IRCloggy #git 2021-09-22

Provider of IRC logs since 2005.

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

	2021-09-22
	→ adlternative joined	00:01
	→ mrfhitz joined	00:02
	← lgc left	00:10
	← adlternative left	00:11
	← computeiro left	00:12
	← sarab left	00:13
	→ computeiro joined	00:13
	→ sarab joined	00:19
	← magic_ninja left	00:20
	← natrys left	00:25
	→ avarab joined	00:33
	← avar left	00:33
	← gioyik left	00:36
	→ igemnace joined	00:37
	← bket left	00:37
	→ ash_worksi joined	00:37
	← durham left	00:40
	→ bket joined	00:49
	← computeiro left	00:49
	← bket left	00:57
	← ash_worksi left	00:58
	← alzgh left	00:59
	→ magic_ninja joined	01:06
	← jwillikers left	01:07
	→ bket joined	01:09
	← hornhack left	01:12
	→ hackinghorn joined	01:13
	← sarab left	01:13
	→ durham joined	01:17
	← ___nick___ left	01:17
	→ ___nick___ joined	01:19
	← m-atoms left	01:23
	→ greyrat joined	01:27
	→ thelounge6619 joined	01:27
	→ lucerne joined	01:27
	← keypusher left	01:29
	← magic_ninja left	01:30
	← alkino left	01:30
	← elkalamar left	01:32
	← anton left	01:32
	→ magic_ninja joined	01:32
	← nyah left	01:33
	→ gioyik joined	01:35
	→ nightstrike joined	01:39
	← magic_ninja left	01:41
	→ magic_ninja joined	01:42
	← sgn left	01:43
	→ anton joined	01:44
	← gioyik left	01:44
	→ sgn joined	01:45
	→ adlternative joined	01:50
	→ a6a45081-2b83 joined	01:59
	→ causasui joined	02:06
	→ Dotz0cat joined	02:14
	→ Guest85 joined	02:17
	→ hoochmonger joined	02:24
	← causasui left	02:25
	→ twb joined	02:26
twb	I keep forgetting this. If I want to start a completely new branch, not connected to anything, how do I do that?	02:26
	→ gioyik joined	02:28
	→ causasui joined	02:32
	← mrfhitz left	02:32
	← gioyik left	02:33
	→ jetchisel joined	02:35
	← deltab left	02:36
	← causasui left	02:41
	← FinnElija left	02:44
	→ causasui joined	02:44
	→ FinnElija joined	02:46
	→ deltab joined	02:46
	← christall left	02:47
	→ christall joined	02:47
	← Guest17 left	02:48
	← jonosterman left	02:48
	→ jonosterman joined	02:50
	→ gioyik joined	02:53
	← christall left	02:58
	← gioyik left	02:59
	← hnOsmium0001 left	03:07
	← ChmEarl left	03:07
	← zebrag left	03:17
	← sgn left	03:19
	→ sgn joined	03:20
	← junktext left	03:21
	← anton left	03:22
	→ zebrag joined	03:22
	← austin987 left	03:28
	→ christall joined	03:28
	→ hnOsmium0001 joined	03:31
	← AnAverageHuman left	03:33
	← pulse left	03:33
	→ gioyik joined	03:35
	← gioyik left	03:40
	→ lpapp_ joined	03:40
	→ austin987 joined	03:43
	← lpapp_ left	03:45
	← christall left	03:49
	← Xaldafax left	03:52
	← ferdna left	03:52
	→ aniruddha joined	03:53
	← hoochmonger left	04:00
	← cweiss left	04:01
	→ cweiss joined	04:02
bradyt	O.O	04:03
	if i knew the name of it, i could give a hint quickly	04:03
	what are those called? parentless commits?	04:03
	but you mean not even a commit?	04:03
	a new root commit?	04:04
twb	bradyt: no commits at all, exactly	04:06
bradyt	eww	04:06
twb	the same as if you did "git init"	04:06
	I ended up futzing it with filter-repo	04:06
	bradyt: you look funny with your work nick on	04:06
bradyt	why would you want _no_ commit?	04:06
	twb: yeah i mention my repos elsewhere and i don't feel a need to express myself elsewhere ;)	04:07
twb	bradyt: so that I can cherry-pick some other commits to be the start of a fresh branch	04:07
bradyt	oh, yeah, hard to manipulate the root commits i guess	04:08
twb	Basically I had a commit history with derp.c derp.c derp.c derp.c good.c good.c <delete derp.c>	04:08
	And I wanted to just go "actually pretend derp.c never existed, it can go in its own 'failed experiments' repo"	04:08
	← zebrag left	04:13
	← subopt left	04:14
	← gast0n left	04:15
	→ christall joined	04:17
	← pyeveryt_ left	04:17
	→ hbautista joined	04:17
	→ pyeverything joined	04:17
	→ fulljab joined	04:18
	→ ThomasD13 joined	04:18
twb	Is it possible to make tig colorize "fixup!" and "squash!" strings?	04:19
