IRCloggy #git 2020-01-21

Provider of IRC logs since 2005.

WARNING: As Freenode became unjoinable and lost all warnings in topics, we cannot log channels on Freenode anymore.

	2020-01-21
	← leeN left	00:00
guzzlefry	kind of?	00:00
rafasc	There must be something running git fetch.	00:00
guzzlefry	yes, me :P	00:00
	I'm ssh'ing in and doing it.	00:00
	← ttoocs left	00:01
guzzlefry	I got around it by setting ACL permissions, but I'd still like to know what's going on if possible.	00:01
	← omnireq left	00:02
	→ Codaraxis joined	00:03
	→ Dirak joined	00:03
	← g00s left	00:05
	← Dirak left	00:08
	← jamiejackson left	00:10
	→ Dirak joined	00:14
	← shabius_ left	00:19
	← mikecmpbll left	00:19
	→ shabius joined	00:24
	← oatmealraisin left	00:26
	→ ttoocs joined	00:28
	← ttoocs left	00:33
	← xcm left	00:34
	← duderonomy left	00:34
	→ xcm joined	00:34
	→ causasui joined	00:36
	→ fphilipe_ joined	00:38
	← CodeSlingerPaul left	00:40
	← dedowsdi left	00:41
	← mowcat left	00:41
	← Anthaas left	00:41
	← fphilipe_ left	00:42
	← paws_ left	00:44
	→ jimender2 joined	00:45
	→ budonyc joined	00:45
	→ Anthaas joined	00:46
	← budonyc left	00:53
	← gioyik left	00:54
	← lefne left	00:57
	→ lefne joined	00:59
	→ ttoocs joined	01:00
	← causasui left	01:00
	→ igemnace joined	01:01
	→ omnireq joined	01:02
	→ dedowsdi joined	01:04
	← ttoocs left	01:05
	← ravndal left	01:06
	← royal_screwup21 left	01:06
	← xcm left	01:06
	→ xcm joined	01:07
	→ Goplat joined	01:15
	→ oatmealraisin joined	01:15
	→ Decobus joined	01:15
	← igemnace left	01:16
	← jstein left	01:16
	→ ttoocs joined	01:32
	← jetchisel left	01:33
	→ rominron_ joined	01:33
	→ jetchisel joined	01:33
	← rominronin left	01:36
	← ketas left	01:36
	← ttoocs left	01:38
	→ ketas joined	01:39
	→ igemnace joined	01:40
	→ ycon_ joined	01:43
ycon_	Im so baffled by this- if my git modules used https, then where I clone it will have to use https too? Does that sounds right?	01:44
	→ fairuz joined	01:45
	← fairuz left	01:46
	← guzzlefry left	01:46
	← Lucas_Gray left	01:49
Ori_B	yes, git submodules are effectively just cloned using 'git clone' from the url in $REPO/.gitmodules	01:55
	← CryptoDavid left	01:56
	← ycon_ left	01:58
	← rafasc left	02:02
	→ fairuz joined	02:03
	→ ttoocs joined	02:04
	→ florekm joined	02:06
	← statusf90 left	02:07
	← sveinse left	02:07
	← ttoocs left	02:09
	← sinhue left	02:10
	← Codaraxis left	02:17
	← beastDiv_ left	02:17
	← _Warl0ck left	02:18
	← dumb_moth left	02:19
	→ beastDiv_ joined	02:26
	→ tpewnag joined	02:34
	← tpewnag left	02:34
	→ ttoocs joined	02:36
	← Brainium left	02:37
	← finalbeta left	02:38
	← RoseBus left	02:40
	→ Thorn joined	02:40
	← justanotheruser left	02:41
	← ttoocs left	02:42
	← jetchisel left	02:43
	→ jetchisel joined	02:44
	→ _Warl0ck joined	02:46
	← yonder left	02:46
	← chandoo left	02:47
	→ finalbeta joined	02:48
	← _Warl0ck left	02:48
	→ justanotheruser joined	02:48
	→ dumb_moth joined	02:49
	→ _Warl0ck joined	02:50
	→ causasui joined	02:51
	← fattredd left	02:51
	← veegee left	02:51
	→ tpewnag joined	02:52
	← CCR5-D32 left	02:52
	→ ravndal joined	02:53
	← F0rTex left	02:55
	← dumb_moth left	02:56
	→ F0rTex joined	02:56
	← _Warl0ck left	02:56
ShapeShifter499	hi	02:57
	How can I merge two repositories together and all the branches they have?	02:58
	jaziz1 → jaziz	02:58
	← dartmouthed left	03:02
	→ ycon_ joined	03:06
ycon_	Ori_B: oh thanks. sorry was disconnected	03:06
	← kantlivelong left	03:07
	→ kantlive- joined	03:08
ycon_	Super wiered- even though my submodule folders show ssh method, as my url github asks for my username and pass. Could something be wrong here?	03:08
	kantlive- → kantlivelong	03:08
	→ ttoocs joined	03:09
	← bvdw left	03:09
	→ bvdw joined	03:10
	→ omnireq_ joined	03:13
	← ttoocs left	03:14
	← howdoi left	03:14
	← omnireq left	03:15
	← causasui left	03:17
	→ ttoocs joined	03:24
	lagothrix → Guest98510	03:25
	← Guest98510 left	03:25
	→ lagothrix joined	03:25
	← beastDiv_ left	03:26
ycon_	Hi all. new to submodules here. What is the way to recursively git fetch all directories?	03:30
	as in , check all submodules for latest changes	03:31
igemnace	ycon_: no dedicated command for it. check out git submodule foreach in the help	03:36
ycon_	I'm also confused with git fetch. If I have done --force-with-lease, does that mean git fetch won't think there are new commits?	03:36
igemnace	it'll still fetch the new commits	03:45
	← ycon_ left	03:45
	→ beastDiv_ joined	03:46
igemnace	you just won't immediately be able to git merge the upstream of the branch you did --force-with-lease in on some other machine	03:46
	...weird sentence but the idea is push --force-with-lease will affect merging local and remote branches, not fetching remote branches	03:48
	→ fattredd joined	03:49
	→ Codaraxis joined	03:52
davidfetter	is there some resources that helps decide among git submodule, git subtree, etc.?	03:54
	→ sauvin joined	03:54
igemnace	davidfetter: good question. i've only been able to stitch together that information from various sources in the past. are you trying to decide for a specific use case? or just collecting resources in general?	03:57
	← Cabanossi left	03:57
	→ ycon_ joined	03:58
ycon_	Hi all, im trying to fetch updates from a repo on my server- where it should override all conflicts/not merge. However, git fetch does not work.	03:58
	→ duderonomy joined	03:58
davidfetter	igemnace, it's for a possible dependency for PostgreSQL, so we need to be able to control pretty carefully what exact commit we're using with each branch	03:58
ycon_	This worked: "git fetch --all;git reset --hard origin/<branch_name>". But is there a way to write that, without having to specify the branch?	03:58
igemnace	davidfetter: is it a dependency you're going to maintain yourselves? or do you expect the dependency's maintainers to work separately?	04:00
davidfetter	...and given the longevity of the PostgreSQL project, we've found ourselves the maintainers of several "external" packages already. Spencer's regex, e.g.	04:00
	we're maintaining the dependencies. others might (from our perspective) send patches to sub-assemblies, which we'd need to review, possibly modify, etc.	04:01
igemnace	ah. the idea is that for subtrees, the entire history of the subtree will be included in the superproject as well. for submodules, the submodule's history will be separate, and the superproject will only keep reference information (remote URL, checked-out hash)	04:01
davidfetter	that latter sounds more attractive, as our history goes back quite a ways, and we'd rather not cloud it with Other People's Pushes	04:02
	← beastDiv_ left	04:02
	davidfetter would ordinarily be down with OPP, but in this case, not so much	04:02
igemnace	right, and apart from that, with subtrees you should be ready to have commits relating only to the subtree being part of the whole superproject history	04:03
	even the ones made by "first-party" maintainers	04:05
davidfetter	that sounds like it'd make a lot more sense than shuffling in others' whole history. are there systems I should consider other than those two?	04:05
	← barteks2x left	04:05
igemnace	for git, i'm only familiar with those two. i'm not familiar with other techniques	04:05
ycon_	Like this seems to work: git fetch --all;git reset --hard origin/$(git rev-parse --abbrev-ref HEAD)	04:06
	davidfetter has seen reference to submodule, subtree, subrepository, and Composer so far	04:06
igemnace	yep, outside git, there might already be dependency management systems for the language/stack you're working with. composer is one of those	04:07
	← dedowsdi left	04:08
	→ Cabanossi joined	04:09
igemnace	ycon_: if your branch already has an upstream set up, you can do git reset --hard @{u}	04:09
	→ dedowsdi joined	04:09
ycon_	igemnace: oh thanks. That's pretty neat. Basically if my branch is remote, then it will always have an upstream yeah?	04:10
	← fairuz left	04:10
igemnace	depends on how the branch was created. if it was created by doing git checkout branch-name (which will automatically set up a local branch with an upstream if no local branch exists) or by more explicit commands such as git checkout --track, then yes it will have an upstream	04:12
ycon_	Cool. Very good	04:13
	→ nickodd joined	04:16
	← tpewnag left	04:18
	← Dirak left	04:20
	← g4570n left	04:21
	← oatmealraisin left	04:23
	← netz left	04:25
	← dedowsdi left	04:26
	→ dedowsdi joined	04:27
	← crose left	04:36
	← hussam left	04:36
	← fling left	04:42
	→ ferdna joined	04:45
	→ Dirak joined	04:53
	← bukkitgerman left	04:53
	→ bukkitgerman joined	04:53
	← dedowsdi left	04:57
	→ dedowsdi joined	04:58
	← jcbitter left	05:01
ycon_	Git submodules is really confusing. Can someone give me clarity on how to get the lastest changes from remove submodules?	05:01
	git submodule update --init --recursive does not do anything	05:01
	→ pks_ joined	05:01
	← pks left	05:03
	pks_ → pks	05:03
	→ fairuz joined	05:06
	→ orbyt_v7 joined	05:09
	← dedowsdi left	05:09
	→ dedowsdi joined	05:11
	← ycon_ left	05:15
	← orbyt_v7 left	05:19
	← Dirak left	05:25
	→ Dirak joined	05:32
	→ _Warl0ck joined	05:35
	→ dumb_moth joined	05:36
	→ xco joined	05:43
	← ferdna left	05:47
	← bukkitgerman left	06:00
	→ bukkitgerman joined	06:01
	→ beastDiv_ joined	06:02
	→ ycon_ joined	06:10
	← R2robot left	06:12
	← ycon_ left	06:13
	← fairuz left	06:19
	← zrhoffman left	06:19
	→ __salt joined	06:19
	→ zrhoffman joined	06:20
	← _salt left	06:20
	← orbyt_ left	06:20
	← zrhoffman left	06:21
	← jetchisel left	06:22
	→ zrhoffman joined	06:23
	→ fairuz joined	06:25
	→ Mattiaslndstrm joined	06:26
	→ jetchisel joined	06:27
	← jetchisel left	06:28
	→ jetchisel joined	06:28
	→ ycon_ joined	06:29
ycon_	Hi all, is there I way I can do "git reset --hard @{u}" conditionally- only adding the @{u} if it actually points to a branch?	06:29
	→ prit joined	06:29
