
00:57 a_thakur_ joined #git
00:58 e01 joined #git
00:58 d5sx43 joined #git
00:58 Vampire0 ISmithers, is that a synonym for doing it the "right" way? :-D
00:59 ISmithers Basically yeah.
00:59 ISmithers It is a little more than just right, its like right + leveraging the language/tool specifics.
01:00 ISmithers Some people hate it, I just have a friend who is coming from SVN and is reaaaaaally struggling with Git, and I'm somewhat at a loss to explain things to him. He finds the git way of doing things verbose and fiddly.
01:00 Darren__ joined #git
01:01 fuzzybear3965 joined #git
01:02 ojacobson Idiomatic git usage isn't really a thing
01:03 Wolv joined #git
01:03 Wolv left #git
01:03 matoro joined #git
01:03 cdown joined #git
01:06 mischat joined #git
01:06 Vampire0 ISmithers, I'd say he is wrong. The Git way is exteremely powerful and flexible, but you can use it almost like SVN if you want to, you just don't use many of the benefits of Git then
01:07 llamapixel https://www.youtube.com/watch?v=4XpnKHJAok8
01:07 llamapixel so glad I missed SVN and only have git in my head.
01:07 ISmithers He got himself into a mess yesterday using stash where he couldn't pop the stash for some reason. I was suggesting that its easier if he has local changes, to do a WIP commit, pull, then pop your commit back off to have it as unstaged again (at least that is what I do). But he didn't like the idea of committing changes he didn't want to push.
01:07 nedbat llamapixel: ugh, that talk: he spends way too much time mocking svn
01:08 ozmage joined #git
01:09 rwp My observation about people who struggle with git is that they are actually struggling with the underlying idea of patches as changesets.
01:10 joshszep joined #git
01:10 rwp The entire idea of taking current modifications from the working copy and saving them off, then reappying them, is trouble for new people learning it.
01:10 kadoban That sounds a bit weird considering that git is mostly based on snapshots.
01:10 llamapixel There is some great diagramatic sites if you really want to improve your understanding for the visually inclined, however if you don’t have an ongoing test project that you practice with, you won’t improve much.
01:10 rwp It makes perfect senses to *us* but I have seen people just struggle and struggle with that idea.
01:11 rwp I also think that anyone who has dealt with patches through mailing lists and using diff to create patches and 'patch' to apply patches usually gets it easily.
01:12 rwp But people who have never done that just don't have a mental model of how that works. Or even how it can work. This causes them to float at sea.
01:12 ojacobson & yet the only DSCM system I'm aware of that's actually based on patches goes more or less unused (and, at this point, unmaintained, because the author gave up)
01:12 ojacobson (or did, last time I checked)
01:13 rwp Well...  There is quilt.  (ahem)
01:13 ojacobson Not D ;)
01:13 llamapixel I usually deploy a DMZ branch just before master if I have a lot of new developers / artists adding to a repo with game projects
01:13 Vampire0 ISmithers, I also don't like that. It is like using a screwdriver to get a nail in the wall. stash is exactly for this situation and I never had a situation where I couldn`t apply the stashed changes. There may have been conflicts, but those you would have the same with a wip commit
01:13 ojacobson Apparently I'm wrong! Darcs is still actively maintained (last release in sept)
01:14 kadoban Heh, even in the haskell world I don't think anybody uses darcs though.
01:15 red82 joined #git
01:15 ISmithers Vampire0: When I read the docs, and this was some time ago, it was talking about using Stash to shelve changes whilst you swap branches to work elsewhere, and maintain a clean transition branch. Hence why I prefer a WIP commit, especially as then a git pull handles all the merging.
01:15 ojacobson Sure. I _really_ appreciate its approach to changes, though. I'm not convinced it's fully rigorous but it's a huge step forward from "fuck it, whatever diff(1) does"
01:15 nckpz joined #git
01:16 kadoban Hmm, I'll have to read up on that
01:16 Emperor_Earth joined #git
01:16 nowhere_man joined #git
01:17 llamapixel Did anyone get affected by the gitlab meltdown last week ;)?
01:17 Noldorin joined #git
01:18 llamapixel My last freelance assignment had a guy putting down other named cloud git repos and I bet he got stung lol
01:19 ianmethyst joined #git
01:19 llamapixel Is anyone using annex or LFS as well extensively here?
01:20 waroux joined #git
01:21 llamapixel I developed a system that allows me to store binary files in LFS and after 4 commits you start to save on space, essentially bin 2 hex and back again.
01:21 Vampire0 ISmithers, stash does the same. If you do stash pop or stash apply it applies your changes and leaves them in the same conflicted state as a merge or rebase does
01:21 Vampire0 ISmithers, you then can use git mergetool to resolve the conflicts
01:22 Gsham joined #git
01:22 Vampire0 ISmithers, the only thing is, if there was a conflict to prevent accidental loss a stash pop worked like a stash apply and you have to manually discard the stash with stash drop
01:22 acetakwas joined #git
01:23 Vampire0 ISmithers, and the docs do say nothing about when you use stash, besides that you temporarily undo the changes in worktree and index to reapply them later to the same or a different commit, but not that you only use it when switching changes
01:24 cspotcode joined #git
01:25 waroux left #git
01:25 Iacobus__ joined #git
01:26 sgtbigman joined #git
01:29 F0rTh3J3st joined #git
01:30 plos joined #git
01:32 ISmithers I think it was this I was reading https://git-scm.com/book/en/v1/Git-Tools-Stashing which is the Git book. It talks about a few use-cases, but this was some time ago I read it.
01:33 ISmithers They are more suggestions as opposed to prescriptions though granted.
01:34 seni joined #git
01:35 annoymouse joined #git
01:35 Vampire0 ISmithers, And I was referring to the official subcommand description / help at https://git-scm.com/docs/git-stash
01:35 ISmithers Yeah I realized that when I looked it up :)
01:36 Vampire0 ISmithers, this even lists the stash pull unstash as first example
01:36 Gsham joined #git
01:36 Vampire0 ISmithers, and even git merge (and thus pull) suggests to stash or commit if I remember correctly
01:36 ISmithers Oh I'm not saying its wrong or impossible, I just came from a studio where it seemed like doing the temp WIP commit was 'the way'. Githonic, if you will hehe.
01:38 mvensky joined #git
01:39 ISmithers I think another issue is that he insists on using a UI and I use Git bash, and I find it really hard to figure out how the particular flavor of UI he is using has wrapped Git functionality. That being said, we have settled on using Git Kraken for this project (its just us 2) so at least we have some common UI there.
01:40 wiesel joined #git
01:40 rgrinberg joined #git
01:41 andream joined #git
01:41 llamapixel clients are always troublesome against raw terminal without any additional addons.
01:42 cqi joined #git
01:42 k_sze[work] joined #git
01:42 llamapixel git cig and a bash log for colour is about all I would want to add to the mix.
01:42 ISmithers Yeah I did need to use some sort of git log pretty thing IIRC, as the default red on black was like - I could not see what was going on.
01:43 fuzzybear joined #git
01:44 k_sze[work] Is there an elegant way to diff my dirty working copy against a stash without first committing or stashing and then undoing?
01:45 kadoban git diff stash@{0} or something
01:45 Gsham joined #git
01:46 ISmithers I think yesterday he got into a state where he stashed some changes, and then Visual Studio modified the project file, and he had unstaged changes again. In that instance the stash would fail to apply is that right?
01:46 k_sze[work] kadoban: what if I want to reverse the diff?
01:47 ojacobson ISmithers: "maybe."
01:47 k_sze[work] kadoban: I mean, `git diff stash@{0}` shows the changes in my stash as the '-', and the stuff in my working copy as '+'.
01:47 k_sze[work] I want to invert that.
01:47 ojacobson From there, it's possible that git stash apply/git stash pop would successfully merge the stashed changes with the workspace changes
01:47 ISmithers He was getting an error that said something like failed to apply due to unstaged changes (or something long those lines).
01:47 ojacobson also possible it wouldn't
01:47 ISmithers OK
01:47 ojacobson depends on the nature of the two changes
01:47 ojacobson if they conflict in the merge sense, then git stash apply will abort.
01:48 ISmithers The solution in that case would be to stash those unstaged changes as well? Can you 'stash amend' or similar?
01:48 JelmerD joined #git
01:48 ojacobson The solution would be to either manually combine the changes or to pick one side or the other
01:48 kadoban k_sze[work]: Dunno, a trip through the manpage may yield something.
01:48 ojacobson "manually" may include using an external merge tool
01:49 ISmithers (We have BC4 setup with Git)
01:51 ResidentBiscuit joined #git
01:52 jameser joined #git
01:52 gugah joined #git
01:52 xaviergmail joined #git
01:52 andream joined #git
01:54 marvi joined #git
01:55 shinnya joined #git
01:55 andream joined #git
01:56 Gsham joined #git
01:58 e01 joined #git
01:59 Sasazuka joined #git
02:01 ShalokShalom hi there
02:01 ShalokShalom i am used to Github
02:01 ShalokShalom where can i find a tarball here?
02:01 ShalokShalom https://code.qt.io/cgit/pyside/shiboken.git/log/
02:01 andream joined #git
02:02 kadoban ShalokShalom: You can try 'git archive' if the web interface doesn't expose tarballs.
02:02 ShalokShalom i wanted to use it as source= in my PKGBUILD
02:02 ShalokShalom anyway, so there is no tarball?
02:02 ShalokShalom so what to do?
02:03 ShalokShalom a version control system without tarballs?
02:03 ShalokShalom can i download the full tree and create one by my own?
02:03 * kadoban wanders off
02:04 red82 joined #git
02:06 Vampire0 ShalokShalom, didn't kadoban answer your question already?
02:06 Vampire0 ShalokShalom, man git archive
02:06 gitinfo ShalokShalom: the git-archive manpage is available at http://jk.gs/git-archive.html
02:06 ShalokShalom he did
02:07 ShalokShalom i ask for something new
02:07 ShalokShalom thanks for the link
02:08 mizu_no_oto joined #git
02:09 Vampire0 ShalokShalom, btw. as far as I can see in cgits feature list it says "on-the-fly archives for tags and commits", so I'd assume you can download a tarball
02:10 Nilesh_ joined #git
02:10 ShalokShalom Ah, i misread kadobans text as "You can`t try..."
02:10 ShalokShalom sleepy :P thanks a lot great help
02:11 ShalokShalom and beta english skills :)
02:11 Vampire0 ShalokShalom, ah, yes, cgit *does* support tarballs, I've just seen it. I guess the page that you want just disabled it, as it is not there
02:11 ShalokShalom ah, ok i see
02:12 ShalokShalom they switch from Github to 'this'
02:12 Gsham joined #git
02:12 pulec joined #git
02:12 Vampire0 ShalokShalom, if you look at https://git.zx2c4.com/cgit/tag/?h=v1.1 or https://git.zx2c4.com/cgit/commit/ you see there are tarball links
02:12 PtxDK joined #git
02:13 Vampire0 maybe you need to enable it, maybe they did disable it, I don't know
02:13 ShalokShalom solved, thanks
02:13 Levex joined #git
02:13 XenophonF joined #git
02:13 Klumben joined #git
02:14 gitinfo ShalokShalom: This channel tracks karma based on who has gotten lots of thanks for being helpful. If you want to help someone reach karmic nirvana, please mention their name when thanking them with "thank you", "thankyou", "thanks", "thx" or "cheers". Try ".karma <nick>" or ".topkarma" to show karma status of a person.
02:14 alcohol joined #git
02:14 Vampire0 yw ShalokShalom
02:14 ShalokShalom ^^
02:14 phroa since when is karma a thing
02:15 bremner phroa: since Buddha??
02:15 ShalokShalom thx Vampire0
02:15 phroa oh, how unenlightened of me
02:15 bremner or eternity, Iguess
02:15 ShalokShalom thx kadoban
02:15 Roconda joined #git
02:16 ShalokShalom how knows gitinfo when to put this message?