	← pyeverything left	04:22
	← vishal left	04:35
	← christall left	04:36
	→ vishal joined	04:39
	← bket left	04:39
	→ bket joined	04:42
	→ saroy joined	04:44
	← fulljab left	04:47
	← locrian9 left	04:47
	→ fulljab joined	04:48
	→ Brumaire joined	04:50
	→ pyeveryt_ joined	04:52
	→ gioyik joined	04:52
	← k-man left	04:54
	← a6a45081-2b83 left	04:54
	→ k-man joined	04:55
	← hbautista left	04:58
	← Guest85 left	04:59
	→ gnoo joined	05:01
	→ hbautista joined	05:01
	→ Mo joined	05:12
	← durham left	05:14
	← lavos left	05:14
	→ lavos joined	05:15
	← squirrel left	05:15
	← bradyt left	05:15
	→ lpapp_ joined	05:15
	→ bradyt joined	05:15
	← hexology left	05:17
	→ blmt8 joined	05:17
	← deepy left	05:18
	← cluelessperson left	05:18
	→ deepy joined	05:18
	→ hexology joined	05:18
	→ cluelessperson joined	05:18
	← jonosterman left	05:18
	← blmt left	05:19
	blmt8 → blmt	05:19
	← Brumaire left	05:21
	→ jonosterman joined	05:21
	← LiENUS left	05:27
	→ LiENUS joined	05:33
	← hbautista left	05:41
	→ dimi1947 joined	05:46
	← dimi1947 left	05:47
	→ ash_worksi joined	05:47
	→ alkino joined	05:49
	→ beshr joined	05:50
	← thiago left	05:51
lpapp_	hi, what is the best way to show our QA engineers how branching works in git? What gets to where when we backport to a release branch, what commits are not their from main, etc? Is it gitk or something similar?	05:52
	or just a generic git branching model picture?	05:52
	← pyeveryt_ left	05:54
	← ash_worksi left	05:54
	← thekingofbandit left	05:54
ikke	I think something generic would work better	05:55
lpapp_	is there a stock git photo for this	05:55
	screenshot*	05:55
	but in general, what do you use for the branching	05:55
	I remember there was even a command line output similar to gitk	05:55
	but I cannot remember the exact command.	05:55
	→ thekingofbandit joined	05:56
lpapp_	it was some "ascii chart" of the way branches happened to be.	05:56
	with \| / and \ indicators.	05:56
ikke	!lol	05:58
gitinfo	A nifty view of branches, tags, and other refs: git log --oneline --graph --decorate --all	05:58
	→ kenanmarasli joined	05:58
	← kenanmarasli left	05:58
	→ kenanmarasli joined	05:58
lpapp_	hmm, this is not what I have seen once.	05:59
	This does not seem to show graphs.	05:59
	it seems to show: * hash commit-title only.	06:00
	git log --oneline --graph --decorate --all	06:00
bookworm	that's what oneline does....	06:00
lpapp_	ah, without --all, it is better	06:00
bookworm	without all you'll only see commits reachable from the current branch	06:01
lpapp_	without all, I get \| for the main flow, and then / \| \ for branching off and then merging back to the main flow.	06:01
	without -all*	06:01
	without --all*	06:01
	should @gitinfo be corrected to remove --all? Or is it a Mac specific thing?	06:03
	← R2robot left	06:03
lpapp_	oh, I see, it actually works, just need to scroll up, it prints a lot more than just the graph itself!	06:04
	sorry then	06:04
	oh no, I am lost.	06:05
ikke	Normally a pager is opened	06:06
lpapp_	is there a page that explains the output?	06:06
	\|/\| \| \| \| \| \| \| \| \| \| \|	06:06
	when I scroll down, I see a lot of that	06:06
	it feels overwhelming at first why so many \|s	06:06
ikke	it means there are lots of parallel branches that were later merged	06:07
lpapp_	but if I scroll down more, they seem to grow like anything	06:07
	the \|s	06:07
	\| \| * \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \| \|	06:08
	→ R2robot joined	06:08
lpapp_	\| \| \| \| \| \| \| \| \| \| \| \|_\|_\|_\|_\|_\|_\|_\|/ / / / / / / / / / / / / /	06:08
	this output is really heavy	06:08
	← jonosterman left	06:08
lpapp_	maybe not the best to show to QA, even I am confused myself as a dev.	06:08
ikke	yes, history can become quite complex over time	06:09
	that's why I mentioned a generic branching overview might be better	06:09