ycon_	Will this acehive it?	06:31
mud	You should be able to use `git rev-parse --verify @{u}` to check if points to something, then do a very short bash script based on that?	06:33
	→ pandakekok9 joined	06:34
	→ trittweiler joined	06:35
	gitinfo set mode: +v	06:36
	← FH_thecat left	06:36
	→ Codaraxis_ joined	06:40
trittweiler	What's the best way to "categorize" files in a repository? I.e. I would like to find out whether files are part of the repository, untracked, or ignored? It doesn't seem like `git ls-files` can do that.	06:40
	→ fphilipe_ joined	06:40
	← jetchisel left	06:41
	→ netz joined	06:41
	← mat001 left	06:41
	← Codaraxis left	06:43
osse	trittweiler: it can list each category separately	06:44
	→ jetchisel joined	06:45
osse	a small script could produce a long merged list with an indication of which list it came from	06:45
	http://sprunge.us/z6H07G - something like this maybe?	06:50
	← fairuz left	06:51
osse	the command with --ignored also needs --others	06:52
	← Trieste left	06:57
	← xco left	06:57
	→ Trieste joined	06:58
	← kjartan left	06:59
	→ kjartan joined	07:00
	→ xco joined	07:12
	← theorangeone left	07:18
	→ Noti joined	07:19
	→ theorangeone joined	07:20
	→ sQVe joined	07:21
	← ycon_ left	07:27
	→ BeerLover joined	07:28
	← nickodd left	07:28
	← duderonomy left	07:32
	→ chele joined	07:32
	← Goplat left	07:33
	→ im0nde_ joined	07:35
trittweiler	Cheers.	07:35
	→ ijash joined	07:35
	← im0nde left	07:36
	→ shady_traveller joined	07:36
	→ duderonomy joined	07:36
	← theoceaniscool left	07:39
	← beastDiv_ left	07:41
	← thiago left	07:44
	→ reprazent joined	07:44
	→ gxt joined	07:50
	← gxt left	07:50
	→ gxt joined	07:53
	← lankanmon left	07:53
	→ robamman2020 joined	07:54
	→ beastDiv_ joined	07:54
	→ rfuentess joined	07:55
	pandakekok9 → CommunistMonotyp	07:55
	CommunistMonotyp → pandakekok9	07:56
	→ lankanmon joined	07:56
	→ graungaard joined	07:58
	→ graungaard83 joined	07:58
	← graungaard83 left	07:58
	→ glosoli joined	08:02
	← glosoli left	08:02
	→ glosoli joined	08:02
	← BonusPlay left	08:08
	→ BonusPlay joined	08:08
	→ oxymoron93 joined	08:08
	← robamman2020 left	08:17
	← dumb_moth left	08:18
	← _Warl0ck left	08:18
	→ dumb_moth joined	08:21
	→ _Warl0ck joined	08:22
	→ dartmouthed joined	08:23
	← xco left	08:23
	→ dre_ joined	08:31
	→ mobidrop joined	08:34
	→ sbeyer joined	08:36
	→ xco joined	08:40
	→ enoq joined	08:42
	← Helenah left	08:44
	→ Helenah joined	08:46
	→ Lope joined	08:47
Lope	if I commit to the wrong branch and have already pushed it to bitbucket, is there a way to roll it back?	08:48
	osse, ^ :)	08:48
	← dre_ left	08:49
oxymoron93	sure, it requires force push	08:49
igemnace	git revert is the safe way	08:49
Lope	igemnace, `git revert $id; git push ...` ?	08:50
oxymoron93	I assume roll back really means that, but yeah, if you weren't aware of git revert, check it out if it is applicable for your case: man git revert Lope	08:50
gitinfo	the git-revert manpage is available at https://gitirc.eu/git-revert.html	08:50
igemnace	yep. that'll add a new commit (your history will move forward), but whose tree is a "roll back" (your files will go back to the state before the reverted commit)	08:51
oxymoron93	it will create another commit, which cancels out changes made by commit used in command	08:51
Lope	are you guys familiar with `git undo` ?	08:52
oxymoron93	resulting in no history rewrite, and you can just push it	08:52
Lope	it's part of some git scripts available in debian.	08:52
oxymoron93	I am not	08:52
igemnace	neither am i	08:52
oxymoron93	from what I see, it is basically `git reset` :D	08:53
	which is not what `git revert` does	08:53
Lope	it's part of the git-extras package.	08:53
	→ floppydh joined	08:54
oxymoron93	yeah, found the manual, is that what you want to do with this latest commit?	08:54
Lope	I use `git undo` often	08:54
	but I've pushed to bitbucket already	08:54
	so how can I undo there?	08:54
oxymoron93	only with force push , but check this out !rewrite	08:55
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	08:55
Lope	this is the git undo command: https://github.com/tj/git-extras/blob/master/bin/git-undo	08:56
oxymoron93	well, even better, now I know it is `git reset` :D	08:56
	← tsujp left	08:56
	→ yonder joined	08:58
	← dionysus69 left	09:00
	← bigfondue left	09:00
	← Kaisyu7 left	09:02
gitter1234	Hey! How come when I rebase to an old commit I can't see the files that are staged for that commit?	09:06
	← bvdw left	09:09
	→ bvdw joined	09:10
	← ijash left	09:10
oxymoron93	elaborate more, rebase is done with clean working directory (no staged, or modified files) and files staged for the existing commit isn't clear to me either	09:11
gitter1234	When I do git commit --amend, then I can see the all the files that are gonna be part of that commit	09:12
	→ tonythomas joined	09:12
osse	gitter1234: have you marked a commit as "edit" ?	09:12
gitter1234	Yeah	09:13
	Makes it kinda hard to see what changes I'm supposed to make. I have to make a copy of the repo and then git show <the commit I'm rebasing> to get an overview.	09:13
osse	it's set up so that you're supposed to use git commit --amend	09:13
gitter1234	Yeah I know	09:13
osse	you can git reset --soft @~ and commit the usual way	09:13
	→ bigfondue joined	09:13
osse	make a copy?!?!	09:13
gitter1234	But I'd like to see all the files that are gonna be applied to the current commit when I do git status	09:14
osse	my suggesteed reset thing does that	09:14
	but remember to not use --amend when you're done	09:14
gitter1234	Cool, thanks	09:14
	Ah so that's why I fucked up my repo just now :-)	09:15
osse	also you could just "git show" when you're there. making a copy sounds silly	09:15
gitter1234	Ah yeah git show shows it	09:16
	Although I'd like just a list of the modified files. I'll try out your command, cheers	09:17
osse	git show --name-only	09:17
	git diff HEAD~ HEAD --name-onlu	09:17
	← igemnace left	09:17
gitter1234	Ohhh nice!	09:17
	Onemore thing if you don't mind	09:17
	When I do git commit --amend and get taken to the editor, how come the commit is still amended even though I exit the editor instead of saving it?	09:18
	Like, once I do git commit --amend, there's no turning back?	09:18
	→ fphilipe joined	09:18
osse	you can empty the editor	09:18
	or make it exist with a failing exit status	09:19
gitter1234	I see. Sounds good!	09:19
	← fphilipe_ left	09:21
canton7	gitter1234, or you can go ahead with the amend, and then undo it	09:22
gitter1234	I'm good :-)	09:24
	canton7: Are you a Spree/Solidus dev by any chance?	09:24
	Your nick seems familiar.	09:25
	→ YuGiOhJCJ joined	09:25
	→ tsujp joined	09:27
gitter1234	Half German half Turkish?	09:28
	Nevermind	09:28
	→ igemnace joined	09:30
	→ Kaisyu7 joined	09:32
	→ Syzygy joined	09:33
	← voker57 left	09:34
	→ feriman joined	09:36
	→ voker57 joined	09:37
	← graungaard left	09:38
	→ vitaminx joined	09:41
	→ jstein joined	09:42
	← xco left	09:49
	← sbeyer left	09:51
	→ sbeyer joined	09:53
	← yonder left	09:54
	← kleisli left	09:58
	← fphilipe left	09:58
	← Lope left	10:00
	← rfuentess left	10:03
	→ rfuentess joined	10:03
	← enoq left	10:04
	→ enoq joined	10:06
	← Dirak left	10:08
	← palooka-jock left	10:09
	→ dbacc joined	10:09
dbacc	hey, I had a forced update to some branch with git rebase. I'm the only user of the git repo and wanted to continue to work on that version on my second client. However, when I pull, I get a nasty merge conflict. How do I force getting the version from origin?	10:10
osse	don't pull.	10:11
	git reset --hard origin/master (or whatever the branch name is)	10:11
	← YuGiOhJCJ left	10:11
	→ xco joined	10:12
osse	(assuming you don't have any non-push commits for your second client)	10:12
dbacc	osse: that worked, thanks! And no, it just kept rechanging certain things	10:14
	→ libertyprime joined	10:19
	← sunri5e left	10:21
	→ sunri5e joined	10:25
	→ fphilipe joined	10:25
	→ charolastra joined	10:26
	→ statusf90 joined	10:29
	← fphilipe left	10:30
charolastra	i've got a customer that wants the web-deployment through git on a self hosted repo. the project is already composed of various submodules. would you copy everything, strip all of git from the directory tree and then commit/push it to that private repo or are ther bette ways?	10:30
	→ palooka-jock joined	10:30
	→ FH_thecat joined	10:31
	← BonusPlay left	10:37
	→ BonusPlay joined	10:37
	← pandakekok9 left	10:42
	← igemnace left	10:44
	→ figuinhas joined	10:47
	→ pandakekok9 joined	10:50
	← glosoli left	10:52
	→ barteks2x joined	10:58
	→ fphilipe joined	10:59
	→ igemnace joined	11:00
	→ glosoli joined	11:03
	← fphilipe left	11:04
	← glosoli left	11:06
	→ doublex__ joined	11:10
	← ledx left	11:11
	← AndrewPRS left	11:14
	← doublex_ left	11:14
	→ AndrewPRS joined	11:14
	← chele left	11:14
	← jimender2 left	11:14
	→ chele joined	11:14
	← beastDiv_ left	11:14
	← JanC left	11:15
	→ JanC joined	11:15
	→ glosoli joined	11:17
	← xco left	11:22
	→ CCR5-D32 joined	11:28
	→ Brainium joined	11:29
osse	!deploy	11:31
gitinfo	Git is not a deployment tool, but you can build one around it (in simple environments) or use it as an object store(for complex ones). Here are some options/ideas to get you started: http://gitolite.com/deploy.html	11:31
charolastra	thanks	11:38
	← statusf90 left	11:39
	→ torbjorn joined	11:40
	→ Lucas_Gray joined	11:40
	← Matej8 left	11:41
torbjorn	In a particular application I need a file to be checked out and not have its newlines modified. the rest of the files in the repo I don't care about, and the default handling is probably fine.	11:41
	well this is part of a test case. Do I need to just handle this problem in my test code, ie have it string replace on all newlines in the file and write them to a temp file, or can I tell git not to touch the newlines of this particular file only?	11:42
	ah .gitattributes perhaps	11:43