02:19 Vampire0 !beer
02:19 gitinfo Beer! It's what's for dinner!
02:19 solenodic joined #git
02:21 Aleric joined #git
02:21 finalbeta joined #git
02:22 ISmithers Slack is so nice to me sometimes :> https://db.tt/WZKKv00SI9
02:22 bremner creepy
02:23 dreiss joined #git
02:24 johnny56_ joined #git
02:25 Gsham joined #git
02:27 druonysus_ joined #git
02:27 wilbert joined #git
02:29 fstd joined #git
02:29 watabou joined #git
02:30 pur3eval joined #git
02:33 LeMike joined #git
02:34 PtxDK_ joined #git
02:35 ISmithers Hehe, it has all these random sayings when you login.
02:35 arescorpio joined #git
02:37 Vampire0_ joined #git
02:39 brent__ joined #git
02:40 pulec joined #git
02:41 _xor joined #git
02:46 aavrug joined #git
02:46 lindenle joined #git
02:46 aavrug joined #git
02:47 ianmethyst joined #git
02:47 _28_ria joined #git
02:48 matoro joined #git
02:50 jimi_ joined #git
02:50 ankit01ojha joined #git
02:51 kasual joined #git
02:53 lindenle joined #git
02:59 e01 joined #git
03:00 mizu_no_oto joined #git
03:00 NeverDie joined #git
03:00 dermoth joined #git
03:00 pks joined #git
03:01 d^sh joined #git
03:02 austin987 joined #git
03:03 fstd joined #git
03:09 mischat joined #git
03:12 Gsham joined #git
03:15 duderonomy joined #git
03:18 Vortex34 joined #git
03:20 red82 joined #git
03:22 phenix joined #git
03:22 FabTG joined #git
03:25 mizu_no_oto joined #git
03:25 govg joined #git
03:27 mizu_no_oto joined #git
03:29 Goplat joined #git
03:30 druonysus__ joined #git
03:32 gfixler joined #git
03:34 JelmerD joined #git
03:35 Vampire0_ joined #git
03:37 SwiftMatt joined #git
03:39 red82 joined #git
03:40 Iacobus__ joined #git
03:40 modin joined #git
03:40 zezba joined #git
03:40 linduxed joined #git
03:41 Derperperd joined #git
03:42 sunri5e joined #git
03:44 a_thakur joined #git
03:46 Neo joined #git
03:50 a_thakur joined #git
03:51 hexagoxel joined #git
03:52 zacts joined #git
03:53 Vampire0 joined #git
03:56 Gsham joined #git
03:56 mizu_no_oto joined #git
03:59 wilbert joined #git
04:00 e01 joined #git
04:00 e01 joined #git
04:01 pur3eval joined #git
04:02 chachasmooth joined #git
04:03 PioneerAxon joined #git
04:06 zezba joined #git
04:09 mischat joined #git
04:11 kent\n joined #git
04:15 Gsham joined #git
04:15 navidr joined #git
04:16 ozmage_ joined #git
04:16 ozmage_ joined #git
04:25 red82 joined #git
04:26 blackwind_123 joined #git
04:28 mehola joined #git
04:31 lindenle joined #git
04:32 govg joined #git
04:32 SporkWitch joined #git
04:33 rivarun joined #git
04:34 dsdeiz joined #git
04:34 Saturn812 joined #git
04:35 zezba joined #git
04:36 SporkWitch so i have a production and a development branch.  I did a squashed merge from devel into production, which went fine: shows as a single commit, with a nice, clean change summary.  Now devel shows as being out of sync with production, which is to be expected because of the squash.  When I try to rebase devel onto production, though, it seems to be trying to merge older versions of files, when it
04:36 SporkWitch should really just rebase automatically; the branches are identical to each other, except for commit messages.
04:36 acln joined #git
04:37 SporkWitch i could just nuke the devel branch and create a new one from the current production branch, but that seems a bit hamfisted.
04:38 ozmage joined #git
04:38 nckpz joined #git
04:40 UniFreak joined #git
04:41 Cabanoss- joined #git
04:42 ojacobson A squash merge isn't a merge
04:42 ojacobson so git can't reason about it like a merge
04:43 ojacobson (it's an entirely novel commit that happens to have a similar-looking diff, produced by applying a sequence of diffs and committing the result. Git records nothing whatsoever about its relationship with the branch being merged.)
04:43 diogenese joined #git
04:45 SporkWitch ojacobson: i get that, i'm more trying to figure out the unusual behaviour being exhibited.  The current version in each branch is identical, yet the results of the rebase have me comparing the current version (which is the same on both) with the PREVIOUS version, not the current of either branch.  THat's the weirdness I'm not getting.
04:45 ojacobson How far back is the last commit the production and development branches have in common? `git rebase` on the development branch is trying to reapply every single one of those diffs (other than merges) as a new commit
04:46 ojacobson even if those diffs no longer make sense because the new base is too different from the current base
04:46 SporkWitch ah, ok
04:46 SporkWitch i get you now
04:47 ojacobson A rebase is (probably) the wrong tool, is I guess my point - but the "right" tool depends on the history you're trying to create :)
04:48 TechnoTony joined #git
04:48 andream joined #git
04:48 SporkWitch yeah, one solution i found suggested git branch -f, but that wipes out the history, when all i want is to tell it that devel IS in fact current with production, heh
04:49 ojacobson You could make a "merge" from production to staging whose result is exactly the state of production
04:49 justanotheruser joined #git
04:49 SporkWitch as far as the history i'm trying to create: i'd like to maintain the detailed commit history on the devel branch, while keeping the commit history on production clean.  Dunno if that's necessarily sane.
04:49 ojacobson That would accurately describe the relationship between the two branches (as far as is possible) so that Git can do further merges from there
04:49 ojacobson in the longer term I would spend some serious time with !workflow - any process that involves rebasing long-lived branches is a recipe for trouble
04:49 gitinfo Finding the right workflow for you is critical for the success of any SCM project.  Git is very flexible with respect to workflow.  See http://sethrobertson.github.com/GitBestPractices/#workflow for a list of references about choosing branching and distributed workflows.
04:50 ojacobson so designing a branch pattern that accomplishes what you want without rebases may be useful
04:50 SporkWitch that's rather the purpose here: playing around with workflows trying to figure out something sane :)
04:51 SporkWitch the repo in question is just a website for a class with a largely useless prof (her "workflow" consists of a flash drive and non-secure FTP using the same password and account info as your university email, billing, etc...)
04:51 SporkWitch so i just set up a simple two-branch config, with a hook on the server to automatically deploy when i push on the production branch.
04:52 freecoder hi, i cloned a repo from a local repo using "git clone -ns". initially .git/objects on cloned repo was empty. then i added some files to it and ran "git add *". now "git status -s" shows those files marked as "A" but .git/objects is still empty
04:52 freecoder how is this possible?
04:52 ojacobson freecoder: have a look at .git/objects/info/alternates
04:53 ojacobson The two repos have a single object store (physically kept with the parent repo)
04:53 ojacobson If you move or delete the parent repo, the child repo will break badly
04:53 inflames joined #git
04:53 freecoder oh, so the objects for cloned repo are stored in the path pointed to by .git/objects/info/alternates?
04:53 ojacobson Sometimes, yes.
04:54 SporkWitch IIRC this is unique to cloning things on the local machine; no such issue if it's using ssh/http/etc.
04:54 ojacobson SporkWitch: it's unique to cloning with -s
04:55 zezba joined #git
04:55 SporkWitch ah, thought it was the default behaviour when cloning something on the local filesystem, to reduce duplications; was that changed at some point, or am i just misremembering?
04:55 ojacobson That's a whole separate thing -
04:55 ojacobson Cloning from a directory in the same filesystem uses hard links to share the objects, instead of git's alternates system
04:55 ojacobson hard links don't have the "renaming the other repo will break them" issue
04:56 ojacobson (however, repo corruption or a rogue write on the other repo will break them both)
04:56 ojacobson Alternates work across FS boundaries, though, so they have that going for them
04:57 SporkWitch gotcha
04:57 JanC_ joined #git
04:59 ayogi joined #git
05:01 robotroll joined #git
05:01 e01 joined #git
05:01 e01 joined #git
05:02 pur3eval joined #git
05:05 solenodic joined #git
05:05 red82 joined #git
05:05 roelmonnens joined #git
05:10 mischat joined #git
05:16 m4sk1n joined #git
05:20 __builtin joined #git
05:23 CussBot joined #git
05:24 _28_ria joined #git
05:26 aidalgol joined #git
05:27 Shakeel_ joined #git
05:29 Guest14970 joined #git
05:30 Rotaerk joined #git
05:32 Rotaerk is there some way to create a folder outside of my git repo containing the HEAD snapshot (but no .git folder)
05:32 Rotaerk git archive seems to create an archive file containing this, but I just want a folder
05:32 grawity Rotaerk: see !deploy
05:32 gitinfo Rotaerk: Git is not a deployment tool, but you can build one around it (in simple environments) or use it as an object store(for complex ones). Here are some options/ideas to get you started: http://gitolite.com/deploy.html
05:34 sbulage joined #git
05:35 Rotaerk hmm don't think my situation requires all those rules; I don't want to deploy to an existing folder, but to create a new folder
05:35 _ikke_ Rotaerk: just untar the tar immediately
05:35 _ikke_ tar is just a streaming format
05:36 Rotaerk k, thanks
05:36 _ikke_ Rotaerk: git archive --format=tar HEAD --prefix=directory/ | tar xf -
05:36 chachasmooth joined #git
05:36 Rotaerk nice, thanks again
05:37 Shakeel_ joined #git
05:41 rewt`` joined #git
05:42 CussBot_ joined #git
05:42 plos joined #git
05:43 PioneerAxon joined #git
05:55 freecoder if i use "git clone -s" and then do "git repack -df" in the cloned repo, shouldnt this store the packfiles in cloned repo?
05:55 freecoder according to https://git-scm.com/docs/git-clone#git-clone---shared
06:01 rchavik joined #git
06:02 e01 joined #git
06:02 e01 joined #git
06:02 pks joined #git
06:02 pur3eval joined #git
06:02 Vampire0_ joined #git
06:03 zerorax joined #git
06:05 zerorax Hi, there is a project on github, a plugin for the Atom code editor... The guy who made it hasn't been active since october. Is there anything I can do to contact this person? They have 2 critical bugs and I fixed them... Not sure what to do
06:06 rscata joined #git
06:07 mohabaks joined #git
06:07 zerorax should I just fork the project and let the Atom people know?