lpapp_	I guess what would be useful to show them is that the release branches branched off at a certain point	06:09
	and after that, whatever goes to main, it does not go to them	06:09
	← twb left	06:09
lpapp_	it would be good to show this to them visually, maybe it is better than words	06:09
	and when we fix something, it goes into main, and if we backport a change, it goes to old release branches to as and when.	06:09
ikke	Something like https://nvie.com/posts/a-successful-git-branching-model/	06:09
	→ jonosterman joined	06:10
	→ ttree joined	06:11
	← beencubed left	06:13
lpapp_	thanks, I find this image complex: https://nvie.com/img/git-model@2x.png	06:14
	I was hoping to see a main, topic branches, and release branches, that is all	06:14
ikke	yes, git flow is a little bit more advanced	06:14
lpapp_	is there a simple representation	06:15
	basically, we use main, release branches, and topic branches	06:15
	I think that is all to it.	06:15
ikke	https://githubflow.github.io/, but that does not have a diagram	06:15
lpapp_	probably a standard way of working, I would imagine?	06:15
	yeah, a graphical presentation would probably be better perceived.	06:15
	"We consider origin/develop to be the main branch where the source code of HEAD always reflects a state with the latest delivered development changes for the next release. Some would call this the “integration branch”. "	06:19
	oh, okay, so it is the same as main for us	06:19
	that is why our flow is slightly simpler than git flow.	06:19
	or perhaps, this develop branch could be considered a long-lived topic branch in our workflow.	06:20
	but definitely not an overarching branch.	06:20
	→ wernerf joined	06:21
	wernerf → wernerfred	06:21
lpapp_	features branches and hotfix branches, not sure what the difference is	06:22
	← fulljab left	06:22
lpapp_	other than one is bug, other is feature	06:22
	but that is just semantics, mechanically, they are pretty much the same?	06:22
	it is a branch for a "topic"	06:22
ikke	how do you use release branches	06:22
	lpapp_: in git flow, hotfix branches are directly created on master, while topic branches are created on develop	06:23
	other than that, not a lot of difference	06:23
lpapp_	oh, I see, that is why we do not have a difference... because we do not have develop.	06:24
	thanks	06:24
	what I found over the years working for different companies and on different projects, some prefer fixing bugs in main first, then backport the first to release branches, and some people prefer fixing a bug in a release branch first and then backport to main.	06:25
ikke	lpapp_: https://imgur.com/a/oeNjWih	06:25
	basis	06:25
lpapp_	e.g. git-flow: Must merge back into:	06:25
ikke	Created with https://excalidraw.com/	06:25
lpapp_	develop and master	06:25
	we do the opposite	06:25
	we fix in main and then backport, unless it is really only release specific and not applicable in main	06:26
	ikke: thanks, I usually use inkscape	06:26
	→ pyeverything joined	06:26
ikke	Ok, so you use release branches to continue support on released versions?	06:27
lpapp_	yep!	06:27
ikke	right, makes sense	06:27
lpapp_	because our customers do not often upgrade	06:27
ikke	in git-flow, releases branches happen before the actual release	06:27
lpapp_	once they get a major release, they practically stick to it for a very long time	06:27
	and we want them to benefit from bugfixes we are doing for newer releases.	06:27
	yes, that is when they are branched off	06:28
	before RC1	06:28
	but we do not fix bugs there first and then merge back to main	06:28
	→ christall joined	06:34
	← alkino left	06:36
lpapp_	ikke: when was the github flow written? It still refers to master.	06:37
ikke	long time ago	06:37
lpapp_	"GitHub is 35 employees now"	06:37
	yes...	06:37
ikke	https://i.imgur.com/tHPVkOS.png	06:38
	→ srinidhi joined	06:40
ikke	Something like that/	06:40
	← Betal left	06:41
lpapp_	ikke: looks like what I described is closer to github flow	06:42