	→ troulouliou_dev joined	11:44
	→ catphish joined	11:45
	→ sickdyd joined	11:45
	← pandakekok9 left	11:46
osse	torbjorn: yeah. 'filename -text' > .gitattributes	11:46
sickdyd	hello! how can I create a github repo by command line?	11:46
osse	!hub	11:46
gitinfo	Hub is a wrapper around git which provides specific github features to the git cli. See https://github.com/github/hub for details.	11:46
osse	!spindle	11:46
gitinfo	[!git-spindle] https://github.com/seveas/git-spindle is a handy tool to use the features of GitHub (and GH Enterprise), GitLab, BitBucket and possibly more from the command line.	11:46
sickdyd	I have my project and I'd like to push it to github without having to go on the website	11:46
osse	those two probably do the trick	11:47
sickdyd	osse: cheers	11:47
pandem1	sickdyd: https://hub.github.com/	11:47
sickdyd	pandem1: thanks	11:47
catphish	i have a branch that i've reset back a few commits, and added some new commits, i would like to push it to a remote, is there a way i can create a new commit that points to my new tree, completely disregarding what's at the remote, but which has the remote as a parent, essentially i want a push -f of my branch but i don't want to actually force a push	11:48
cadabrax	Hey, what was this webpage to help you out with most git problems where it would ask questions to understand what's happening and then suggest commands to run to fix it?	11:48
torbjorn	osse: thanks!	11:49
bremner	cadabrax: google? (sorry, could not resist)	11:49
cadabrax	haha no	11:50
	it was a few years ago on this channel IIRC	11:50
	→ mikecmpbll joined	11:51
catphish	looks like "git merge -s ours origin/branch" is what i want	11:52
	← lefne left	11:52
	→ flaviodesousa joined	11:53
	→ prc94 joined	11:53
	← prc94 left	11:53
	→ Rashad joined	11:54
	← OnkelTem left	11:54
	→ cdown_ joined	11:55
	← cdown left	11:57
	cdown_ → cdown	11:57
	→ statusf90 joined	12:00
	→ royal_screwup21 joined	12:00
	→ cd joined	12:01
	→ emihir0 joined	12:02
	→ oatmealraisin joined	12:02
	← royal_screwup21 left	12:04
	→ thevishy joined	12:04
	→ royal_screwup21 joined	12:10
	← milkt_ left	12:13
oxymoron93	cadabrax is this it: https://sethrobertson.github.io/GitFixUm/fixup.html	12:13
	← Lucas_Gray left	12:13
cadabrax	oxymoron93: yes, thank you	12:15
	← Rashad left	12:15
oxymoron93	it is under _fixup if you ever need to recover it	12:16
cadabrax	good to know, thank you	12:16
	→ milkt joined	12:16
	→ Lucas_Gray joined	12:17
oxymoron93	tho I never read it, it seems like a nice project this Seth did	12:18
gitter1234	man i love this feeling... when you're done with your app and can finally lower your shoulders	12:19
	and actually enjoy rebasing going from the initial commit to the last, cleaning up things as you go :)	12:20
	← sbeyer left	12:20
	→ fphilipe joined	12:26
	→ mowcat joined	12:27
	→ lucasb joined	12:28
	→ haiiokarin joined	12:29
	← enoq left	12:32
	→ enoq joined	12:36
trittweiler	I have a file which was deleted locally. Is there a way to find out when that file was deleted?	12:40
oxymoron93	git diff --diff-filter=D -- file something like that	12:42
	git log	12:42
	pardone me	12:42
	yeah, seems like I haven't misspelled option: https://git-scm.com/docs/git-log#Documentation/git-log.txt---diff-filterACDMRTUXB82308203	12:43
trittweiler	The file is only deleted locally, not as part of a commit. `git status` lists it as "deleted"	12:45
	an unstaged delete	12:45
	It's a silly question, thinking about it. There is no way for git to have a track of this	12:46
	→ aw1 joined	12:47
oxymoron93	sure, if just deleted, git status will give you the info :)	12:47
_ikke_	trittweiler: correct, git does not know when it was deleted	12:48
oxymoron93	just recently	12:48
_ikke_	you could delete it, and run git status a year later	12:48
	← catphish left	12:55
	← sybariten left	12:56
	→ azwieg103 joined	13:02
	← thevishy left	13:06
	→ R2robot joined	13:10
	→ TomyWork joined	13:13
	← johnny56 left	13:18
	← royal_screwup21 left	13:20
	→ clime joined	13:21
	→ crose joined	13:22
	← ShapeShifter499 left	13:23
	← glosoli left	13:23
	← BeerLover left	13:24
	→ glosoli joined	13:25
	→ johnny56 joined	13:26
	→ milkt_ joined	13:30
	→ chrisml` joined	13:30
	← chrisml left	13:30
	← milkt left	13:32
	← clemons1 left	13:32
	← aw1 left	13:32
	← feriman left	13:33
	→ clemons1 joined	13:33
	→ feriman joined	13:35
	→ royal_screwup21 joined	13:36
	→ CryptoDavid joined	13:42
	← Lucas_Gray left	13:42
	→ Paradisee joined	13:43
	→ cfoch joined	13:46
	← royal_screwup21 left	13:52
	→ fercell joined	13:53
	← fercell left	13:58
	← hendry1 left	14:04
	→ sidetrack joined	14:05
Tobbi_	So, apparently, I changed my default push target using the command git push -u azure --all. Assuming I want to set it back to origin, do I simply do git push -u origin --all	14:11
	?	14:11
	Or can I destroy stuff?	14:11
	I mean, will that destroy stuff?	14:12
	← kjartan left	14:16
	→ crater2160 joined	14:16
	← crater2160 left	14:17
	← crater2150 left	14:18
	→ crater2150 joined	14:18
	→ kjartan joined	14:18
	→ causasui joined	14:20
_ikke_	Tobbi_: git branch --set-upstream-to	14:24
Tobbi_	_ikke_: Can I update multiple branches like that?	14:24
_ikke_	Not in one command	14:25
	← justanotheruser left	14:25
	← glosoli left	14:25
	← dedowsdi left	14:25
Tobbi_	Hrm...so that command above git push -u origin --all won't re-set it to origin? (I can't test because origin is currently not available)	14:25
	→ glosoli joined	14:27
	← glosoli left	14:27
	→ xco joined	14:31
	← oxymoron93 left	14:34
	→ g4570n joined	14:34
	→ justanotheruser joined	14:37
	→ SteffanW joined	14:39
	← Noti left	14:40
	→ dionysus69 joined	14:43
	← enoq left	14:44
	← oatmealraisin left	14:44
	← seraphire left	14:44
	← duderonomy left	14:50
	→ thiago joined	14:53
	← donofrio left	14:53
	← shady_traveller left	14:55
	→ CodeSlingerPaul joined	14:56
	→ glosoli joined	15:00
	→ Noti joined	15:01
	→ Dirak joined	15:01
	← statusf90 left	15:01
	← SteffanW left	15:05
	← Dirak left	15:06
	← bvdw left	15:09
	→ bvdw joined	15:10
	← jaziz left	15:10
	← im0nde_ left	15:10
	← Anthaas left	15:10
	← comptroller left	15:11
	→ Anthaas joined	15:11
	← g4570n left	15:11
	→ zher joined	15:15
	← ttoocs left	15:15
	→ hussam joined	15:17
henrebotha[m]	Curious about how Git stores linked working trees. I see .git/worktrees/my-branch has a huge file called index	15:20
	Is that file like a patch on top of the main working tree? Or what's happening there?	15:21
	← sQVe left	15:22
	→ Dirak joined	15:22
	→ comptroller joined	15:22
	← clime left	15:23
	→ g4570n joined	15:23
	← haiiokarin left	15:25
selckin	can have one entry for every file in the repo, shouldn't be 'huge'	15:25
	→ sperry joined	15:26
	← sickdyd left	15:27
	← deltam left	15:27
	← Dirak left	15:27
	← sauvin left	15:28
cadabrax	how could I build a list of people who committed to a repo up to a certain commit in the past? I'd like to know who worked on a repo up until a commit from a year ago, but not include people who have worked on it later.	15:28
	→ thevishy joined	15:29
	← xSke left	15:29
selckin	git log --pretty=format:'%an' --abbrev-commit --before='Fri Jan 17 08:05:24 2015 +0100' \| sort -u	15:30
	← omnireq_ left	15:34
	← trittweiler left	15:39
	→ sauvin joined	15:40
	→ xalen7 joined	15:41
	→ truthseeker joined	15:41
	→ roadie joined	15:42
	→ Lucas_Gray joined	15:42
	← xalen7 left	15:43
	→ bolovanos joined	15:44
	→ enoq joined	15:44
	← bolovanos left	15:45
	→ bolovanos joined	15:46
	← floppydh left	15:47
	← comptroller left	15:48
	→ cousteau joined	15:54
cousteau	Hi!	15:54
	← sperry left	15:54
osse	henrebotha[m]: that's just like .git/index, except for the other working tree	15:55
cousteau	Version control of MS Word documents? (any rich text document that can be opened in MS Word, not necessarily .docx)	15:55
henrebotha[m]	Ok so I guess I need to know how .git/index works then! Lol	15:55
cousteau	I was wondering if git can deal with RTF, but I'm not sure it's texty enough	15:56
selckin	its just a binary file to git	15:56
	← is_null left	15:57
cousteau	Specifically I'm interested in whether there is a way to merge Word documents, because Word's merging tool sucks (and I can prove it, it's not a personal opinion)	15:57
osse	henrebotha[m]: the index is the "middle man" between commits and the actual files on disk. it's used for several purposes, most notably keeping track of what you've git add'ed	15:57
cousteau	Basically it lacks the three-way merge features a proper version control system has	15:57
selckin	cousteau: no	15:58
cousteau	:(	15:58
	RTF is a text format btw. But I don't know if it's the kind of text format that works fine with git	15:59
	← thiago left	16:01
	→ durham joined	16:02
osse	I say just commit it and deal with it if it becomes a problem	16:03
henrebotha[m]	cousteau: this article covers some basic techniques https://blog.martinfenner.org/2014/08/25/using-microsoft-word-with-git/	16:04
	→ comptroller joined	16:04
henrebotha[m]	Git allows you to shell out to external tools for various purposes such as diffing	16:04
	It may be possible to use those to get a reasonable workflow	16:05
	→ prepend joined	16:06
	→ duderonomy joined	16:06
	← gordonfish left	16:07
	→ gordonfish joined	16:07
cousteau	I don't know how reliable pandoc is for doc diffing	16:07
	→ Anon34623 joined	16:08
	← Noti left	16:09
sdgathman	90% of the problems people have with Word or LibreOffice are caused by using explicit formatting instead of styles.	16:10
	Once that finally gets drilled into their heads, the light goes on, and they realize that WYSINWYG is not such a bad idea. Latex is maybe too complex, but markdown is simple to learn.	16:11
	→ rafasc joined	16:12
	→ jackedin joined	16:13
	← chele left	16:13
	← _Warl0ck left	16:17
	← dumb_moth left	16:17
	→ omnireq joined	16:18
	→ _Warl0ck joined	16:19
	→ dumb_moth joined	16:20