06:07 thiago joined #git
06:08 _ikke_ zerorax: This is more of a #github question (I wouldn't know how, if even possible)
06:08 _ikke_ You can always fork it
06:08 zerorax oh sorry, i'll ask in github
06:08 xall joined #git
06:09 ToBeCloud joined #git
06:09 MineCoins joined #git
06:11 albel727 joined #git
06:12 mischat joined #git
06:12 Hudu joined #git
06:13 dec0n joined #git
06:13 mweshi joined #git
06:20 Mikerhinos joined #git
06:21 ngm joined #git
06:25 bocaneri joined #git
06:28 a3Dman joined #git
06:28 freimatz joined #git
06:28 lektrik joined #git
06:30 Skitz0 joined #git
06:32 rivarun joined #git
06:37 mingrammer joined #git
06:37 Skitz0 joined #git
06:40 hahuang61 joined #git
06:41 b1tchcakes joined #git
06:42 red82 joined #git
06:42 kaldoran joined #git
06:43 Drzacek joined #git
06:43 Drzacek left #git
06:44 Mikerhinos joined #git
06:47 Cogitabundus joined #git
06:48 brent__ joined #git
06:49 JelmerD joined #git
06:49 overlord_tm joined #git
06:50 zeroed joined #git
06:51 ThomasLocke joined #git
06:52 redhedded1 joined #git
06:55 roelmonnens joined #git
06:55 seni joined #git
06:57 solenodic joined #git
06:57 freimatz joined #git
06:58 JeroenT joined #git
06:58 cyphase joined #git
06:58 freimatz_ joined #git
07:00 MattMaker joined #git
07:00 Raging_Hog joined #git
07:01 ThomasLocke_ joined #git
07:01 ThomasLocke_ joined #git
07:03 e01 joined #git
07:03 e01 joined #git
07:03 JeroenT joined #git
07:05 seni joined #git
07:08 nemesit|znc joined #git
07:09 chele joined #git
07:09 Skitz0 joined #git
07:10 Cogitabundus joined #git
07:10 nemesit|znc joined #git
07:11 nemesit|znc joined #git
07:12 mweshi_ joined #git
07:14 zeroed joined #git
07:15 zeroed joined #git
07:15 _ng joined #git
07:17 JeroenT joined #git
07:23 JeroenT joined #git
07:27 ihradek joined #git
07:28 roelmonnens joined #git
07:29 roelmonnens joined #git
07:30 Vampire0 joined #git
07:31 mingrammer joined #git
07:34 mischat joined #git
07:34 c0c0 joined #git
07:36 byte512 joined #git
07:36 psyb0t joined #git
07:36 jnavila joined #git
07:36 elect_ joined #git
07:38 qt-x joined #git
07:38 matoro joined #git
07:39 psyb0t_ joined #git
07:42 acetakwas joined #git
07:45 Snugglebash joined #git
07:45 troulouliou_div2 joined #git
07:46 lordjancso joined #git
07:47 blackwind_123 joined #git
07:47 zeroed joined #git
07:48 a_thakur joined #git
07:49 vrlx joined #git
07:50 vrlx left #git
07:51 pur3eval joined #git
07:52 theoceaniscool joined #git
07:53 dersand joined #git
07:53 mathematic-alpha joined #git
07:54 spacelord_ joined #git
07:55 _28_ria joined #git
07:57 math-alpha joined #git
08:02 jagob joined #git
08:03 freimatz_ joined #git
08:08 absence left #git
08:10 synchronous joined #git
08:11 madduck so I have a rev-list command that gives me a set of commits and I'd like to format-patch each one of them. Is there a way to do so as a series? Or must I shell-while-loop?
08:12 thierryp joined #git
08:12 svm_invictvs joined #git
08:13 osse madduck: git format-patch accepts a range, so you can try that
08:14 madduck the problem is that it's not a range. I am using rev-list to filter the list of commits by path
08:14 madduck so it's the sequence of commits touching a file ./foo, which is non-contiguous
08:15 osse | xargs -n1 git format-patch -1
08:15 Ruzzy joined #git
08:15 Ruzzy I just did a chmod in my working tree
08:15 Ruzzy and git status won't reflect it
08:15 madduck this will yield 3 dozens of files all numbered 0001 and thus not in order…
08:16 madduck osse: ^
08:16 troulouliou_div2 joined #git
08:17 osse madduck: while read hash ; do git format-patch --start-number=$(++i)) -1 "$hash"; done
08:17 osse Ruzzy: what does git config core.fileMode say ?
08:17 osse madduck: oops, I mean $(( ))
08:18 Ruzzy osse: nothing
08:19 Ruzzy but my git always used to report these changes even when I didn't care about them, now I did a chmod on purpose and nothing is shown in status
08:19 jast what kind of chmod did you do?
08:19 Ruzzy 755
08:19 Ruzzy they were 644s before
08:20 King_Hual joined #git
08:20 Ruzzy I think...
08:20 Ruzzy or they used to be similar to 644
08:20 Ruzzy in any case, they are now all 755 as desired.
08:20 JeroenT joined #git
08:20 reallyfu joined #git
08:21 jast let's check what Git thinks it was before... git ls-files --stage | grep filename
08:21 Ruzzy ls-files?
08:22 Ruzzy wait nvm
08:22 Ruzzy okay, no output
08:22 jast you did grep for one of the actual names, I hope? :)
08:22 Ruzzy hang on. some files have an output
08:23 Ruzzy others do not.
08:23 Ruzzy jast: yeah, after writing "filename" like an idiot
08:23 borkr joined #git
08:23 Ruzzy but still, mixed results. the files that were already 755 are showing as such. the files that were changed to 755 have no output
08:24 jast that would indicates Git isn't tracking them at all
08:24 thierryp joined #git
08:24 Ruzzy oh no. what if I left an outdated .gitignore somewhere...
08:24 jast actually try again with 'git ls-files -c -m -s', I can never remember which combinations of flags do what exactly. it may have been skipping some of the files.
08:24 Ruzzy give me a momen :3
08:24 Ruzzy moment*
08:28 solenodic joined #git
08:28 Ruzzy yup it was an old gitignore
08:28 Ruzzy got rid of it and tracked a tonne of new files. all is right with the world. thanks for your help.
08:30 mikecmpbll joined #git
08:30 jast great. I like straightforward solutions. :)
08:31 Junior joined #git
08:31 madduck osse: heh. that works. thanks. ;)
08:31 zeroed joined #git
08:31 zeroed joined #git
08:39 reallyfu Am I the only one suffering from the fact that git isn't able to resolve rename/rename merge pseudo-conflicts? The second time now and it's going to eat away my working time. It screws up my repo and it's not even a real problem.
08:40 ocbtec joined #git
08:42 Alienpruts joined #git
08:42 Shashikant86 joined #git
08:42 Balliad joined #git
08:45 prxgnb joined #git
08:46 mozzarella joined #git
08:47 anyonebutme joined #git
08:50 hahuang61 joined #git
08:50 mathematic-alpha joined #git
08:50 solenodic joined #git
08:50 math-alpha joined #git
08:51 Shashikant86 joined #git
08:55 warthog9 joined #git
09:03 rorro joined #git
09:04 tobiasvl reallyfu: the second time with the same file? if so, man git rerere
09:04 gitinfo reallyfu: the git-rerere manpage is available at http://jk.gs/git-rerere.html
09:04 tobiasvl or else what is the problem
09:04 brent__ joined #git
09:05 reallyfu tobiasvl: I don't know actually. It's been a while. I'm working with Joomla and they use to put index.html files with the same content everywhere and git appears to not be able to handle that on different branches when I add/delete/copy these files arbitrarily.
09:06 seni joined #git
09:06 mikecmpbll joined #git
09:06 acetakwas joined #git
09:08 reallyfu I'd rather not automate the resolution. Don't want to screw up even more. I'll probably save me a workflow now somewhere. After all, it's retry merge with high verbosity, find conflicting commit, rebase to somewhere suitable, apply magic.
09:08 jast with 'git merge' you can disable rename detection: git merge -Xno-renames
09:08 Ardethian joined #git
09:10 overlord_tm joined #git
09:12 zeroed joined #git
09:12 zeroed joined #git
09:14 anyonebutme left #git
09:14 bongjovi joined #git
09:15 marcogmonteiro joined #git
09:16 mingrammer joined #git
09:18 chele can i stash just some files from unstaged changes?
09:18 jast you can stash individual changes: git stash save -p
09:19 ac3takwas joined #git
09:19 reallyfu jast: Thank you very much for this tip. I think you saved me some time there. I hope I won't have to use that always now, though.
09:19 jast reallyfu: fingers crossed :)
09:22 chele jast thanks
09:23 solrize joined #git
09:23 solrize joined #git
09:24 Random832 joined #git
09:27 kurkale6ka joined #git
09:30 Levex joined #git
09:30 Hudu joined #git
09:32 lyxus joined #git
09:33 _noblegas joined #git
09:33 Mikerhinos joined #git
09:34 xall joined #git
09:38 Mikerhinos joined #git
09:39 Ardethian\ joined #git
09:39 salamanderrake joined #git
09:40 pur3eval joined #git
09:42 digidog joined #git
09:43 ferr1 joined #git
09:45 Shashikant86 joined #git
09:47 cyphase joined #git
09:48 enkrypt joined #git
09:49 Vgr_ joined #git
09:49 Vgr_ joined #git
09:50 Vgr_ joined #git
09:50 Vgr_ joined #git
09:54 Serpent7776 joined #git
09:55 Snugglebash joined #git
09:56 basiclaser joined #git
09:56 joshszep joined #git
09:56 miczac joined #git
09:57 Shashikant86 joined #git
09:58 Guest9 joined #git
09:58 Guest9 joined #git
10:00 chll_ joined #git
10:01 overlord_tm joined #git
10:02 thierryp joined #git
10:02 pijiu joined #git
10:03 Shashikant86 joined #git
10:04 Anticom joined #git
10:06 TomyLobo joined #git
10:07 delboy1978uk joined #git
10:08 delboy1978uk my work use PHP's composer (like maven on java) all wrong, and are committing vendor code into our git repo. It's a pain in the ass! Anyway, I'm trying to add it, but it whines about submodules. So I deleted the .git in the vendor package folder. But I still get fatal pathspec submodule error, can anyone assist?
10:10 Darcidride joined #git
10:10 jast what's the error message?
10:10 Ardethian joined #git
10:11 jast and the command you tried
10:11 jimi_ joined #git
10:11 delboy1978uk git add vendor/blah/*
10:12 delboy1978uk fatal: Pathspec vendor/blah/some-file is in submodule vendor/blah
10:12 lss8 joined #git
10:12 lss8 joined #git
10:13 delboy1978uk jast ^
10:13 ddd_ joined #git
10:13 Darcidride joined #git
10:16 jimsio joined #git
10:16 howitdo joined #git
10:17 Shashikant86 joined #git
10:17 dsdeiz joined #git
10:17 dsdeiz joined #git
10:18 zerorax how do I revert to previous commit states, and is there a way to bookmark commit states with git, or should i make a script to keep track of important commits?
10:18 tobiasvl delboy1978uk: did you run git submodule deinit ?
10:19 tobiasvl just removing .git from the submodule isn't enough
10:19 Juzzika joined #git
10:19 jbitdrop joined #git
10:19 Mikerhinos joined #git
10:19 zerorax like, I commit every time I make any progress at all, but there are times when i've reached a point where i want to make a more notable copy, so I archive the project to save that state
10:20 tobiasvl zerorax: !revert
10:20 gitinfo zerorax: That's a rather ambiguous question... options: a) make a commit that "undoes" the effects of an earlier commit [man git-revert]; b) discard uncommitted changes in the working tree [git reset --hard]; c) undo committing [git reset --soft HEAD^]; d) restore staged versions of files [git checkout -p]; e) move the current branch to a different point(possibly losing commits)[git reset --hard $COMMIT]?
10:20 pijiu2 joined #git
10:20 tobiasvl zerorax: as for bookmarks, man git tag
10:20 gitinfo zerorax: the git-tag manpage is available at http://jk.gs/git-tag.html
10:20 nickabbey joined #git
10:20 jast delboy1978uk: does this help: !submodule_rm
10:20 gitinfo delboy1978uk: [!submodules_rm] To delete a submodule, do this in the superproject: Edit/delete .gitmodules to remove the submodule, and `git add .gitmodules`. Then `rm -rf submodulepath; git rm -f --cached submodulepath; git commit -am "Removed submodules!"`  Inspect .git/config for "submodule" entries to remove.  Inspect .git/modules for caches to remove.  Possible alternatives to submodules: "!gitslave" or "!subtree"
10:20 tobiasvl zerorax: depends what you need the "bookmarks" for though. I assume you don't just want a branch
10:21 zerorax thank you all
10:21 zerorax appreciate the answers
10:22 delboy1978uk jast: thanks, but I DON'T have a .gitmodules!!!
10:22 moritz good for you!