ikke	yes	06:43
	though they don't have release branches	06:43
lpapp_	yes, looks good, so only need adding the backporting workflow to it and then it is more or less complete.	06:43
	right, so we are in-between.	06:43
ikke	a service like github mostly only needs to support one deployment	06:43
lpapp_	yep	06:44
	is there a "standard flow" out there like ours?	06:44
	github flow + release branches?	06:45
ikke	Not aware of any fancy name for it	06:45
	← jazzy left	06:45
	← Furai left	06:47
lpapp_	not just name, but like a description	06:48
	so that I do not need to describe it from scratch. I can refer QA to an existing description.	06:48
ikke	Not aware of anything written down	06:48
lpapp_	ok, thanks.	06:48
	→ rsx joined	06:48
lpapp_	so, these are the two main well-known flows documented in public?	06:49
	← AbleBacon left	06:49
ikke	There are probably others, but I don't know / recollect them	06:50
lpapp_	ok, thanks.	06:50
	→ Furai joined	06:55
	← Furai left	06:55
lpapp_	ikke: how about this? https://www.endoflineblog.com/oneflow-a-git-branching-model-and-workflow	06:55
	→ Furai joined	06:55
	→ palasso joined	06:57
ikke	seems like they merge their release branches back	06:57
	← ttree left	06:57
ikke	so sounds more like git flow without an intermediate develop branch	06:58
	← jonosterman left	06:58
lpapp_	ikke: https://www.endoflineblog.com/img/oneflow/feature-branch-rebase-final.png I do not get this. How do the 3 purple circles become two?	07:00
ikke	Maybe they squashed one commit when rebasing/	07:01
	→ jonosterman joined	07:01
	← gnoo left	07:01
lpapp_	ikke: yes, it looks like they remove the release branches, so they probably do not need to maintain old releases	07:02
	← srinidhi left	07:02
lpapp_	they also use hotfixes from the gitflow model	07:02
	hotfix branches, that is	07:02
	← christall left	07:03
ikke	yes, but at this point there is very little difference between them and feature branches	07:04
	→ AnapodoPsalidaki joined	07:06
	← unluckyshrubbery left	07:09
	← gioyik left	07:12
	→ average joined	07:21
	← jrofd left	07:24
	→ srinidhi joined	07:24
	← dviola left	07:27
	→ dviola joined	07:29
	→ elkalamar joined	07:31
	← jetchisel left	07:34
	→ christall joined	07:35
	← tejr left	07:39
	→ tejr joined	07:40
	← christall left	07:40
	← Xenguy left	07:43
	→ Xenguy joined	07:43
	← jonosterman left	07:49
	→ EvilDMP joined	07:51
	→ jonosterman joined	07:51
	← Gustavo6046 left	07:57
	→ Gustavo6046 joined	07:57
	→ xiongxin8802 joined	07:58
	← xiongxin8802 left	07:58
	→ xiongxin joined	07:59
	→ TomyWork joined	08:00
	→ sarab joined	08:03
	← sarab left	08:03
	← thekingofbandit left	08:03
	← tejr left	08:07
	→ tejr joined	08:07
	→ john_johnk joined	08:08
	← Tobbi_ left	08:18
	→ Tobbi_ joined	08:18
	→ nyah joined	08:20
	← jimklimov1 left	08:21
	← lpapp_ left	08:23
	→ virr joined	08:25
virr	hello	08:25
	← rahl left	08:25
virr	I did a mistake, did the file changes in wrong branch (forget the checkout), is there a way to bring the changes to the right branch without do it manually?	08:26
ikke	virr: !float	08:26
gitinfo	virr: If you have made a change in your working directory and have NOT YET COMMITTED, you may "float" that change over to another (`git checkout existing_branch`) or new (`git checkout -b newbranch`) branch and commit it there. If the files you changed differ between branches, the checkout will fail. In that case, `git stash` then checkout, and `git stash apply` and go through normal conflict resolution.	08:26
	→ rahl joined	08:27
virr	I did the commit :(	08:27
baltazar	virr: you can git reset HEAD^ then I think	08:28
virr	but I lose the changes, right	08:28
	?	08:28
	→ mrfhitz joined	08:28
ikke	baltazar: they have not committed yet	08:29
baltazar	virr: by default it does --mixed, which only changes the index, not the worktree	08:29
	ikke: 10:27 < virr> I did the commit :(	08:29