	→ oxymoron93 joined	16:20
	← dumb_moth left	16:20
	← _Warl0ck left	16:20
	← neckutrek left	16:24
	← thurstylark left	16:25
charolastra	is there no way to treat a submodule inside a master repository as just files? for the containing repo the git of the contained plugins doesn't matter	16:25
	← gxt left	16:26
	← zher left	16:27
	← mobidrop left	16:30
	← durham left	16:31
	← vitaminx left	16:32
	← tsprlng left	16:35
	→ is_null joined	16:35
	→ tsprlng joined	16:36
	→ fission6 joined	16:38
	← thevishy left	16:38
	→ Cyrano joined	16:43
henrebotha[m]	I mean making it a submodule is explicitly saying you want to treat it as a Git repo, not just files	16:43
	← rominron_ left	16:43
	→ fling joined	16:43
	→ thevishy joined	16:43
	← omnireq left	16:46
	→ omnireq joined	16:47
charolastra	well, i just called it a submodule	16:47
osse	sdgathman: i find using styles is hard too :( am I a moron?	16:47
	→ beastDivision joined	16:47
	← beastDivision left	16:47
	→ Silicium joined	16:48
Silicium	hi there	16:48
	→ greggerz joined	16:49
Silicium	when i do a git rebase to take changes from one branch into another, git will not change the commit ID. if i do it using a github pullrequest and rebase&merge, it changes them, why?	16:49
	→ durham joined	16:50
	← Joel left	16:50
	← fission6 left	16:51
	← rfuentess left	16:53
	→ fission6 joined	16:54
	← Anthaas left	16:54
	← durham left	16:55
	← xcm left	16:56
	→ xcm joined	16:57
	→ Anthaas joined	16:57
	→ z1haze joined	16:59
	← thevishy left	16:59
z1haze	general process question here, but in our process, we use mostly a trunk based process, but we do have release branches in which features are merged into, and then when the release is stable, that gets merged into master.. I keep getting into arguments with someone who is rebasing their feature branches onto the release that its going into rather than master	17:00
	Am i wrong for this? My thought process is if you rebase your feature onto the release, and then the release gets something pulled out, then your feature could potentially have commits in there that shouldn't be	17:01
	→ durham joined	17:02
	← emihir0 left	17:02
	← durham left	17:02
osse	Silicium: I think you're wrong. git rebase almost always changes commit IDs	17:03
	unless you're already rebased and there's nothing to do	17:03
z1haze	that is not for me, right?	17:05
osse	no	17:05
	→ thevishy joined	17:05
z1haze	cool, thanks. just making sure because "i think you're wrong" came right after asking if im wrong :D	17:05
Silicium	osse: i dont know what i'am doing different but it doesnt in my case	17:06
	same timestamp same id	17:06
osse	that sounds very strange	17:06
	can you !repro	17:06
gitinfo	[!transcript] Please paste (using https://gist.github.com/ or similar) a transcript ( https://git.io/viMGr ) of your terminal session so we can see exactly what you see	17:06
Silicium	nope, since it is an internal one, sorry	17:07
	← is_null left	17:07
	← igemnace left	17:07
osse	i don't need to see the code. just the commands you run and some shas	17:08
	but anyway, if you can do git rev-parse HEAD and then git rebase whatever and then git rev-parse HEAD again, and it prints the same thing then you are a wizard	17:08
	→ thiago joined	17:08
	→ micro4lpha joined	17:09
	→ lefne joined	17:09
micro4lpha	what is the best practice to develop on a dev branch and then merge to a master branch (I need to modify some values after I merge to the master branch) and I don't want it to be an individual commit on the master branch. Basically to have two environmnets at the same time	17:09
	dev and prod	17:09
charolastra	you keep those individual settings out of git. if it's for web you can use environment variables	17:11
	→ prc94 joined	17:12
	← flaviodesousa left	17:13
	→ is_null joined	17:15
rafasc	z1haze: usually once you have a release-branch, only bugfixes get merged.	17:17
	← lefne left	17:17
	→ lefne joined	17:18
rafasc	z1haze: but creating feature branches on the oldest relevant commit is also common. Because it allows you to merge features independently.	17:18
z1haze	rafasc, thanks.. so in that case.. what do you do with a feature that still needs to be tested? does it not get merged into a release until it is tested?	17:18
micro4lpha	charolastra: the thing is those settings are java spring boot and they are needed for profile settings	17:18
Anon34623	Good $timeoftheday, everybody. I am fairly new to this channel, so excuse me for probably not formatting my message in an appropriate manner. In any case, I would like to ask a git-related question. I am using Arch and I've been trying to set up my own git server on nginx with gitweb (for web interface) lately, and it works great, except for the fact default git http backend doesn't really have any authentication options besides those on the webserver	17:19
	side, such as basic password-based authentication, which is much less than I need. My goal is to have an authentication level sufficient to provide account-based access to repos (not just git itself). I found that gitolite is a good way to achieve this, but I haven't been able to find any relevant documentation on integrating it into my git (nginx/git-http-backend/gitweb) setup, only some information in regards to Apache. Can anyone provide me with such a	17:19
	piece of documentation or with a sequence of actions I am required to take in order to make it work? Alternatively, if implementation of gitolite is impossible in such a setup, are there any similar alternatives for gitolite with similar authentication capabilities? My current nginx conf file looks like this: https://upaste.de/P88	17:19
z1haze	rafasc, that is my understanding too.. we have an issue though where a feature will get merged into the release, because QA tests the release branch, and they find too many issues so that feature has to get pulled out of the release, which requires we rebuild the relase	17:19
micro4lpha	like while it's on dev I want it to load spring.boot.profile=dev, when it's on prod i want it to load spring.boot.profile=aws	17:19
charolastra	micro4lpha: no idea about java but you need to keep it out of git	17:19
z1haze	so i guess we are just missing the piece of what do we do with a feature that we aren't sure is ready to go into the release, because it hasnt been given a green light by QA to get merged	17:19
	currently, we deploy the feature solo to a test instance and they test it there and give the thumbs up, if that happens they it gets merged to the release.. but the flow for that is clunky	17:20
rafasc	z1haze: depends, there are numerous way to make an workflow, for example in the git project the various feature branches are merged into something called pu (proposed updates) where various features are tested together to see if they may interfere between them.	17:20
	then each feature goes throw a promoting process until it reaches master. pu -> next -> master.	17:21
	← cd left	17:21
micro4lpha	charolastra: can't keep it out of git as it gets built on Amazon into a docker container and has to be inside the file	17:21
	← nutron left	17:22
micro4lpha	i'm sure someone has another solution	17:22
z1haze	so the pu branch isnt an eternal branch?	17:22
	like a 'develop' branch.. because i def don't want to go that way	17:22
rafasc	No, it gets rewritten all the time. But the git process is different, because you do not use branches. You submit patches via email, and all the branches are managed by the maintainer.	17:23
z1haze	currently we have likea hybrid of trunk based with releases.. maybe i should look into going full into trunk based process	17:23
	← thevishy left	17:23
z1haze	via email?	17:23
	oh geez	17:23
rafasc	git and the linux kernel are both managed via email.	17:25
dzho	Anon34623: I don't know about making such integrations. I do know that pagure builds atop gitolite to offer web access to repos.	17:25
	beyond that there are things like self-hosted gitlab, gitea, gogs, kallithea ...	17:26
	trac, I suppose	17:26
	← mikecmpbll left	17:27
	← Khisanth left	17:27
	→ mikecmpbll joined	17:28
	→ thevishy joined	17:29
z1haze	rafasc, so people are emailing their code changes to files and a single individual patches that into a branch?	17:29
	why wouldnt the dev just make a PR	17:29
rafasc	because it's the workflow they use. There isn't a public repo you can PR agaist. All public repos are just mirrors.	17:30
	← TomyWork left	17:30
	← thevishy left	17:31
rafasc	if your really want to contribute to git via github PRs, you can use https://gitgitgadget.github.io/ which is a bot that converts the PRs into mail submissions. It's managed by the git for windows maintainer.	17:31
	→ localhorse joined	17:31
rafasc	remember that github and git are different things. !git_vs_github	17:32
gitinfo	[!github_vs_git] github.com is not git; it's a hosting platform for git. There are alternative hosting offers (you can even host repositories yourself, or work on something on your own without any hosting whatsoever), and the range of features on the corresponding websites may differ. Git has little influence on what happens in github's web interface.	17:32
	→ mat001 joined	17:32
	← fphilipe left	17:33
	→ zher joined	17:34
	← mikecmpbll left	17:36
	→ thevishy joined	17:37
	→ Khisanth joined	17:40
	← oxymoron93 left	17:41
	← fission6 left	17:42
	→ kleisli joined	17:46
	← thevishy left	17:46
	→ nickodd joined	17:49
	← is_null left	17:50
	← Jackneill left	17:50
	← azwieg103 left	17:51
	→ azwieg103 joined	17:51
dzho	speaking of email-based workflows, this also reminds me that SourceHut is a thing	17:51
	→ fission6 joined	17:52
	→ Dirak joined	17:53
	→ orentanay joined	17:53
	← prc94 left	17:54
	← FFY00 left	17:58
	→ FFY00 joined	17:58
	← Dirak left	17:59
	→ sgen joined	18:05
	→ clime joined	18:05
	← truthseeker left	18:07
	→ prc94 joined	18:12
	→ sbeyer joined	18:12