10:22 aard_ joined #git
10:23 Shashikant86 joined #git
10:24 tobiasvl delboy1978uk: assumedly you need to remove an entry in .git/config (either manually or with "git submodule deinit" like I suggested)
10:26 tvw joined #git
10:27 Joost joined #git
10:32 janx joined #git
10:32 irqq joined #git
10:33 afuentes joined #git
10:35 achadwick joined #git
10:35 DieguezZ joined #git
10:36 jast delboy1978uk: ignore that step, then
10:36 janx joined #git
10:37 Shashikant86 joined #git
10:38 fuchstronaut What is a good strategy of merging into the master? Our lead dev seems to cherry-pick (or somehow copy) commits ontop of master. So changes exist twice, once in the branch and once in master. This seems odd, doesn't it?
10:38 zefferno joined #git
10:39 delboy1978uk sorted, thanks for your help jast
10:39 tobiasvl fuchstronaut: well, the best strategy is to use git merge
10:39 tobiasvl fuchstronaut: a bit weird to cherry-pick into master, I'd say, but there's probably a reason they use that strategy?
10:42 ShapeShifter499 joined #git
10:44 thierryp joined #git
10:44 manuelschneid3r joined #git
10:47 SwiftMatt joined #git
10:49 MineCoins joined #git
10:52 hahuang61 joined #git
10:53 ShapeShifter499 joined #git
10:54 muldover joined #git
10:56 Snugglebash joined #git
11:00 navidr joined #git
11:00 wrouesnel1 joined #git
11:02 cdown joined #git
11:02 jnavila joined #git
11:04 adamru joined #git
11:06 vuoto joined #git
11:08 YuGiOhJCJ joined #git
11:13 plos joined #git
11:14 markovh joined #git
11:14 rivarun joined #git
11:17 dimi1947 joined #git
11:20 Anticom joined #git
11:21 brent__ joined #git
11:21 nettoweb joined #git
11:21 rorro joined #git
11:24 mathematic-alpha joined #git
11:25 _ngz_ngzz joined #git
11:25 seni joined #git
11:28 Shashikant86 joined #git
11:28 fahadash joined #git
11:29 pur3eval joined #git
11:31 Tobbi joined #git
11:32 ozkavosh joined #git
11:32 jost_ joined #git
11:32 re1 joined #git
11:38 djb-irc joined #git
11:41 ToBeCloud joined #git
11:43 delboy1978uk hi guys, i'm trying to rebase 900 commits, but i constantly get conflicts. the remote version appears to be the new correct version, whereas the local version is some ancient half coded commit. i accept the server version, only to get the same file conflicting whilst rebasing the next commit. am i doing something wrong here?
11:43 JeroenT joined #git
11:43 delboy1978uk should i be accepting the local?
11:47 BackEndCoder joined #git
11:48 ismithers joined #git
11:48 tobiasvl delboy1978uk: man git rerere ?
11:48 gitinfo delboy1978uk: the git-rerere manpage is available at http://jk.gs/git-rerere.html
11:51 mischat_ joined #git
11:51 nobitano_ joined #git
11:52 uranellus joined #git
11:52 uranellus joined #git
11:53 ertesx joined #git
11:54 theoceaniscool joined #git
11:56 UniFreak joined #git
11:57 Hudu joined #git
11:57 JeroenT joined #git
11:59 dsdeiz joined #git
12:02 robattila256 joined #git
12:03 reallyfu I popped a stash entry and it gave me conflicts, some of which are “deleted by us”. I tried to git rm them, but that process just doesn't finish. It's just three files that need to be removed.
12:03 mischat joined #git
12:04 reallyfu Oh, I need to add them?!
12:05 mgver guys, whitch best alternative to github to private repo?
12:06 JeroenT_ joined #git
12:06 roelmonn_ joined #git
12:07 reallyfu Can't give you a best, but I think bitbucket is free for up to five users with a private repo, mgver.
12:07 _ikke_ either gitlab or bitbucket
12:08 mgver reallyfu: and gitlab? is free to unlimited users?
12:09 reallyfu I don't know that one, sorry.
12:11 a3Dman joined #git
12:11 sgfgdf joined #git
12:11 DaveTaboola joined #git
12:11 mgver reallyfu: no problem, thanks
12:12 tobiasvl mgver: looks like it https://about.gitlab.com/products/#compare-options
12:13 sgfgdf hello, guys! anyone can share a good git workflow which is not so complicated as gitflow is? is anybody familiar with this http://www.bitsnbites.eu/a-stable-mainline-branching-model-for-git/. what do you think?
12:13 _ngz_ngzz left #git
12:13 _ng joined #git
12:16 hobodave joined #git
12:17 mohabaks joined #git
12:18 UrsoBranco joined #git
12:19 rorro joined #git
12:20 Vampire0 sgfgdf, !workf
12:20 gitinfo sgfgdf: [!workflow] Finding the right workflow for you is critical for the success of any SCM project.  Git is very flexible with respect to workflow.  See http://sethrobertson.github.com/GitBestPractices/#workflow for a list of references about choosing branching and distributed workflows.
12:21 nettoweb joined #git
12:22 pietv joined #git
12:23 cdown joined #git
12:23 nettoweb1 joined #git
12:25 seni joined #git
12:26 fuchstronaut tobiasvl: i don't really know, maybe I should just ask.
12:27 nettoweb joined #git
12:29 Cavallari joined #git
12:30 jimi_ joined #git
12:30 JeroenT joined #git
12:31 inflames joined #git
12:32 osse https://arstechnica.com/information-technology/2017/02/microsoft-hosts-the-windows-source-in-a-monstrous-300gb-git-repository
12:34 _main_ joined #git
12:42 jameser joined #git
12:43 daivyk joined #git
12:44 jast Vampire0: I think that qualifies you as a candidate for most efficient user of triggers
12:45 mmlb joined #git
12:45 jast Vampire0: however, !rkf is even shorter
12:45 gitinfo Vampire0: [!workflow] Finding the right workflow for you is critical for the success of any SCM project.  Git is very flexible with respect to workflow.  See http://sethrobertson.github.com/GitBestPractices/#workflow for a list of references about choosing branching and distributed workflows.
12:45 delboy1978uk left #git
12:46 redeemed joined #git
12:48 kpease joined #git
12:52 _Cyclone_ joined #git
12:52 hahuang61 joined #git
12:56 overlord_tm joined #git
12:56 Snugglebash joined #git
12:57 mattcen joined #git
12:57 m4sk1n joined #git
12:59 ojdo joined #git
13:03 reallyfu git checkout ftp://bill:linusisapenguin@git.microsoft.com/win11
13:04 synthroid joined #git
13:04 ash_workz joined #git
13:05 synthroi_ joined #git
13:05 Mikerhinos joined #git
13:07 ruxu joined #git
13:08 Levex joined #git
13:10 nowhere_man joined #git
13:10 Sick joined #git
13:11 garethdaine joined #git
13:11 kfc_ left #git
13:11 averell joined #git
13:11 Vampire0 jast, I should write a macro that shortens each trigger I write to the shortest possible variant. :-D
13:12 kfc joined #git
13:12 HardlySeen joined #git
13:12 mmlb joined #git
13:12 JeroenT_ joined #git
13:12 Vampire0 reallyfu, whatever you mean with your message, you should probably not use the FTP protocol. It is deprecated and only supported for backwards compatibility. It is dumb, legacy and inefficient. So if you have any chance, don't use FTP.
13:13 Snugglebash joined #git
13:14 reallyfu Vampire0: It was just meant as a joke refering to the article linked 42 minutes ago.
13:14 jast Vampire0: EWHOOSH
13:17 JeroenT joined #git
13:18 m1m3-50 joined #git
13:19 jameser joined #git
13:19 solenodic joined #git
13:22 Vampire0 reallyfu, ah, ok :-)
13:22 Vampire0 jast, what?
13:22 jast never mind :)
13:22 lmatteis joined #git
13:22 Vampire0 I don't :-)
13:22 osse Vampire0: it's the new meme. Use E for ERROR and then a made up error code
13:22 wiesel joined #git
13:22 jast new?
13:22 osse "No such file or directory" is ENOENT. "Didn't get the joke" is EWHOOSH
13:22 Vampire0 Ah, I see, thx osse :-)
13:22 jast EAGAIN
13:23 tvw joined #git
13:23 osse also shouldn't it be -EWHOOSH ?
13:23 jast in my stupid jokes I make the rules
13:23 Vampire0 Who cares about signs :-D
13:24 theoceaniscool joined #git
13:26 zefferno joined #git
13:28 grawity hmm, freebsd actually has -EDOOFUS
13:28 lmatteis guys. say i have a file.txt. i add "foo" to it and commit such as "Add foo". then i add bar to it, before the foo, and the file becomes: "bar\nfoo" and commit it as "Add bar". at a later point then i decide that the commit "Add foo" needs a change, it should've been "foo-" rather than "foo". so i go ahead and commit a fixup to the "Add foo" commit with such
13:28 lmatteis change.
13:28 lmatteis when i rebase, i get conflicts and in more complicated scenarios it's really hard/confusing to understand what should be deleted/changed
13:28 irqq joined #git
13:28 lmatteis any help?
13:29 bongjovi joined #git
13:30 zerorax left #git
13:30 tobiasvl lmatteis: !eek
13:30 gitinfo lmatteis: [!eekaconflict] Merge conflicts are a natural part of collaboration. When facing one, *don't panic*. Read "How to resolve conflicts" in man git-merge and http://git-scm.com/book/ch3-2.html#Basic-Merge-Conflicts then carefully go through the conflicts. Picking one side verbatim is not always the right choice! A nice video explaining merge conflicts: https://www.youtube.com/watch?v=zz7NuSCH6II
13:30 Aleric lmatteis: I don't think you should do a rebase for that, just fix the commit with a third commit.
13:31 Aleric If you want to create a "clean git history" all you do is create unnecessary work for yourself.
13:31 lmatteis we try to keep commits atomic. having useless commits in a history doesn't make much sense, especially for reviews
13:32 Aleric That is a source of endless pain imho.
13:32 osse atoms come in different sizes
13:32 lmatteis a PR is a good enough balance to have a clean history, for reviews
13:32 seni joined #git
13:32 osse lmatteis: there is no good way around that.
13:33 reallyfu osse: NEW meme? My classmate used that half a decade ago.
13:33 Aleric Anyway, you might want to look into three-way merging if you need to resolve conflicts.
13:33 osse if you expect all these commits to be squashed in the future anyway then using --amend as you develop might be better
13:33 reallyfu EURLATE
13:33 osse reallyfu: i've just seen an increase in them lately
13:34 * reallyfu goes to the I-did-X-before-it-was-cool-room.
13:34 lmatteis osse: how so?
13:34 reallyfu However, I need to leave you now. Bye and thanks for the help.
13:35 osse lmatteis: because then add foo and and bar would be in the same commit so it's easier to fixup
13:35 lmatteis osse: not really, they would be in separate commits. just the last modification should be in the "Add foo" commit
13:36 borkr joined #git
13:36 osse lmatteis: not if what I said applied
13:37 rical joined #git
13:38 rical I have added a file, not commited, and then modified again, I want to reset it to the added state, anyone know how?
13:38 rical checkout?
13:39 nowhereman joined #git
13:39 _ikke_ yes
13:39 _ikke_ git checkout <file>
13:41 DaveTaboola joined #git
13:41 rical thx
13:46 P4nther joined #git
13:47 Wulf joined #git
13:47 Wulf Good Morning!  Can I use git to fetch a single file in its current version from a git:// url?
13:47 grawity no
13:48 Wulf :(
13:48 grawity you can use `git clone --depth=1` to fetch the latest commit of the whole repo's default branch, but it's a whole commit or nothing
13:48 grawity use regular HTTP for serving individual files
13:48 p4trix joined #git
13:48 grawity there are cgi scripts for serving them straight out of Git, though I dunno about performance
13:49 cr34ton joined #git
13:50 jeffreylevesque joined #git
13:50 nettoweb joined #git
13:52 noc_ joined #git
13:53 Wulf --depth is interesting, but doesn't help me here. Just out of interest, How can I "repair" my clone after I used --depth?