	→ igemnace joined	18:14
	← gitter1234 left	18:15
	← Lucas_Gray left	18:15
	→ savolla joined	18:15
	← amcclure left	18:15
	← lungaro left	18:17
	→ amcclure joined	18:18
	→ xalen7 joined	18:19
	→ Phylock joined	18:19
	→ leeN joined	18:20
	← Paradisee left	18:24
	← igemnace left	18:29
	→ plexigras joined	18:31
	→ Dirak joined	18:41
	← dbacc left	18:41
	→ igemnace joined	18:42
FLHerne	z1haze: Note that git has various commands like `send-mail` and `mailsplit` that mostly automate that process	18:44
	← fission6 left	18:44
FLHerne	(most projects/devs with such a workflow have various shell-scripts as well)	18:45
	→ H2Q joined	18:46
	← vicfred left	18:46
	→ vicfred joined	18:47
	→ oatmealraisin joined	18:48
	← zher left	18:49
	→ fission6 joined	18:50
z1haze	im just trying to fight this git-flow argument	18:52
	we have a guy whos just dead set on changing to that	18:52
	most of us prefer trunk, but what we currently have isnt really either.. maybe a mashup	18:52
	→ erebel55 joined	18:53
erebel55	lo	18:53
	← enoq left	18:53
z1haze	definitely curious to hear the rooms opinion	18:53
_ikke_	those workflow models are mostly just that, models	18:54
erebel55	I canceled an in progress branch change and now I'm stuck with a ton of staged and unstaged files. I hadn't changed any files myself, so there is nothing to lose. But is there anyway to fix my repo without deleting and repulling everything?	18:54
_ikke_	I think no one hear would argue you have to adhere exactly to a certain workflow	18:54
rafasc	erebel55: yes, but depends on what you mean by "fix my repo".	18:54
orentanay	I'd like to setup 3 repos for a Magento 1.x multi store scenario where the base installation and package/theme are at the root and sub directories (this would be repo1), but then the packages for my other stores themes are split between 2 root level subfolders (app and skin), and within those sub folders are the files and sub folders for templates and styling that I want to be my seperate repos	18:55
	(repo2 /app/package1/ and /skin/package1/) and (repo3 /app/package2 and /skin/package2). I'm guessing I can just add the use the gitignore for repo1 to exclude the directories for what I want in repos 2&3, but how would I setup Repos 2&3? sorry for the long cryptic description. Thanks.	18:55
_ikke_	z1haze: You have to use what makes sense for your project	18:55
	← glosoli left	18:55
_ikke_	z1haze: git flow is mostly geared towards projects that work with (sometimes multiple) stable releases	18:55
rafasc	_ikke_: I think what z1haze is trying to figure out is if git-flow makes sense for their project.	18:55
erebel55	I just want to get back to a point where I don't have any staged/unstages files. So get back to a clean slate on either the branch I had checked out when I canceled or the branch I was switching to.	18:55
_ikke_	rafasc: yes, I was following up on that	18:56
z1haze	we USUALLY have a single running release, 2 week sprints	18:56
	there are occasions when the sprints are a bit longer, and MAY result in 2 release being in development at the same time, but this is very rare	18:56
rafasc	erebel55: if you use git status, it should tell you how to unstage changes.	18:57
_ikke_	z1haze: THe nice thing is that you can adopt your workflow to what makes sense at any given moment	18:57
	← fission6 left	18:57
	→ fphilipe joined	18:57
rafasc	erebel55: on recent versions of git, "git restore --staged :/" will unstage everything.	18:57
_ikke_	Have a single release, just work on master with feature branches and make a release when you're ready	18:57
	working on multiple releases? Maybe branch of for a certain release while still continue to develop for the next version	18:58
z1haze	_ikke_, that's exactly what we do now	18:58
	we branch features off master, then create a release when ready	18:58
	merge those features into the release	18:58
	and releae gets tested	18:58
_ikke_	z1haze: And I would argue you could still call that kind of git-flow (maybe a lighter version, which is sometimes also called github flow)	18:59
z1haze	the issue we have though is that features don't get tested (by qa) until they're merged into the release. and if qa finds a lot of issues, we end up having to back that feature out of the release, and its a pain in the ass	18:59
rafasc	erebel55: with older versions, git reset; will do it.	18:59
	→ glosoli joined	19:00
	→ zher joined	19:00
_ikke_	z1haze: Not sure how git-flow would solve this though. At some point, you have to wait until features get merged, or they will have to test individual feature branches	19:00
rafasc	what kind of qa is that? quality assurance only after the code is considered releases? o.O	19:00
	→ fission6 joined	19:01
_ikke_	Kind of a waterfall model	19:01
	you first do all the development, once everything is finished, you hand over to QA	19:01
	← nickodd left	19:02
_ikke_	z1haze: Is there no way for QA to test whatever is in master at any given time?	19:02
z1haze	heh, look at production i guess	19:02
	that is master	19:03
_ikke_	But isn't that already too late?	19:03
rafasc	once everything is finished should be different than considering it a release. Ideally, finish - handover to QA - get thumbs up from QA - promote it to release.	19:03
z1haze	we dont merge into master until qa says its good	19:03
_ikke_	Wouldn't you want to test things before they go to production?	19:03
z1haze	of course dude	19:03
	hold on a second, i think you've misunderstood me	19:03
_ikke_	yes, I think so	19:03
z1haze	day one of sprint is like this: release branch created off master, then all features are created off master.. developers submit PR to release branch as they finish, PR is merged into release	19:04
	the issue is we dont really have a process for QA to test individual or even combined features before they get into the release branch	19:04
	and if durign testing the release branch qa finds issues witha particular feature, we have to revert that feature from the release, and if there were a lot of commits/changes it can be a pain	19:04
_ikke_	z1haze: ok, git flow is more like this: everyone branches of from master for their feature branches. Once they are considered ready, they get merged to master	19:05
z1haze	uh?	19:05
	git flow is where everything is merged into the develop branch	19:05
_ikke_	Euh, right	19:05
	Which is your integration branch basically	19:05
z1haze	we don't have that.	19:06
	we have a release branch	19:06
	the release branch is what is tested	19:06
	i guess we need to just think through a new process	19:06
_ikke_	Yeah, but it sounds like you have a need for an integration branch	19:06
	where you can integrate new features which can be tested combined	19:06
z1haze	we were using the release branch as that branch	19:07
rafasc	And to consider common nomenclature. What you guys are doing are not what people commonly use release-branches for.	19:07
	→ munsterberg joined	19:07
z1haze	so the release was the integration tested branch	19:07
	but they want to be able to test these features independent before the release	19:07
	so i just need a some process to deploy features to an instance where they can be tested	19:08
munsterberg	for release branches, should merges be done directly to the release branch, or should they go through the devleopment branch first? I realize this question could probably use some context, so if you have any clarification questions, shoot	19:08
_ikke_	munsterberg: sounds similar to what z1haze is asking about	19:08
	z1haze: another interesting aproach might be what git itself is doing	19:09
munsterberg	sorry, just joined missed the previous conversation	19:09
rafasc	munsterberg: !logs in case you're interested	19:09
gitinfo	munsterberg: [!irclog] Public logs of #git are kept at: https://gitirc.eu/log	19:09
	← fission6 left	19:10
munsterberg	perfect thank you	19:10
_ikke_	z1haze: the maintainer basically merges topics / features that are somewhat polished to pu (proposed updates)	19:10
z1haze	i heard someone say this earlier	19:10
	but then he said sometyhing along the lines of its all done via email	19:10
_ikke_	But this branch is regularly rewritten	19:10
	yes, but that's just a detail	19:10
	← cousteau left	19:11
z1haze	so basically its a 'develop' branch that isnt eternal	19:11
_ikke_	It's basically a living integration branch	19:11
z1haze	yeah	19:11
	← Codaraxis_ left	19:11
z1haze	so its just the same as my release is now	19:11
	except the release can be sure it has only stuff that will stay	19:11
_ikke_	Well, the difference is that people don't build on it	19:11
z1haze	no one builds on the release either	19:11
	they shouldnt be anyway	19:11
_ikke_	If some feature should be dropped, the branch is rebuilt	19:12
z1haze	correct	19:12
	that's how our release branch works	19:12
rafasc	No, you said you revert.	19:12
	pu actually drops the commits out of the log. The branch is rewritten.	19:12
_ikke_	ok, but calling it a release branch can be confusing (people have expectations with that kind of name, like rafasc already mentioned earlier)	19:12
	← fphilipe left	19:13
z1haze	well we dont use revert commits actually, we usually just literally rebuild the release	19:13
	yeah i totally get it	19:13
	but in my case it really is a release branch	19:13
_ikke_	ok, but then there should be no issue merging feature so that they can be tested, right?	19:13
	→ fission6 joined	19:13
z1haze	because it is what ultimately will get merged into master	19:13
	← Khisanth left	19:14
z1haze	not if we just keep rebuilding the release	19:14
	but i dont want to have to do that	19:14
_ikke_	git itself has another intermediate branch	19:14
z1haze	its a pain, especially if there are a lot of commits	19:14
_ikke_	pu -> next -> master	19:14
z1haze	so feature -> pu ... ?	19:14
	is next the release?	19:14
_ikke_	next will be the next release :)	19:15
z1haze	and it goes from features into pu?	19:15
	← xalen7 left	19:15
rafasc	For me the pain is that you guys are calling 'release branch' to something that the community call something else.	19:15
z1haze	rafasc, is it not a release though, this branch, though it gets rewritten (if necessary) ultimately gets tagged and merged into master when it goes to production	19:15
	how is that not a release	19:15
_ikke_	yes, topic branch gets merged into pu after the first round of review	19:15
munsterberg	for my own clarification, what is 'pu'?	19:15
_ikke_	munsterberg: proposed updates	19:16
munsterberg	ok thank you	19:16
_ikke_	It's the first point of integration	19:16
	can be unstable	19:16
	features can be dropped again	19:16
	is regularly rewritten	19:16
z1haze	_ikke_, so how does the feature get a first round of review.. how can it be given a code review if there is nowhere to submit a PR to.. how does it get build to a test instance if it is just a feature	19:16
_ikke_	z1haze: in the case of git or in your case?	19:16
	→ Sasazuka joined	19:17
z1haze	git	19:17
	curious how they do it	19:17
	that's our sticking point	19:17
_ikke_	You post a patch series to the mailing list (which is basically similar to making a merge / pull request)	19:17
z1haze	its hard to qa a feature without first submitting the PR to release and then merging then testing	19:17
rafasc	z1haze: doesn't make sense to call it release imho. release is something that you release to the public. Once you merge that to master, You could fastforward the release branch to that point, but call that branch that gets rewritten "release" is outright confusing because that's not what the community expect from a release branch.	19:18
z1haze	we do fast forward master at that point	19:18
	and we tag that commit as a release commit	19:18
rafasc	I bet part of the confusion is people trying to apply things they read about "release branches" to your workflow.	19:18
_ikke_	in git-flow, a release branch is the branch you create right before a release	19:18
rafasc	But it's simply not the same thing. Hence fights...	19:18
	← lefne left	19:18
_ikke_	It mostly contains commits to prepare the release, and maybe last-minute fixes	19:18
z1haze	fair enough	19:18
	← jetchisel left	19:19
z1haze	so in your opinion, a release branch should be used for what? simply branching off the pu branch and .. ?	19:19
	sounds like in the end, once all dev is done, the PU branch and the RELEASE branch would be identical	19:19
	because they'd all have the same code	19:19
_ikke_	No, the pu branch is always updated with new features	19:20
z1haze	but when it gets reset.. where does it get reset to	19:20
	to master?	19:20
	you guys said that it goes pu -> next -> master	19:20
	→ jetchisel joined	19:21
z1haze	does it actually go feature -> pu, then feature -> next?	19:21
_ikke_	yes	19:21
z1haze	yes to which statement	19:21
_ikke_	the latter	19:21
z1haze	so pu is literally just a intermediate integration/testing grounds	19:22
_ikke_	exactly	19:22
	so it might be a good source for deploying somewhere to test around	19:22
z1haze	im trying to think of a logical jira workflow for that then	19:22
	and another question then -- i've always though it best to test individual features off master, so that this feature is the only thing in play.. if we are deploying the pu branch to test features, then we dont always know what the issue is	19:23
	← sgen left	19:23
_ikke_	A good CI system might help there	19:23
	but you need good automation to be able to actually deploy each individual feature	19:24
rafasc	there's also the personal jch, which is the attempt of the maintainer for the pu branch. But that's not part of the official flow, Junio just lets us peek.	19:24
z1haze	we have a basic one, well we use bitbucket's piplines and its decent	19:24
	← caveman left	19:24
z1haze	the ci isnt our issue, its the number of environments we have	19:24
	the platform we develop on is a saas	19:24
	so we require their system, and we are limited on environments	19:24
_ikke_	right, so you might need to adopt your workflow then as well	19:25
z1haze	we just deploy code to the 'instance' and it runs on their server	19:25
_ikke_	I don't think it's feasible to test each individual feature in isolation (would be a point of contention probably)	19:25
	← sauvin left	19:26
	→ Samian joined	19:26
Samian	how do I undo my last push to remote?	19:26
z1haze	reset and force push	19:26
Samian	I type git reset	19:26
	git force push ?	19:26
	huh?	19:26
rafasc	Samian: !rewrite	19:26
gitinfo	Samian: 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	19:26
z1haze	`git reset --hard "HEAD^"` i believe will reset it. use --soft if you want to keep your changes staged	19:27
_ikke_	that would just reset the last commit	19:27
z1haze	that's what hes asking	19:27
_ikke_	They could have pushed more then one commit	19:27
z1haze	ah, good point	19:27
	← jackedin left	19:27
rafasc	z1haze: please let they give more context, we don't even know if they know the consequences.	19:28
_ikke_	git push origin <branch>@{1}:<branch>	19:28
	git push origin --force-with-lease <branch>@{1}:<branch>	19:28
	but yes	19:28
	context is important	19:28
z1haze	right on	19:28
rafasc	--hard --force* is all things that may put them in a situation harder to recover than what they have now.	19:28
	→ def_jam joined	19:29
Samian	when I type git rev-parse HEAD	19:29
	I get 5098016bf0afa573b33244cfc5ef5d35bd86017d	19:29
	→ eb0t_ joined	19:29
Samian	when I type git branch -a, for master it shows remotes/origin/HEAD -> origin/master	19:30
	← eblip left	19:30
	← eb0t left	19:30
	← azwieg103 left	19:31
	→ lefne joined	19:31
Samian	fuck I hate fucking git	19:31
	→ Khisanth joined	19:31
	→ azwieg103 joined	19:31
Samian	I just want to undo the last push	19:31
	no one else has pulled from git since	19:31
	because it's just me and one other coworker	19:31
rafasc	the line remotes/origin/HEAD -> origin/master only tells you that the default branch of the remote is master.	19:31
Samian	okay	19:32
	← CryptoDavid left	19:33
Samian	so the last commit to remote is 5098016bf0afa573b33244cfc5ef5d35bd86017d	19:33
rafasc	Samian: if you know the consequences, git reset --hard <ID>; git push --force-with-lease; for example	19:33
Samian	dangit	19:33
rafasc	but if there's anything that diverged after that commit, you will potentially be in trouble.	19:33
Samian	rafasc I'm trying to make sure I get the id stuff right	19:33
	I type git log remotes/origin/HEAD	19:34
	but it shows local log it appears	19:34
rafasc	no, it shows the same thing as origin/master	19:34
	← figuinhas left	19:34
rafasc	which is your remote tracking branch.	19:34
Samian	okay yes nevermind	19:34
rafasc	!rtb	19:35
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>'	19:35
Samian	so I want the last commit on remote to be ad94ddd6095d908427217415c3198f77c6e27786	19:35
rafasc	git reset --hard ad94ddd6095d908427217415c3198f77c6e2778; git push --force-with-lease;	19:35
Samian	got it	19:35
rafasc	but your buddy, needs to reset as well. Not pull	19:36
_ikke_	or git push --force-with-lease origin ad94ddd6095d908427217415c3198f77c6e2778:master	19:36
	then you don't need to reset your current branch	19:36
rafasc	otherwise he will merge the thing again.	19:36
	← lefne left	19:36
Samian	darn it says I need git ForcePush permission	19:37
	aha I'm figuring it out	19:37
	how to get permission	19:37
	woohoo it worked	19:39
	← xcm left	19:40
	→ xcm joined	19:40
	→ oncall-pokemon joined	19:41
erebel55	@rafasc Thank you so much, "git restore --staged :/" did it and was surprisingly fast	19:41
rafasc	erebel55: next time, type git status and read the hints. It tells how to undo most operations that affect it.	19:42
	→ lungaro joined	19:42
	← Dirak left	19:42
erebel55	I did try git status. I guess I should have pipped the output to a file	19:42
rafasc	it wouldn't mention the :/ part, I think it says <file>	19:42
	but yea	19:42
erebel55	I think the hint is at the top and there were thousands of files, so I didn't see it	19:42
Samian	thank you guys!	19:43
rafasc	erebel55: ah, makes sense.	19:43
	→ caveman joined	19:43
erebel55	But that is a good tip. I'll pipe to a file next time I run into it :)	19:43
rafasc	erebel55: or to a pager. e.g. git --paginate status	19:45
	← Helenah left	19:45
erebel55	ah I didn't know about pagers	19:46
rafasc	You probably use it when you use git log.	19:46
	But I think status doesn't paginate by default.	19:47
	→ Dirak joined	19:47
	→ Helenah joined	19:47
	← foul_owl left	19:47
	← glosoli left	19:48
	→ gioyik joined	19:48
Samian	rafasc my buddy never pulled after I did the screw up push	19:51
	but he still needs to do a reset too?	19:52
	← g4570n left	19:52
Samian	he needs to type a reset command?	19:53
	← Mattiaslndstrm left	19:53
rafasc	if he never pulled or fetched (some tools fetch automatically) the "wrong" commit, they should be fine.	19:53
Samian	woohoo	19:53
	thanks rafasc	19:54
	← jetchisel left	19:54
	→ keynaris joined	19:55
	→ jetchisel joined	19:56
	← caveman left	19:56
rafasc	you're welcome.	19:56
	→ relipse joined	19:59
	→ g4570n joined	19:59
relipse	how do I add all untracked folders with /ja/ in their name	19:59
	← Trieste left	19:59
relipse	listed under Untracked files	19:59
	→ glosoli joined	20:02
	← orentanay left	20:02
rafasc	relipse: git add ://ja/	20:03
	note that git doesn't track directories, only files.	20:04
	→ Trieste joined	20:04
rafasc	So only things that have files at the end will be added. If you have a structure of empty dirs you'll need to add a placeholder, e.g. a .gitignore with an '*' pattern.	20:05
	← keynaris left	20:06
relipse	ok thanks what is : for	20:06
	← Dirak left	20:07
rafasc	the :/ is git pathspec to denote the root of the repository.	20:08
osse	TIL pathspec works for untracked files	20:09