13:54 grawity `git fetch --depth=100` to get some more (e.g. deepen to 100 commits total)
13:54 qqx Wulf: `git fetch --unshallow`
13:54 grawity `git fetch --depth=99999999999` or recently `git fetch --unshallow` to make it full
13:54 Gsham joined #git
13:55 Wulf thanks :)
13:55 aleb joined #git
13:55 knigitz joined #git
13:58 red82 joined #git
13:59 Erati joined #git
13:59 nowhereman joined #git
14:01 Vampire0 joined #git
14:01 _ADN_ joined #git
14:02 jast the closest you can get with native methods is (a) created an annotated tag on the file/blob object or (b) fetching the contents of a tree object (using git archive). the latter requires that feature being enabled on the server.
14:03 Wulf I'll do something completely different now
14:05 NullableTruth joined #git
14:06 Tobbi joined #git
14:09 __main__ joined #git
14:11 rorro joined #git
14:11 guampa joined #git
14:14 Phil-Work joined #git
14:15 YuGiOhJCJ joined #git
14:16 Phil-Work what's the best way to back/forward port a change (I'd normally cherry-pick) between branches when the file the change was made on is in a different directory on the target branch?
14:16 red82 joined #git
14:16 grawity first try cherry-pick regardless
14:16 Phil-Work yeh, it's not happy
14:17 grawity it can deal with renames up to a point
14:17 grawity if it falls over, then one manual equivalent would be `git format-patch <hash>^! > foo.patch`, adjust the paths, switch branch, then `git am < foo.patch`
14:17 Phil-Work ok
14:18 m1m3-50 joined #git
14:19 aleb I run "git fetch origin refs/notes/reviews" and it reports "refs/notes/reviews -> FETCH_HEAD". Can I reference that by something else besides FETCH_HEAD? (without specifying a temporary destination) I was looking into refs/remotes/origin/notes/reviews, but it does not exist.
14:20 publio joined #git
14:21 rnsanchez joined #git
14:22 aleb I run "$ git notes --ref refs/notes/reviews merge alskjfslfkjsldfkj " and it does not return any error. Should I file a bug?
14:24 jimi_ joined #git
14:24 JeroenT joined #git
14:25 Hudu joined #git
14:25 ShekharReddy joined #git
14:26 mkoskar joined #git
14:26 Phil-Work grawity, the am doesn't apply cleanly, which is possible - how do I get it to give me a conflict file to resolve?
14:28 byte512 joined #git
14:28 e14 joined #git
14:29 mischat_ joined #git
14:29 rivarun joined #git
14:30 tvw joined #git
14:32 Derperperd joined #git
14:33 ikelso joined #git
14:34 King_Hual joined #git
14:34 King_Hual joined #git
14:41 gugah joined #git
14:41 dviola joined #git
14:41 gugah joined #git
14:42 ShalokShalom joined #git
14:44 mikecmpbll joined #git
14:44 ProbabilityMoon joined #git
14:44 Vampire0 Phil-Work, after using `git am` the conflicted files should be in conflicted state like usual so you can use a text editor or `git mergetool` to resolve the conflict
14:45 Phil-Work Vampire0, that's what I assumed, but I don't see anything
14:50 fahadash joined #git
14:51 solenodic joined #git
14:51 Tykling joined #git
14:51 Murii_ joined #git
14:53 Nnarol joined #git
14:53 Es0teric joined #git
14:53 hahuang61 joined #git
14:53 jayaura joined #git
14:57 cdg joined #git
14:58 red82 joined #git
14:59 Nnarol left #git
14:59 nickabbey joined #git
14:59 Nnarol joined #git
15:03 Nnarol Hi! I am trying to clone a shared local repo through LAN. Both the local repo and my PC have Windows. The command I'm trying is "git clone file://PC_name\\name_of_shared_folder" and I get the error "fatal: '<full_path_to_shared_repo_on_hosting_PC>' does not appear to be a git repository\nfatal: Could not read from remote repository". The path does seem to be OK, since Git expands it to the full path.
15:04 mathematic-alpha joined #git
15:04 Nnarol The folder which is the git repository is also properly shared with me and I have all the necessary permissions. So much so that I can in fact change it through Windows Explorer.
15:04 Nnarol What could be the problem here?
15:06 Dougie187 joined #git
15:06 pur3eval joined #git
15:06 tvw joined #git
15:06 Abbott left #git
15:07 shinnya joined #git
15:07 pietv joined #git
15:09 holodoc joined #git
15:10 Snugglebash joined #git
15:10 jimi_ joined #git
15:10 jimi_ joined #git
15:10 nckpz joined #git
15:11 AaronMT joined #git
15:12 steelnwool joined #git
15:12 steelnwool Hi. I believe I started a new branch and commited/pushed changes with it before doing a pull. How can I remedy this and bring it up to speed, then re commit/push?
15:14 LeBlaaanc joined #git
15:14 cdg joined #git
15:15 jimi_ joined #git
15:18 adrianh joined #git
15:19 adrianh Hello?
15:19 gitinfo adrianh: hi! I'd like to automatically welcome you to #git, a place full of helpful gits. Got a question? Just ask it — chances are someone will answer fairly soon. The topic has links with more information about git and this channel. NB. it can't hurt to do a backup (type !backup for help) before trying things out, especially if they involve dangerous keywords such as --hard, clean, --force/-f, rm and so on.
15:20 adrianh I'm having some difficulty with using *xfuncname*
15:20 Guest9 joined #git
15:20 bannakaffalatta joined #git
15:20 DevAntoine joined #git
15:20 DevAntoine hi
15:21 DevAntoine if I do "git merge --squash foo" it won't actually merge the branch right? (under bitbucket I can see that my squashed/merged branch is not merged)
15:21 adrianh I'm trying to define a hunk header for a natvis file (XML), but it doesn't seem to be working.
15:21 DevAntoine so it's a bad practice?
15:21 treia joined #git
15:21 DevAntoine Ideally when I merge a branch into develop I'd like to only have the merge commit but idk how to do it
15:21 jast DevAntoine: correct, it's basically a plain commit that has the same file contents as a proper merge would
15:21 adrianh I can describe it here, but I've already posted a question on SO if you want to look at that.
15:21 adrianh http://stackoverflow.com/questions/42078376/why-isnt-my-xfuncname-working-in-my-gitconfig-file/42080115#42080115
15:22 jast DevAntoine: when you have real merges, you can tell your history viewer to simplify the history, e.g. leave out the commits that got merged in: git log --first-parent
15:22 DevAntoine jast: yeah, not good.
15:22 DevAntoine Can I only have the merge commit (without rebasing and squashing) for everyone and not just for me because I tell git to only show the first parent?
15:23 jast nah, that's not how merges work. other people have to choose themselves what they want to see.
15:24 ash_workz joined #git
15:24 a_thakur joined #git
15:25 nickabbey joined #git
15:25 adrianh Apparently, someone has tried this:
15:25 adrianh [diff "natvis"]
15:25 adrianh xfuncname = "^[ \\t]*<Type Name=\"([^\"]*)\".*$"
15:26 adrianh and it worked for them, but it doesn't work for me.
15:26 adrianh I'm wondering if this is a bug in my version or not.
15:28 wiesel joined #git
15:29 le_melomane joined #git
15:31 matsaman joined #git
15:32 DevAntoine okay so tje only way to have a single commit is to rebase the branch I want to merge
15:33 freimatz3 joined #git
15:33 adrianh I guess no one is free at this time.
15:34 adrianh If someone could look at the question at SO, that would be great.  Here's the link again:  http://stackoverflow.com/questions/42078376/why-isnt-my-xfuncname-working-in-my-gitconfig-file/42080115#42080115
15:35 mikecmpb_ joined #git
15:35 nettoweb joined #git
15:36 le_melomane joined #git
15:36 zacts joined #git
15:38 rgrinberg joined #git
15:39 bannakaf_ joined #git
15:41 eroux joined #git
15:42 jast DevAntoine: rebase doesn't normally reduce the number of commits
15:43 Derperperd joined #git
15:43 jast the order just ends up different
15:44 jast adrianh: "free" doesn't factor in. I just don't have anything helpful to say. have you double-checked that your .gitattributes file assigns that diff driver to the right files?
15:44 sbulage joined #git
15:45 jast adrianh: never mind, that's in your SO post, I just saw
15:45 adrianh @jast: yeah, did that.
15:45 adrianh Np.
15:46 DevAntoine jast: no because with rebase you can squash them
15:47 jast adrianh: have you copied the regex from that other guy on SO? because there's another unmentioned change in it, the added '*' in the part that matches Type="..." - if you edited your regex to try and match the details in the comments, you may have missed that bit.
15:48 jast DevAntoine: if you squash, there's no real difference to a merge --squash
15:48 adrianh Yeah, I cut and pasted, no joy.
15:48 adrianh :(
15:49 jast I don't know what regex implementation git is using, maybe yours is different and a little stricter... have you tried [[:space:]] instead of your [ \\t]?
15:51 adrianh Funny thing is if I have a non-blank line before the added line, and I use _xfuncname = "^.*$"_, it still doesn't work. :?
15:52 zivester joined #git
15:52 nickabbey joined #git
15:52 DevAntoine jast: of course there is, I squash on the branch I want to merge then I merge this branch we result in a real merge not a merge --squash
15:52 a3Dman joined #git
15:52 adrianh It's like it doesn't even know what xfuncname is for.
15:53 le_melomane joined #git
15:53 jast DevAntoine: okay, didn't know you were going to use rebase *and* merge
15:53 jast so that will actually give you two commits... the squashed thing and the merge commit bringing it in
15:54 e14 joined #git
15:55 solenodic joined #git
15:56 alwyn joined #git
15:57 jayaura Hello, I'm using gogs for my git server. Gogs doesnt have "protected branches" feature yet. And I want to prevent merge/commit rights to specific users to say "stable" branch. I should be able to find a solution using pre receive hooks. But not sure how to check for user
15:57 jayaura any suggestions ?
15:58 alwyn Hey, I just committed something on a server (dirty, I know) using --author="My Name <my@email.com>" and even though it displays the author fine it says the committer is the local server's user/hostname, and adds that I should configure it (which I should, of course). I thought --author would set both the author and committer?
15:58 GodGinrai joined #git
15:58 pietv joined #git
15:58 le_melomane joined #git
15:59 Snugglebash joined #git
15:59 roelmonnens joined #git
16:00 jayaura alwyn: I think commite changes when you take a patch from some person (created with format-patch) and apply to the repo using am
16:00 thethorongil joined #git
16:01 jayaura s/commite/commiter
16:02 Noldorin joined #git
16:02 m1m3-50 joined #git
16:03 xall joined #git
16:04 Gsham joined #git
16:04 alwyn The docs say ... --author=<author>
16:04 alwyn Override the commit author.
16:05 alwyn Which doesn't mention committer...
16:05 alwyn I guess it can't really hurt but it'd still be nice
16:06 math-alpha joined #git
16:06 Derperperd joined #git
16:06 amagawdd joined #git
16:07 pur3eval joined #git
16:09 roelmonn_ joined #git
16:11 gugah joined #git
16:12 hashpuppy joined #git
16:15 wrouesnel1 joined #git
16:15 texinwien joined #git
16:16 osse alwyn: The only way to overwrite the committer is with the env var
16:17 invisbl joined #git
16:17 durham joined #git
16:17 re1 joined #git
16:18 invisbl joined #git
16:21 e14 joined #git
16:23 a_thakur joined #git
16:23 miczac joined #git
16:24 Esya joined #git
16:24 crose joined #git
16:25 solenodic joined #git
16:25 jast jayaura: you can only do that in a hook if the Git service (gogs in this case) exposes user information to hooks in some way. I'm not familiar with how this is done in gogs, maybe you should ask the gogs folks?