rafasc	it also helps ensure that what you pass isn't substituted by the shell.	20:09
	osse: I think this might be new, because I had the impression it didn't.	20:10
	but tried and it worked.	20:10
	← kleisli left	20:11
osse	it's been like that at least since 2008 :O	20:12
	or maybe "pathspec" meant something else then :p	20:12
	→ skelletor joined	20:15
	→ caveman joined	20:17
	→ funhouse joined	20:17
funhouse	Hi there, just created a private repository, how do I add my folder and files to it in linux?	20:17
	← jess left	20:17
	→ jess joined	20:17
pandem1	you created a local git repo or a repo on lab, hub or some other online service?	20:19
	← glosoli left	20:20
	→ _Warl0ck joined	20:21
	→ dumb_moth joined	20:21
	← kreyren left	20:22
	→ kreyren joined	20:23
	→ owlscientist joined	20:24
	← fission6 left	20:24
	→ fphilipe joined	20:25
	→ Lucas_Gray joined	20:26
	← troulouliou_dev left	20:26
	→ jamiejackson joined	20:27
	← skelletor left	20:27
	→ goh joined	20:28
goh	What's the best tutorial for git conflict resolution?	20:28
_ikke_	!eek	20:28
gitinfo	[!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	20:28
pandem1	damn, the bot has everything	20:29
_ikke_	We make sure it does :)	20:29
	← Syzygy left	20:30
goh	If the key advice is don't panic, maybe the keyword shouldn't be "eek" :)	20:30
_ikke_	goh: the response to "eek" is don't panic :)	20:30
goh	Is vimdiff a decent merge tool? I'm working remotely via SSH, so need something terminaly	20:31
	← comptroller left	20:31
_ikke_	goh: The default colorscheme is awfull to me, so you might want to tweak that	20:31
	← zher left	20:31
_ikke_	goh: it's a decent tool, you just need to know how to work with it	20:32
goh	I'm just really confused by what REMOTE, LOCAL, BASE mean	20:32
	I want to be able to quickly jump to the points of conflict, see why that conflict occured, and resolve it.	20:32
_ikke_	Those are the 3 components of a 3-way merge	20:33
goh	heh, 3-way	20:33
	I guess the page in the link will explain?	20:33
_ikke_	https://www.rosipov.com/blog/use-vimdiff-as-git-mergetool/	20:33
	→ zher joined	20:33
	→ Codaraxis joined	20:34
_ikke_	]c and [c to jump between conflicts	20:34
goh	nice	20:34
	In this case I'm not sure why there's even a conflict	20:34
_ikke_	:diffget <pattern> to get the change from one of those files (:diffget RE to get REMOTE, etc)	20:34
goh	LOCAL and BASE appear to be identical	20:35
	→ kleisli joined	20:35
_ikke_	add / add to the same location is a conflict	20:35
	touching the context of of another change is a conflict	20:35
goh	I should probably understand what a 3-way merge actually means.	20:36
	→ veegee joined	20:36
_ikke_	progit does not seem to explain that	20:38
	https://www.drdobbs.com/tools/three-way-merging-a-look-under-the-hood/240164902	20:38
	→ skelletor joined	20:38
	← duderonomy left	20:39
	→ comptroller joined	20:39
	→ duderonomy joined	20:40
	← learning1 left	20:41
	← skelletor left	20:42
goh	thanks!	20:44
	← sidetrack left	20:45
	← munsterberg left	20:48
	← duderonomy left	20:48
	← cdown left	20:48
	→ learning1 joined	20:49
	→ AndyAndyBoBandy joined	20:49
	→ cdown joined	20:50
	→ royal_screwup21 joined	20:50
AndyAndyBoBandy	Can anyone recommend a good TUI for github issues and/or PRs? I currently use stephencelis/ghi which is ok, but last updated 2016. I also see jsmits/github-cli last updated 2012. I know hub, but that seems to just open the pages in a browser, whereas I'm looking for a TUI or CLI.	20:50
	← azwieg103 left	20:51
	→ azwieg103 joined	20:51
_ikke_	Not aware of anything	20:53
rafasc	AndyAndyBoBandy: git-bug has a github bridge.	20:55
goh	So a 3-way merge is just merging two conflicting changes based on the file state before these changes, right?	20:56
	(That's the "third" part)	20:56
_ikke_	goh: it's a way to find what is actually a conflict and what is change just made on one side	20:57
rafasc	goh: yes, all that happened in between is ignored.	20:57
_ikke_	If you just compare two files, you have no idea what part to pick from what side	20:57
	By comparing each side to a common ancestor, you know what part of a file changed, and what part remained the same	20:58
goh	So if I look at my conflict, and LOCAL and BASE are the same, that means REMOTE just introduced a change and that's it?	20:58
_ikke_	yes, but is LCOAL and BASE exactly the same?	20:58
	s/is/are	20:58
goh	BASE is the common ancestor, what are LOCAL and REMOTE?	20:59
rafasc	remote is what you are merging in, local is what you have.	20:59
_ikke_	LOCAL represents HEAD	21:00
goh	I'm cherry-picking, not merging in this case	21:00
rafasc	git checkout master; git merge feature; local=master remote=feature base=common ancestor between master and feature	21:00
goh	about about rebase/cp?	21:00
_ikke_	LOCAL is HEAD	21:00
goh	REMOTE is the commit I'm rebasing/cping?	21:01
AndyAndyBoBandy	rafasc: thanks, I'll look into git-bug	21:01
_ikke_	goh: yes	21:01
rafasc	remote is the thing being picked, base is the parent of the thing being picked.	21:01
goh	the darkness of ignorance is banished by the light of knowledge!	21:01
rafasc	(cherry-pick actually a kind of merge in the background)	21:01
	(so is each 'pick' from rebase)	21:01
goh	rafasc: interesting, I thought it was a one-commit rebase	21:01
	_ikke_: LOCAL and BASE are not exactly the same	21:02
	That part of the file is exactly the same.	21:02
	The part REMOTE changed	21:02
rafasc	goh: It's more like rebase is a multiple cherry-picks sequentially, but yeah.	21:03
goh	But LOCAL has changes over BASE	21:03
	→ fission6 joined	21:03
goh	I _suspect_ some trailing whitespace was removed from the area of REMOTE's change	21:04
	which caused the conflict.	21:04
_ikke_	yes, that would explain it	21:05
goh	The lesson is clear: never remove trailing whitespace ;)	21:05
rafasc	or never put them in.	21:05
goh	It's a joke	21:06
	I finally understand conflict resolutions, sorta	21:06
	← Shentino_ left	21:06
goh	so I'm no longer afraid of conflicts	21:06
	I confirmed it's the whitespace btw	21:06
_ikke_	:-)	21:06
rafasc	just wait until you hit one of those created by the recursive strategy.	21:06
	:D	21:06
goh	LOCAL removed some trailing whitespace from BASE, then REMOTE made a completely different change	21:06
_ikke_	rafasc: :D	21:07
goh	rafasc: I'll prepare my resume so I can quit my job immediately in that case.	21:07
_ikke_	rebase conflicts can be daunting as well	21:07
rafasc	kdiff3 helps me a lot.	21:08
goh	how is BASE identified?	21:08
_ikke_	man git merge-base	21:08
gitinfo	the git-merge-base manpage is available at https://gitirc.eu/git-merge-base.html	21:08
rafasc	for merges ^	21:08
goh	OK, so it's just the merge-base of LOCAL and REMOTE, got it	21:08
rafasc	for cherry-picks it is the parent of you are picking	21:08
goh	i.e. most recent ancestor	21:08
	right, because in CP, REMOTE isn't part of the graph	21:09
rafasc	if you pick a merge, you'll need to specify the parent you want with -m N	21:09
goh	"pick a merge"?	21:09
	→ Dirak joined	21:09
rafasc	you can cherry-pick a merge commit.	21:09
	← bvdw left	21:09
goh	Insanity.	21:09
	Did I tell you today I figured out the DAG is a Merkle Tree?!	21:10
	→ bvdw joined	21:10
goh	You seem less than impressed :\|	21:12
rafasc	:)	21:12
goh	I wonder what the next quantum leap in VC will be.	21:12
	Like the sort of leap Git made over SVN. What will be the next such leap over Git?	21:12
	← Codaraxis left	21:12
rafasc	brain upgrades	21:13
goh	anyway, since the DAG is a Merkle Tree, BRB adding "blockchain" to resume	21:15
rafasc	Not really, because usually when we say blockchain we also have the part about consensus.	21:15
	→ jcbitter joined	21:15
rafasc	git doesn't have that.	21:15
	Although, there was an implementation for jgit that I can't seem to find.	21:16
	← irrgit left	21:16
rafasc	https://www.eclipse.org/lists/jgit-dev/msg03073.html	21:16
_ikke_	"a digital database containing information (such as records of financial transactions) that can be simultaneously used and shared within a large decentralized, publicly accessible network"	21:16
	Sounds like git	21:16
	afaik, concensus is not a requirement for blockchains	21:17
	← erebel55 left	21:17
	← fphilipe left	21:17
	→ irrgit joined	21:18
rafasc	90% of the time people say blockchain they mean the consensus kind. But I guess you're technically correct.	21:19
	https://dictionary.cambridge.org/dictionary/english/blockchain	21:21
_ikke_	but a blockchain is not limited to cryptocurrencies	21:21
rafasc	directly mentions the crypto currency kind.	21:21
_ikke_	^	21:21
	wikipedia: "A blockchain, originally block chain, is a growing list of records, called blocks, that are linked using cryptography Each block contains a cryptographic hash of the previous block, a timestamp, and transaction data (generally represented as a Merkle tree). "	21:23
	"A blockchain, originally block chain, is a growing list of records, called blocks, that are linked using cryptography Each block contains a cryptographic hash of the previous block, a timestamp, and transaction data (generally represented as a Merkle tree). "	21:24
	wrong one	21:24
	"The first work on a cryptographically secured chain of blocks was described in 1991 by Stuart Haber and W. Scott Stornetta They wanted to implement a system where document timestamps could not be tampered with."	21:24
	so the origin is not even cryptocurrency	21:24
rafasc	I agree, just wanted to show that even a seemly reputable source refer defines 'blockchain' as the consensus one.	21:24
	It will be probably one of those things that the meaning either gets blurred or changes.	21:24
	just my opinion.	21:25
_ikke_	same with the term "crypto"	21:25
rafasc	hung up the phone.	21:25
	It's not the first time this happens.	21:26
	→ m0viefreak joined	21:26
_ikke_	ofcourse not	21:26
goh	blockchain inherently has nothing specific to do with crypto	21:26
	there's applications in contracts for example	21:26
rafasc	anyway, git ketch would be the consensus blockchain version of git blockchain.	21:27
goh	git kvetch	21:27
	a revolutionary consensus algorithm	21:27
	← Dirak left	21:27
	→ bolovanos_ joined	21:28
goh	who thought up all this stuff? like the DAG and how it should work	21:29
	→ Dirak joined	21:29
	← bolovanos left	21:32
	→ duderonomy joined	21:33
	← bolovanos_ left	21:36
	← royal_screwup21 left	21:37
	← eirc left	21:37
	→ eirc joined	21:37
	← duderonomy left	21:38
	← AndyAndyBoBandy left	21:40
	→ Inside joined	21:42
Inside	heyo! Anyone having problems with github right now?	21:42
_ikke_	https://www.githubstatus.com/	21:43
	for me it's working	21:43
	→ fstd_ joined	21:43
_ikke_	Though, it does feel sluggish	21:43
	← learning1 left	21:44
Inside	I'm trying to deploy using capistrano and the script fails on a 'git remote update --prune' command. I can run the command manually and about 50% of the time it fails with a "Could not read from remote repository" error	21:44
	← Sasazuka left	21:44
goh	Is Linus the one who designed the DAG and all the basic Git design?	21:44
canton7	goh, most likely, unless he got the idea from somewhere else	21:45
	→ Sasazuka joined	21:45
Inside	Are you asking if he invented DAGs?	21:45
canton7	I assumed, came up with the idea of using a DAG to represent history in a VCS	21:45
	← owlscientist left	21:45
	← xco left	21:46
	→ yonder joined	21:46
Inside	Don't all VCS use a DAG of some sort to represent history?	21:47
goh	Inside: Yeah, I'm asking if Linus invented graph theorey and modern computer science ;)	21:47
	Inside: nope.	21:47
	← fstd left	21:47
	fstd_ → fstd	21:47
goh	Git was the first one afaik.	21:47
	→ royal_screwup21 joined	21:47
Inside	This is beyond me :)	21:47
goh	SVN and CVS use a very simple linear changeset model.	21:47
	canton7: using a DAG as a Merkle Tree to represent the commit graph is pretty genius.	21:48
	→ owlscientist joined	21:49
	← owlscientist left	21:49
Inside	Hmmm. Anyone want to clone a repo from github and hammer 'git remote update --prune' a couple of times to comiserate with me? :)	21:53
	→ owlscientist joined	21:57
	← zher left	21:59
	→ cdown_ joined	22:00
	← oncall-pokemon left	22:00
	← cdown left	22:02
	cdown_ → cdown	22:02
	→ shabius_ joined	22:04
	← gioyik left	22:04
	← voker57 left	22:07
goh	Can there ever be a case of conflict / resolution in a two-way merge?	22:07
	← shabius left	22:07
_ikke_	goh: in a two way merge, every change is a conflict	22:07
goh	!	22:08
	→ voker57 joined	22:08
goh	so every merge in git is a 3-way merge?	22:08
	→ nutron joined	22:09
_ikke_	yes	22:09
	unless git cannot find a common ancestor (ie, when merging unrelated histories)	22:09
goh	what's the recursive thingie rafasc mentioned before?	22:10
_ikke_	man git merge, look for merge strategies	22:10
gitinfo	the git-merge manpage is available at https://gitirc.eu/git-merge.html	22:10
	→ gioyik joined	22:10
goh	I have so much appreciation for Linus from learning Git	22:10
	→ is_null joined	22:11
	→ Shentino joined	22:11
goh	It's a bit ironic that he wrote an operating system in his 20s, and then much later wrote a VCS	22:11
	One wonders what a great OS he'd write if he was to do it now.	22:11
_ikke_	That was mostly out of necessity	22:11
	They used to use bitkeeper, but they had a conflict (:p)	22:12
	→ FrozenKow joined	22:12
goh	Sure, but Git is pretty brilliant, with the DAG and all	22:12
	Imagine he used the same brilliance to write a new kernel from scratch	22:13
rafasc	more impressive that git was made in a week. If i recall correctly.	22:13
	at least the core part.	22:14
_ikke_	It was a very bare-minimal version of git	22:14
rafasc	goh: linus also made subsurface, a scuba divelog manager. But no one talks about that.	22:14
goh	Linus designed the DAG in a week?!	22:15
	I should quit my job as an engineer, I'm not worthy	22:15
Eugene	implemented*	22:15
goh	It wasn't his design/idea?	22:15
Eugene	It's been an idea on/off for a long time; the original git implementation was really just "hash some objects together". It wasn't a full-fledged system.	22:16
	← tjbp left	22:16
_ikke_	It's barely comparibly (as a complete system) with how it works now	22:16
goh	Who came up with this idea?	22:16
	→ thevishy joined	22:17
goh	"The idea" == using a DAG to implement a Merkle Tree of commits.	22:17
_ikke_	https://www.youtube.com/watch?v=4XpnKHJAok8	22:17
	← m1dnight_ left	22:17
_ikke_	very early presentation from Linus on git	22:17
rafasc	goh: a merkle tree is a type of dag.	22:18
	→ pie_ joined	22:18
	→ Codaraxis joined	22:18
	→ m1dnight_ joined	22:18
goh	rafasc: fair, I'm interested in who thought of using it to implement a VCS	22:18
	→ eb0t joined	22:20
	← feriman left	22:20
	→ eblip joined	22:21
	← def_jam left	22:22
	← eb0t_ left	22:22
	← Shentino left	22:23
	← pie_ left	22:23
	← thevishy left	22:24
FLHerne	goh: Bitkeeper, Bazaar-ng and mercurial all predated git	22:24
goh	Bitkeeper already used Merkle Trees?	22:24
	goh moves on to worship the developer of Bitkeeper	22:25
	← royal_screwup21 left	22:25
	→ Shentino joined	22:26
	→ zher joined	22:26
_ikke_	FLHerne: I recall Mercurial and git being about the same age	22:27
FLHerne	goh: hg does; not 100% sure about Bitkeeper	22:27
	(but that definitely pioneered a lot of the conceptual model that git uses)	22:28
_ikke_	Linus credits bitkeeper a big deal for the way he designed git	22:28
goh	per Wikipedia, Git and Hg initial releases are the same month	22:28
	April 2005	22:28
	→ yashi joined	22:29
FLHerne	_ikke_: I remember the "what VCS should we use for the kernel...hey look, I wrote git" thread features Linus testing (and complaining about) hg	22:29
	Too damn slow at the time, IIRC	22:29
	→ thevishy joined	22:29
FLHerne	(possibly still)	22:29
_ikke_	FLHerne: They have been using BitKeeper for some time	22:29
FLHerne	_ikke_: I know	22:29
goh	FLHerne: makes sense, as I believe Hg was originally written in pure Python	22:30
	much of the internals have since been replaced with C	22:30
	but Git is probably still faster.	22:30
_ikke_	isn't it still mostly writtin in python?	22:30
goh	most or even all of the hot loops are in C nowadays	22:31
	so classic Python-with-C program.	22:31
	IIRC there's other stuff that makes it less efficient	22:32
	like the way the object database is stored.	22:32
	← rafasc left	22:32
	→ _salt joined	22:33
	→ Sasazuka__ joined	22:33
	← __salt left	22:33
mud	Ya, I'd imagine the "other stuff" is the big difference, since even git is a mix, it's not only C. And just the fact that linus heavily focused on speed as a UI goal from I'd assume the very beginning.	22:33
_ikke_	yes	22:34
mud	Tech can matter a lot less than just what you focus on optimizing.	22:34
_ikke_	"If it's not fast, it's not worth using"	22:34
goh	I wouldn't be so sure. There's still a big difference between a Python program that gradually replaced most of its hot loops with C, and a mostly pure C program	22:34
	← m1dnight_ left	22:34
goh	AFAIL Perl is only used as glue in one-time operations?	22:35
	→ m1dnight_ joined	22:35
_ikke_	Most scripts have been rewritten in C	22:35
	there are just a hand-full left	22:35
goh	It's sort of like buying a regular car and upgrading it to a racing car, vs getting a racing car with some very minor compromises :)	22:36
	← Sasazuka left	22:36
FLHerne	goh: Quite a lot of git is/was made of shell scripts :P	22:37
	← prepend left	22:37
	← Arguggi left	22:38
	→ Arguggi joined	22:39
_ikke_	mostly was	22:39
	← kleisli left	22:42
	← mud left	22:42
goh	FLHerne: a lot of shell with most of the key performance parts written in C would still outperform pure Python :)	22:46
	especially when the C parts are written by Linus	22:46
	BTW, is there a goal to eliminate Perl as a dependency of Git?	22:46
	← thevishy left	22:48
	→ emmanuelux joined	22:49
	← prc94 left	22:50
	← azwieg103 left	22:51
	→ azwieg103 joined	22:51
	← jetchisel left	22:51
	→ thevishy joined	22:53
	← thevishy left	22:57
	→ ARoxdale joined	22:57
	← fission6 left	22:58
ARoxdale	I'm working on a cross platform project and I'm asking about autocrlf settings. If I selected "autocrlf=input" apparently I will commit as LF. But are only by diff lines commited as LF, or is the whole file changed to LF?	22:58
	→ Mattiaslndstrm joined	23:00
	→ thevishy joined	23:02
	← Mattiaslndstrm left	23:02
_ikke_	goh: there is no direct goal	23:04
	← alexandre9099 left	23:05
	← thevishy left	23:06
	→ alexandre9099 joined	23:08
	← Lucas_Gray left	23:08
	← xcm left	23:09
	→ Lucas_Gray joined	23:09
	→ xcm joined	23:10
	← Phylock left	23:11
	→ thevishy joined	23:12
	← thevishy left	23:14
	→ fphilipe joined	23:14
	← is_null left	23:15
	← Dirak left	23:15
	→ SpeakerToMeat joined	23:17
	← fphilipe left	23:19
	→ thevishy joined	23:19
	← dartmouthed left	23:20
	→ Dirak joined	23:20
	← gioyik left	23:20
	← xelxebar left	23:23
	→ xelxebar_ joined	23:23
	Disco → Disconsented	23:24
	→ Phylock joined	23:24
	→ gioyik joined	23:25
	← thevishy left	23:26
Inside	Welp, I'm still having weird github issues	23:26
	Or SSH issues - I can't tell. In the local VM that I am on, I can do 'git remote update' multiple times with no errors. On the staging environment, that command fails about 50% of the time.	23:28
	← Jonno_FTW left	23:31
	→ thevishy joined	23:31
	← zher left	23:31
	← plexigras left	23:32
	→ Sasazuka joined	23:32
	← TheJollyRoger left	23:33
	← greggerz left	23:34
	← Sasazuka__ left	23:34
	← leeN left	23:34
	→ TheJollyRoger joined	23:36
	← H2Q left	23:36
	← thevishy left	23:37
	← greatgatsby left	23:41
	→ ferdna joined	23:44
	← oatmealraisin left	23:44
	→ kleisli joined	23:44
	→ dedowsdi joined	23:45
	← Aelius left	23:47
	← gioyik left	23:50
	→ Aelius joined	23:51
	← thiago left	23:52
	← clime left	23:53
	← ferdna left	23:54
	← Shentino left	23:55

Logs Search ←Prev date Next date→

Channels Documentation