16:25 shgysk8zer0_ joined #git
16:26 jagob joined #git
16:27 jast alwyn: for future dirty hacks: git -c user.name="Your Name" -c user.email="your@email.example" commit
16:28 brent__ joined #git
16:29 jayaura jast: you're right. I just dropped a line waiting for response. but would this work: I import the gpg keys of the people with commit access to my server, and do `git verify-commit` for all the pushes, and decline the push is verify says public key not found. But then the issue would be that what I'm trying to push will have some commits from users who do not have push rights., these commits will make the script fail
16:30 overlord_tm joined #git
16:31 jast jayaura: yeah, that's why you should always check the push and not the commits in it
16:31 Snugglebash joined #git
16:31 calfly joined #git
16:32 calfly left #git
16:32 menip joined #git
16:32 jast you can restrict your Git usage such that checking the commits becomes feasible, but your hook will get fairly long and people will take a while to get used to the restrictions
16:32 nowhereman joined #git
16:32 jast I did this in a setup in the past. all commits to the affected branches had to be merge commits. I then validated the merge commit but not the commits it merged in.
16:33 m1m3-50 joined #git
16:33 jayaura Lets assume I write the sciript to allow only merges to stable, and no single commits. Now I would do git verify-commit for that merge to see if the commiter of the merge is a whitelisted person. Would this work? Would the other commits go in through the merge or would the commits interfere ?
16:34 jayaura ah right!
16:34 pijiu2 joined #git
16:34 jast there are edge cases where this becomes tricky, such as force pushes
16:34 jayaura I see
16:36 alwyn jast: thanks! Hadn't thought of that :)
16:38 dendazen joined #git
16:40 Snugglebash joined #git
16:40 Vampire0_ joined #git
16:41 e14 joined #git
16:42 rivarun joined #git
16:44 diogenese joined #git
16:45 shgysk8zer0_ joined #git
16:45 WizJin joined #git
16:46 chipotle joined #git
16:47 mikecmpbll joined #git
16:48 regedit joined #git
16:48 adrianh jast: yes, I've tried [[:space:]] with no joy.
16:49 manuelschneid3r joined #git
16:50 hobodave joined #git
16:50 Emperor_Earth_ joined #git
16:51 synthroid joined #git
16:52 nckpz joined #git
16:53 matoro joined #git
16:54 hahuang61 joined #git
16:55 solenodic joined #git
16:56 Snugglebash joined #git
16:57 tyreld joined #git
16:57 svm_invictvs joined #git
16:58 [0xAA] joined #git
17:04 sbeller joined #git
17:04 pietv joined #git
17:04 madewokherd joined #git
17:07 thiago joined #git
17:08 [0xAA] joined #git
17:13 govg joined #git
17:14 miczac_ joined #git
17:15 svm_invictvs joined #git
17:17 jamessexxual joined #git
17:17 solenodic joined #git
17:17 VladGh joined #git
17:20 nixjdm joined #git
17:21 crose joined #git
17:22 repozitor joined #git
17:23 miczac_ joined #git
17:23 crayon joined #git
17:25 svm_invictvs joined #git
17:25 Cthalupa joined #git
17:27 p4trix joined #git
17:28 nickabbey joined #git
17:28 adrianh Just reposting my question.  Can someone give me a hand with *xfuncname*.  See my question on SO here:  http://stackoverflow.com/questions/42078376/why-isnt-my-xfuncname-working-in-my-gitconfig-file/42080115#42080115
17:30 Macaveli joined #git
17:30 ShapeShifter499 joined #git
17:31 Derperperd joined #git
17:31 jccn joined #git
17:32 k_sze[work] joined #git
17:35 matoro joined #git
17:36 Mikerhinos joined #git
17:38 pur3eval joined #git
17:38 thierryp joined #git
17:38 miczac_ joined #git
17:40 adamtoakley_ joined #git
17:41 [0xAA] joined #git
17:42 Es0teric joined #git
17:42 chardan joined #git
17:45 sbeller joined #git
17:46 drodger joined #git
17:47 lyxus joined #git
17:47 dreiss joined #git
17:48 svm_invictvs joined #git
17:48 chardan joined #git
17:49 repozitor joined #git
17:49 Guest9 joined #git
17:52 repozitor joined #git
17:54 denisMone joined #git
17:55 nickabbey joined #git
17:55 repozitor can i ask question about git/gogs here?
17:56 repozitor i addess my public key to authorized_keys in .ssh directory
17:56 repozitor gogs installed very well.
17:56 repozitor now i can clone url with http and https very well
17:56 repozitor but i can't clone with ssh
17:57 robotroll joined #git
17:57 repozitor actually i ssh ask me the git account password, which is set to disable by gogs installation docs.
17:58 _ikke_ repozitor:
17:58 _ikke_ repozitor: try ssh -t -vv user@host
17:58 _ikke_ see if the key is offered
17:58 _ng joined #git
17:59 deepy joined #git
17:59 repozitor http://paste.ubuntu.com/23949078/
18:00 dviola joined #git
18:00 _ikke_ Ok, it is offered, but not accepted
18:01 repozitor so what is wrong with me?
18:01 u1000 joined #git
18:01 Dougie187 joined #git
18:02 repozitor i don't like to set password for git account on my machine and broadcast between users.
18:02 repozitor do you have any solution?
18:02 _ikke_ Verify if the ssh public key on the server has the correct format?
18:02 jnavila joined #git
18:03 repozitor i don't like to broadcast my git account private key between users.
18:04 repozitor asking for git password is incorrect by gogs
18:04 repozitor actually it should ask for related DB account password
18:04 ruxu joined #git
18:04 repozitor i mean password of owner repository.
18:04 _ikke_ Not sure if gogs has its own ssh implementation
18:05 repozitor we can add ssh public key to our repository, but i don't know why gogs didn't ask it.
18:05 repozitor instead it ask git user of linux.
18:06 _ikke_ That's the default behaviour if the ssh key authentication failed
18:08 _ikke_ Looking at the documentation, but cannot find much about ssh keys
18:08 repozitor do you understand my problem?
18:09 boombatower joined #git
18:09 _ikke_ https://discuss.gogs.io/t/how-to-config-ssh-settings/34
18:09 repozitor something goes wrong with me!
18:09 repozitor setting ssh public key in gogs seems useless.
18:09 _ikke_ See common problems
18:09 _ikke_ "This is why you have to add your public key through Gogs web UI, and want to make sure same public key only exists once in the file, which is generated by Gogs."
18:09 thiago joined #git
18:10 Darren__ joined #git
18:12 Levex joined #git
18:12 Groscheri joined #git
18:12 Emperor_Earth joined #git
18:18 doener joined #git
18:19 tang^ joined #git
18:20 Es0teric joined #git
18:20 mohabaks joined #git
18:20 raynold joined #git
18:21 ShalokShalom joined #git
18:21 hahuang61 joined #git
18:25 synthroi_ joined #git
18:25 kasual joined #git
18:26 johnny56 joined #git
18:29 dviola joined #git
18:31 synthroid joined #git
18:33 yehowyada joined #git
18:34 wad joined #git
18:34 pietv joined #git
18:35 [0xAA] joined #git
18:35 wad Anyone have a favorite linux git gui client? I've found SmartGit, git-cola, GitEye, giggle, gitg, GitKraken.
18:35 * GodGinrai prefers the good old fashioned terminal
18:36 tango_ command line, at most tig for browsing about and sometimes to do partial commits
18:37 nickabbey joined #git
18:38 * wad looks at tig
18:38 ikelso joined #git
18:38 _ikke_ another one for tig
18:38 duderonomy joined #git
18:39 joshszep joined #git
18:41 matoro joined #git
18:41 wad Thanks!
18:42 le_melomane joined #git
18:43 eroux joined #git
18:43 cbreak git command line is ultimate power
18:44 jraccoon joined #git
18:44 Eugene THE REAL ULTIMATE POWER!?!?
18:45 BSaboia joined #git
18:47 sbeller_ joined #git
18:47 osse Power is not a means; it is an end.
18:47 GodGinrai Eugene: foshizzle dizzle dawg
18:48 Levex joined #git
18:50 jfran joined #git
18:50 gkatsev joined #git
18:53 dave0x6d joined #git
18:55 e14 joined #git
18:56 rominronin joined #git
18:56 prg3 joined #git
18:57 finalbeta joined #git
18:58 adrianh Just reposting my question.  Can someone give me a hand with *xfuncname*.  See my question on SO here:  http://stackoverflow.com/questions/42078376/why-isnt-my-xfuncname-working-in-my-gitconfig-file
18:58 adrianh Thanks
19:00 finalbeta joined #git
19:02 Gsham joined #git
19:03 raygn joined #git
19:03 Gsham joined #git
19:04 j7k6 joined #git
19:05 Gsham joined #git
19:05 tmk joined #git
19:05 [0xAA] joined #git
19:07 astrofog joined #git
19:08 ash_workz joined #git
19:08 Gsham joined #git
19:08 pur3eval joined #git
19:08 j7k6 joined #git
19:09 MineCoins joined #git
19:10 MrWoohoo joined #git
19:11 MACscr joined #git
19:11 le_melomane joined #git
19:12 Levex joined #git
19:12 MACscr is there a way to force push just a single file to a repo without pulling any other changes?
19:12 gajus joined #git
19:13 MACscr i have about 12 systems that push to a single repo and im way behind on commits and have a lot of stuff to get sorted out, but i really want to push this one file there
19:14 Sasazuka joined #git
19:15 kadoban Pushing a single file doesn't really make any sense. push deals with commits, not files.
19:15 nettoweb1 joined #git
19:15 raygn joined #git
19:16 vuoto joined #git
19:16 howitdo joined #git
19:16 howitdo joined #git
19:16 raedah joined #git
19:17 raygn I am trying to do a git diff repA/BranchA  repB/BranchB and then copy the diff files from repA/BranchA that are new or modified to local directory  better explanation and git commands used are here http://stackoverflow.com/questions/42080681/git-diff-repa-brancha-rebb-branchb-and-only-to-return-items-from-brancha
19:17 MACscr well can i make sure only this one file is committed?
19:17 nimbleark joined #git
19:18 MACscr from one i can see, its only this one file in the commit
19:19 matsaman joined #git
19:21 matoro joined #git
19:22 MrMojit0 joined #git
19:24 griffindy joined #git
19:24 tmsmith joined #git
19:25 KV joined #git
19:26 The-Compiler joined #git
19:26 The-Compiler I did a "git gc" in a clone of https://github.com/WebKit/webkit and .git growed from 7GB to 16GB... That shouldn't happen, right? :D
19:26 The-Compiler err, in a clone of https://github.com/annulen/webkit
19:26 Atemu joined #git
19:28 Vampire0_ joined #git
19:28 grawity what do you see in .git/objects/? does it shrink after a `git prune`?
19:30 cr34ton joined #git
19:30 The-Compiler I'll try, currently doing a git gc --aggressive (as this was originally an SVN import afaik)
19:32 onmeac joined #git
19:33 Es0teric joined #git
19:34 menip joined #git
19:36 jnavila joined #git
19:36 sgen joined #git
19:37 troulouliou_dev joined #git
19:38 BackEndCoder joined #git
19:38 mikecmpbll joined #git
19:38 jedahan joined #git
19:40 eroux joined #git
19:47 jsolano joined #git
19:47 zerorax joined #git
19:49 Sick joined #git
19:49 jimi_ joined #git
19:49 zerorax left #git
19:51 [mgv] joined #git
19:54 [mgv] joined #git
19:54 devhost joined #git
19:54 Guest9 joined #git
19:54 Balliad joined #git
19:55 caao joined #git
19:56 devhost I'm trying to convert my company to git from svn, and one of the complaints I've had from testers is copying and keeping track of the commit SHA1. I personally think it's lazy to not want to copy and paste this into bugs, but they are use to remembering 4 digit revision.
19:57 devhost Does anyone have an elegant solution? Can they just use the first four characters to reference the IDs or is collision really a problem?
19:57 tang^ six
19:57 perlpilot 5 or 6 chars usually will do you well
19:57 kadoban devhost: You can typically abreviate sha-ids and git won't care. 4 characters sounds a bit short, but 6 or 7 is typical.
19:57 tang^ six is typical
19:57 _ikke_ Right, but that does not scale over time
19:57 tang^ also, have your commits reference your bug id instead
19:57 devhost I knew I could abbreviate, but didn't know 6 would be good.
19:58 perlpilot devhost: if they are cutting and pasting anyway, what difference does it make how long they are?
19:58 devhost Yeah, tang^ sure but that's on the developer side
19:58 _ikke_ (see recent git.git topic mentioning how soon you would already supersede 5/6 characters)
19:58 devhost The problem is they aren't cutting and pasting.
19:58 devhost They remember revision numbers
19:58 perlpilot devhost: then they're doing it wrong  :-)
19:58 devhost perlpilot: I completely agree.
19:58 devhost This whole issue makes me upset.
19:58 _ikke_ revision numbers are not going to work in a decentralized version control system
19:58 devhost But non-the-less I want to convert them over.
19:58 tang^ devs should be writing code to fix issues and reference the issue
19:59 tang^ rather than writing issues that reference a specific commit
19:59 devhost _ikke_: I understand the reason why git uses SHA1
19:59 _ikke_ devhost: question, why do they need to remember revision numbers?
19:59 devhost That's how the USE to do it in SVN.
19:59 devhost They would download r111 and then just work on finding bugs
20:00 tang^ testers
20:00 devhost yep
20:00 perlpilot They do have a point when *talking* about revisions ... it's much easier to say "revision 1537" than "revision 0152316f56121ecd6236b4539308cb8341cec6d8"
20:00 devhost manually testing too
20:00 devhost <_<
20:00 _ikke_ in git you talk more about branches / refs
20:00 tang^ testers should be just downloading latest, and adding bug reports regardless of "revision"
20:00 tang^ let the devs fix the issue and reference it in the commit
20:00 devhost hmm
20:01 tang^ and have your scm tell your bug system what commit updates what bug
20:01 devhost We wanted to provide tarballs on an a server for the testers
20:01 perlpilot devhost: do you use anything like JIRA for tracking work?
20:01 devhost and not even have them use git
20:01 devhost bugzilla
20:01 devhost But
20:01 devhost I setup gitlabs
20:01 devhost for devs
20:01 rominronin joined #git
20:02 devhost gitlabs and bugzilla integration isn't really amazing, I'd have to write some abstractions on their APIs or find some project
20:02 perlpilot devhost: then, if everyone makes named branches for bugs/features/etc., they can talk about those branch names with ease (if that makes a difference)
20:02 devhost Hmm.
20:03 devhost Yes I see what your point is.
20:03 mvensky joined #git
20:03 devhost Thanks for the help everyone.
20:04 devhost wish me luck
20:04 perlpilot devhost: good luck!  (and, more importantly, have fun!  :-)
20:04 ToBeCloud joined #git
20:04 e14 joined #git
20:04 irqq joined #git
20:04 devhost Getting GitLabs as infrastructure has been fun.
20:04 devhost :)
20:06 raygn trying to get the new or modified files from a branch in the main repo using a diff that compares the that branch to the systems nightly backup backup
20:06 j7k6 joined #git
20:06 quackgyver joined #git
20:06 raygn doing git diff repoA/BranchA /repoB/nightlyBackup --name-status --diff-filter=AM
20:06 j7k6 joined #git
20:07 raygn then I want to copy the files that are different in the repoA/Branch to be copied down locally to a new dir
20:07 netj joined #git
20:08 kadoban ... are you building a backup system on git or something?
20:08 zezba9000 joined #git
20:08 j7k6 joined #git
20:09 raygn kind off, I have the backup stuff working, but the new code in that branch I want to download so I can build a package to deploy
20:09 j7k6 joined #git
20:09 gopar joined #git
20:09 raygn but I only want things that are new or modified
20:10 kadoban Why? Don't you need the whole thing to build it or deploy it?
20:11 Darren__ joined #git
20:11 SwiftMatt joined #git
20:11 j7k6 joined #git
20:11 raygn the code base is APEX so it only needs the updated files
20:11 raygn I do not need to do a compile or anything
20:12 raygn currently we do pull it all and deploy that, but it takes time
20:13 j7k6 joined #git
20:14 hobodave joined #git
20:15 freimatz joined #git
20:16 ash_workz joined #git
20:18 Derperperd joined #git
20:19 Joost joined #git
20:19 Endarked joined #git
20:21 mellernoia joined #git
20:22 nidr0x joined #git
20:23 solenodic joined #git
20:24 ash_workz joined #git
20:26 pietv joined #git
20:26 Balliad joined #git
20:27 netj joined #git
20:27 Borg joined #git
20:28 Darcidride_ joined #git
20:28 Borg left #git
20:28 Borg^ joined #git
20:28 Borg^ hi....
20:28 gitinfo Borg^: hi! I'd like to automatically welcome you to #git, a place full of helpful gits. Got a question? Just ask it — chances are someone will answer fairly soon. The topic has links with more information about git and this channel. NB. it can't hurt to do a backup (type !backup for help) before trying things out, especially if they involve dangerous keywords such as --hard, clean, --force/-f, rm and so on.
20:29 Borg^ I wonder about SHA-1 collisions in git.. I know that probability is very very small. but still its not 0..
20:29 Borg^ I understand that adding protection to it.. its very expensive. and thats why it wasnt implemented at all?
20:29 j7k6 joined #git
20:30 osse Borg^: AFAIK it is in fact implemented, but it has never run :P
20:30 kadoban There is no way known to give 0 probability of collisions, except not to use hashes at all, just use the same thing. But having commit IDs gigabytes in length would rather defeat the purpose.
20:31 Borg^ osse: AFAIK.. its not...
20:31 Borg^ kadoban: what do you mean?
20:31 gdg joined #git
20:31 j7k6 joined #git
20:31 kadoban Borg^: Please describe a hashing system that has no chance of collisions, ever.
20:32 Borg^ kadoban: there is none... and you understand me wrongly..
20:32 gdg I am using git 2.7.4 and using a pre-commit hook, it returns git-diff-index: command not found
20:32 Borg^ the protection, I mean.. hash blob... check if such hash already exist.. yes? report to user..
20:32 gdg how is possible that I do not have git-diff-index?
20:32 Borg^ afaik, currently.. unpredictable stuff start to happen..
20:32 nettoweb joined #git
20:33 Borg^ thats why I wonder...  because.. on every git add <> you basicaly neeed to compare with every hash in repo.. and that loooks expensive to me
20:33 m0viefreak joined #git
20:34 _ikke_ Borg^: git does detect hash collisions
20:34 _ikke_ it does not naively overwrite existing blobs with the same hash
20:34 Sasazuka_ joined #git
20:34 qqx gdg: Is the hook referring to `git-diff-index` literally (with the first -)? That's been obsolete for awhile.
20:34 Borg^ _ikke_: is it recent?
20:34 _ikke_ no
20:35 gdg yeah, I am assuming that the pre-commit script may be old
20:35 Borg^ http://stackoverflow.com/questions/9392365/how-would-git-handle-a-sha-1-collision-on-a-blob
20:35 Borg^ _ikke_: how would you comment that then? when dude did experiment with 4bit hash size?
20:35 qqx gdg: You could try changing that first "-" to a space.
20:37 mischat joined #git
20:40 The-Compiler error: pack-objects died of signal 9143712)
20:40 The-Compiler that's... an interesting output
20:41 Borg^ catastrophic ;)
20:41 The-Compiler grawity: but it's back to 6.9 GB after that (even though it failed), and "git prune" doesn't change much
20:44 nunchuck joined #git
20:45 Borg^ _ikke_: no comments?
20:47 _ikke_ Borg^: Still reading / checking
20:49 matsaman joined #git
20:52 hobodave_ joined #git
20:53 j7k6 joined #git
20:54 Balliad joined #git
20:57 j7k6 joined #git
20:57 pur3eval joined #git
20:58 j7k6 joined #git
20:58 Ankhers joined #git
20:59 j7k6 joined #git
21:03 j7k6 joined #git
21:04 j7k6 joined #git
21:04 Atm0spher1c joined #git
21:04 gdg I am moving here from #linux, let me re-iterate a question. I am writing some linux kernel device drivers, and I would like a pre-commit script that check they coding style against the linux kernel coding style
21:04 gdg do you have any suggestion?
21:05 _main_ joined #git
21:05 gdg I tried this script but it seems to be a little dated: https://github.com/mellowcandle/astyle_precommit_hook
21:05 nickabbey joined #git
21:07 __main__ joined #git
21:07 tvw joined #git
21:08 nickabbey joined #git
21:08 GodGinrai joined #git
21:08 GodGinrai joined #git
21:08 rivarun joined #git
21:11 mischat joined #git
21:11 _ikke_ Borg^: Hmm, so the collision check used to be in there one time, but somehow it got removed
21:11 Borg^ hmm... not good.. I wonder why...
21:12 _ikke_ (but it's something different from what I was refering to)
21:12 Borg^ they were probably too taxing...
21:12 ShalokShalom_ joined #git
21:13 _ikke_ A collision check is rather trivial
21:13 dsdeiz joined #git
21:13 _ikke_ At least the one that is referred to in that post
21:14 _ikke_ What suprises me, is that they seem to update the object each time you add it
21:14 _ikke_ let me check
21:14 Borg^ hmm.
21:14 _ikke_ I mean, every time you add the same object under a different name
21:14 The-Compiler left #git
21:15 d5sx43 joined #git
21:15 gugah joined #git
21:15 Borg^ hmm.. so thats bad.. because.. you actualy start to loosing data..  right?
21:16 synthroid joined #git
21:17 aidalgol joined #git
21:17 inflames joined #git
21:18 _ikke_ mtime does change indeed
21:18 _ikke_ added 2 files with the same content
21:18 texinwien joined #git
21:19 rominronin joined #git
21:19 cdown joined #git
21:20 texinwien_ joined #git
21:21 Borg^ _ikke_: okey, thx for info and checking.. so basically.. there is a problem with it.. that probably needs to rethink..
21:22 _ikke_ right, but still, the chance of that happening is not that high
21:22 cdg joined #git
21:22 _ikke_ https://public-inbox.org/git/20050424231605.30173.qmail@science.horizon.com/
21:22 Borg^ I know.. but still. im paranoid about DATA... and >0 zero. means can happen..
21:23 cdg joined #git
21:23 _ikke_ The change is >0 that your pc will blow up in your face :)
21:23 _ikke_ chance*
21:24 _ikke_ Or a plane will land on your house
21:24 Borg^ _ikke_: actualy.. this happened to me.. with CRT..
21:24 _ikke_ heh
21:25 bremner Borg^: make backups.
21:25 Borg^ bremner: well, im using git.. because with it.. i do NOT need to make backups
21:26 bremner no.
21:26 _ikke_ hah
21:26 _ikke_ !backup
21:26 gitinfo Worried about your data while trying stuff out in your repo? The repository in its entirety lives inside the .git directory in the root of your work tree so to backup everything `cp -a path/to/workdir path/to/backup` or equivalent will suffice as long as the repo is not modified during backup. See also http://sethrobertson.github.com/GitBestPractices/#backups
21:26 Borg^ thats why this brought my atention
21:26 _ikke_ git is *not* a backup
21:26 bremner there are _many_ ways for your disk to get hosed that have nothing to do with SHA1
21:26 Borg^ why?
21:26 bremner also, user error.
21:26 Borg^ bremner: well, I usualy have at least... 3 repos...
21:27 Borg^ 2 workstation.. and 1 server (bare repo).
21:27 Borg^ sometimes more.. keeping all the stuff synced..
21:27 bremner *shrug*. You're either serious about data integrity, and you make backups, or not.
21:27 Borg^ so.. if anything explodes.. I still have all my shit..
21:27 _ikke_ git:github:git/git::5c2a7fbc362e4227ced84c32c3fdc9682d085962
21:27 gitinfo Git web link: https://github.com/git/git/blob/HEAD/5c2a7fbc362e4227ced84c32c3fdc9682d085962
21:27 _ikke_ hmm, not what I meant
21:27 _ikke_ git:github:git/git:commit:5c2a7fbc362e4227ced84c32c3fdc9682d085962
21:27 gitinfo Git web link: https://github.com/git/git/blob/commit/5c2a7fbc362e4227ced84c32c3fdc9682d085962
21:28 Borg^ bremner: I do backups.. but not git repos.
21:28 _ikke_ Borg^: https://github.com/git/git/commit/5c2a7fbc362e4227ced84c32c3fdc9682d085962
21:28 _ikke_ That's how someone did the check before
21:28 bremner well, that's my free enterprise consulting for today, take it or leave it.
21:29 Borg^ _ikke_: *sigh*
21:29 Borg^ a portable way to do stuff..
21:29 Borg^ okey.. seems I will have to handle that issue myself.. as usual..
21:30 Borg^ anyway.. thx.. im glad that at least this was somehow confirmed.
21:31 matoro joined #git
21:32 _ikke_ Note that preparation are already being made to be able to switch to another hashing scheme (though there are no concrete plans yet to switch)
21:33 Borg^ wow.. why?
21:33 MattMaker joined #git
21:34 _ikke_ Borg^: Because they try to find a good way to do it and still remain someone backwards compattible
21:34 kadoban SHA1 is on its way towards a bad idea.
21:34 _ikke_ (and also because a lot of code needs to be changed, which happens slowly)
21:34 Borg^ but why change sha-1 ? for what git do with it it looks ok?
21:35 kadoban Git uses it for security in at least important way. Signed tags/commits depend on the security of sha1
21:35 ResidentBiscuit joined #git
21:36 _ikke_ https://public-inbox.org/git/CAPp-Vrb_n6z39RLHZ4AeUaBFiJfL3_xX8Utfq7+bTgzZrza58Q@mail.gmail.com/
21:36 JanC_ joined #git
21:36 Borg^ oh well.. no big deal after all :) git will have to support 2 hash func after "switch"
21:36 Borg^ to handle new and old repos..
21:36 Borg^ so.. finaly. repo version will bump up.. :)
21:37 _ikke_ Borg^: yeah, the discussion is still about the best way to do it
21:37 _ikke_ "Git signed tags and signed commits are cryptographically insecure,
21:37 _ikke_ they're useless at the moment."
21:38 Doow joined #git
21:38 Borg^ well yeah... but sha-1 as data integrity is good..
21:39 askb joined #git
21:39 _ikke_ md5 too :-)
21:39 Borg^ pasky: yoo pasky :)
21:40 Borg^ sure.
21:40 Borg^ but git uses sha-1 for content identification, etc.. so 128bit might be too small.
21:40 _ikke_ but as more and more code relies on git, the more people try to attack it to sneak in backdoors
21:40 _ikke_ git also uses sha1 for integrity
21:41 Borg^ yep
21:41 _ikke_ (hence the merkle-tree style of data storage)
21:41 Borg^ thats why I do not do backups of git repos..
21:41 Borg^ and I dont really get why bremner got so mad. about backups.. but okey.. its his opinion
21:42 _ikke_ Borg^: https://public-inbox.org/git/20130324183133.GA11200@sigill.intra.peff.net/
21:43 _ikke_ real-life case of corruption that was even propagated
21:44 Borg^ hmm
21:44 Borg^ but its hard-link related, right?
21:45 * bremner is not bad, just can't be bother to argue.
21:45 bremner err. not mad.
21:45 bremner I may well be bad.
21:45 _ikke_ not that I recall, and cannot find any reference to
21:45 _ikke_ "too perfect a mirror"
21:46 Borg^ we're using the fast "just hardlink it"  <- comment in mail
21:46 Borg^ I know git can use hardlinking when cloning local repos
21:46 Vampire0 joined #git
21:46 _ikke_ Yes, it does by default
21:47 Borg^ I think I never use hardlinking.. always force git to do copy via file:///
21:47 Borg^ but still, interesting.. I will take a look to it..
21:47 _ikke_ But this is just one case
21:48 _ikke_ detecting corruption is one thing, but recovering from corruption is something else
21:49 Borg^ yeah.. and backups can or cannot save you from that..
21:49 jeffreylevesque joined #git
21:49 Borg^ history showed that doing backup is not as easy as many thinks it is
21:50 vdv joined #git
21:50 _ikke_ Making backups is easy, restoring them not :P
21:50 Borg^ s/can or cannot/may or may not/
21:50 Borg^ _ikke_: hehe well.. depends from what part you are looking at them :)
21:54 Flaghacker joined #git
21:54 Vampire0_ joined #git
21:55 Flaghacker I'm trying to push to a remote repository, bit I get an error saying the remote contains work I don't have locally. Is there a way to preview the difference before actually pulling?
21:55 ToxicFrog Flaghacker: pull is just fetch + merge
21:55 ToxicFrog So run 'git fetch' and then you can inspect it with git diff, git log, git show, gitk, etc
21:56 Flaghacker Right fetch, I forgot about that one. Thanks a lot!
21:56 j7k6 joined #git
21:58 pur3eval joined #git
21:59 nickabbey joined #git
21:59 Borg^ bremner: I have a bit same. but only when other side doesnt give sane arguments
21:59 Borg^ ;)
21:59 nickabbe_ joined #git
21:59 Flaghacker I have a history that looks like this: (A->B->C) and (A->D->E), ie. there are two different branches that start at A. How do I move B and C to be behind E? The end result would look something like this: A->D->E->B->C
22:00 Borg^ anyway.. nite
22:00 kadoban git checkout one && git rebase two    should do
22:02 ToxicFrog Flaghacker: git checkout C; git rebase --interactive E
22:02 cdg_ joined #git
22:02 ToxicFrog And then reorder the commits
22:02 ToxicFrog Is probably the easiest way
22:02 ToxicFrog kadoban: they want the final history to be A-D-B-C-E, not A-B-C-D-E or A-D-E-B-C
22:03 kadoban ToxicFrog: They didn't say that
22:03 m0viefreak actually the latter is what they want
22:04 ToxicFrog Oh wait, so it is. I missed that bit and just saw the "move B and C to be behind E"
22:04 Flaghacker ToxicFrog, Thanks!
22:04 Flaghacker Ah oops
22:05 Flaghacker So what should I do then?
22:05 m0viefreak what kadoban said
22:05 Flaghacker And will that put one behind two or the other way around?
22:05 ToxicFrog Flaghacker: which outcome do you want? If you want the second thing you said (A->D->E->B->C) follow kadoban's instructions.
22:06 Flaghacker ToxicFrog, Yes that's indeed what I want, I"m sorry for the confusion :)
22:07 Flaghacker So it looks like
22:07 Flaghacker "git rebase X" moves the current commit behind X, right?
22:08 m0viefreak moves all history of the current branch, that is not contained already in x, behind x
22:08 Flaghacker I'm a bit afraid to mess up my history...
22:08 ToxicFrog m0viefreak: after X
22:08 cdown joined #git
22:08 ToxicFrog Flaghacker: if you mess up your history, you can always roll back to where you were before
22:09 ToxicFrog If you're really worried, make a !backup
22:09 gitinfo Worried about your data while trying stuff out in your repo? The repository in its entirety lives inside the .git directory in the root of your work tree so to backup everything `cp -a path/to/workdir path/to/backup` or equivalent will suffice as long as the repo is not modified during backup. See also http://sethrobertson.github.com/GitBestPractices/#backups
22:09 Sound joined #git
22:09 blackwind_123 joined #git
22:09 Flaghacker Ah cool, didn't know about that one! Thanks!
22:09 m0viefreak man git-rebase
22:09 gitinfo the git-rebase manpage is available at http://jk.gs/git-rebase.html
22:10 m0viefreak take a look at that, the description contains your requirement
22:10 Flaghacker m0viefreak, Yup I read that but I wanted to be sure.
22:10 hvxgr joined #git
22:10 rnsanchez joined #git
22:10 Atemu joined #git
22:11 ShalokShalom joined #git
22:11 nettoweb joined #git
22:11 zoph joined #git
22:14 Emperor_Earth_ joined #git
22:15 robnester joined #git
22:15 Flaghacker That worked flawlessly, thanks everyone!
22:17 ihradek joined #git
22:17 _ADN_ joined #git
22:17 hvxgr joined #git
22:18 cdown joined #git
22:18 jstein_ joined #git
22:18 Guest17292 joined #git
22:20 johnny56 joined #git
22:23 ToBeCloud joined #git
22:31 ocbtec joined #git
22:31 Levex joined #git
22:34 BackEndCoder joined #git
22:37 NeverDie joined #git
22:38 j7k6 joined #git
22:39 madduck is it possible to import patches with git-am but tell Git about a change in path? The patches fix up /foobar, but that's at /subdir/foobar now. Of course, I could move it, then apply, then move it back, but I'd rather not…
22:39 madduck heh, I suppose I could try sed the patch files…
22:41 pietv joined #git
22:41 m0viefreak madduck: git am -p ?
22:42 madduck no, I tried that. I'd need -p-1 ;)
22:42 m0viefreak ah
22:42 a3Dman joined #git
22:43 madduck sed worked
22:44 m0viefreak --directory then
22:44 jimi_ joined #git
22:44 m0viefreak git am --directory=subdir
22:44 nowhereman joined #git
22:44 madduck aah. how could I miss that???
22:44 madduck sorry.
22:52 BackEndCoder joined #git
22:52 F0rTh3J3st joined #git
22:53 roelmonnens joined #git
22:54 [ADN] joined #git
22:56 chipotle joined #git
22:56 hobodave joined #git
22:56 johnmilton joined #git
23:01 wizardedit joined #git
23:01 wizardedit joined #git
23:01 fmeerkoetter joined #git
23:01 j7k6 joined #git
23:02 nowhereman joined #git
23:06 gugah joined #git
23:09 hahuang61 joined #git
23:09 ShalokShalom_ joined #git
23:11 vdv joined #git
23:13 kyan joined #git
23:14 Dougie187 left #git
23:17 rorro joined #git
23:17 Lynxium joined #git
23:20 Levex joined #git
23:22 dsdeiz joined #git
23:23 spacelord_ joined #git
23:24 ikelso joined #git
23:27 mischat joined #git
23:28 zezba9000 I have custom git credential helper but when git called the "get" command on it and I respond via stdout... I get authentication errors
23:29 pur3eval joined #git
23:29 zezba9000 My helper is written in C#... its very simple and just respond everything including "username=***" and password=*** etc
23:29 rivarun joined #git
23:30 zezba9000 as soon I I pass it these parms, it gets called again with "erase"... which in turn passes back all the stuff I passed git
23:30 zezba9000 is there some char set I need to be using?
23:30 spacelord_ joined #git
23:30 zezba9000 my username is an email with the @ symble. Could that be causing issues?
23:32 hahuang61 joined #git
23:33 hahuang65 joined #git
23:35 miczac joined #git
23:40 ozmage joined #git
23:41 madewokherd joined #git
23:42 Samual joined #git
23:43 mischat joined #git
23:45 overyander joined #git
23:47 DieguezZ joined #git
23:51 F0rTh3J3st joined #git
23:53 daivyk_ joined #git
23:54 sobersabre_ joined #git
23:54 Es0teric joined #git
23:59 dil3mm4 joined #git

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