	→ jimklimov joined	08:29
ikke	then checkout that branch you want it on	08:29
	git cherry-pick <first-branch>	08:29
	then reset the first branch a commit back	08:30
virr	ok, it seems to work	08:31
	now i can commit to the right branch	08:31
	→ lpapp_ joined	08:31
virr	thanks	08:35
	← j`ey left	08:36
	→ gordonfish- joined	08:36
	→ baltazaar joined	08:37
	← cweiss left	08:37
	← Arokh left	08:37
	→ cweiss0 joined	08:37
	→ Arokh joined	08:37
	← baltazar left	08:37
	baltazaar → baltazar	08:37
	← xD- left	08:37
	← gordonfish left	08:38
	← sco1 left	08:38
	← samuelbernardo left	08:38
	← pieguy128 left	08:38
	→ sco1 joined	08:38
	→ samuelbernardo joined	08:38
	→ pieguy128 joined	08:38
	→ xD- joined	08:39
	← jonosterman left	08:39
	→ theoceaniscool joined	08:40
	→ jonosterman joined	08:41
	→ cadmio joined	08:41
cadmio	Hello	08:42
	I want to update my repository via command line on github	08:42
	I've created an access token, but when logging in I get "remote: Support for password authentication was removed on August 13, 2021. Please use a personal access token instead."	08:43
	← xiongxin left	08:43
ikke	cadmio: did you use the access token as password?	08:43
	→ af joined	08:43
	← mrfhitz left	08:43
cadmio	no I didn't	08:45
ikke	Then you should	08:45
	just creating the token is not sufficient	08:45
	→ jetchisel joined	08:46
cadmio	ok thanks	08:47
	← cadmio left	08:47
	← hnOsmium0001 left	08:47
	→ alkino joined	08:49
	→ mrfhitz joined	08:51
	→ xiongxin joined	08:53
	→ xx joined	08:54
	← baltazar left	08:54
	← skapata left	08:54
	← mrfhitz left	08:55
	→ baltazar joined	08:56
	← jinsun left	08:57
	→ hqdruxn08 joined	08:58
	← adlternative left	09:00
	← hqdruxn08_ left	09:01
	← sudoforge left	09:03
	← lavos left	09:05
	→ molt joined	09:06
	→ hornhack joined	09:13
	← virr left	09:14
	→ unluckyshrubbery joined	09:14
	← hackinghorn left	09:16
	→ mrfhitz joined	09:17
	← mrfhitz left	09:18
	→ mrfhitz joined	09:19
	← mrfhitz left	09:19
	← R2robot left	09:21
	→ mrfhitz joined	09:22
	← Murr left	09:27
	→ Murr joined	09:28
	← jonosterman left	09:28
	→ jinsun joined	09:28
	→ jonosterman joined	09:30
	← mrfhitz left	09:33
	→ mrfhitz joined	09:33
	→ christall joined	09:38
	→ robz joined	09:45
	→ natrys joined	09:49
	← Mo left	09:51
	→ keypusher joined	09:56
	→ hackinghorn joined	09:59
	← hornhack left	10:00
	→ jla joined	10:01
	← saroy left	10:01
	→ saroy joined	10:01
	jla .. just cloned a fresh remote repo, and trying to checkout onto another branch (but master)	10:01
jla	would 'git fetch' download all branches to my hard disk so that i can happily 'git checkout otherBranch' from master ? That's not occuring... what am i missing ?	10:03
	→ keypushe- joined	10:03
ikke	jla: depending on the fetchspec	10:04
jla	?	10:04
ikke	if you did git clone -b <branchname> <remote>	10:04
	← keypusher left	10:04
ikke	then git gets setup to just fetch that single branch	10:04
jla	ah	10:04
	i just made standard 'git clone git@...reponame '	10:05
ikke	git config remote.origin.fetch	10:05
	keypushe- → keypusher	10:06
	→ computeiro joined	10:06
	→ ConstantDesc joined	10:06
jla	when i do 'git branch --list' i just see 'master' branch... even after 'git fetch (--all )'	10:06
ikke	git branch --remote	10:07
	← christall left	10:08
wyre	hi guys, could I do a rebase in a fork of a branch in the upstream that isn't present in the fork?	10:08
	I mean, guess I have an upstream with `master` and `branch1` and I do a fork of this and I create a branch like `branch2` in my fork, but people in the upstream keeps working on `branch1` ... could I do a rebase of my `branch2` in that `branch1`?	10:10
ikke	You can fetch from upstream and then rebase branch2 onto branch1, yes	10:11
wyre	ikke, so the commits in branch1 will be created in branch2, right?	10:12
ikke	wyre: commits don't belong to a single branch	10:12
	the commits in branch2 are appended to the commits in branch1	10:13
	← igemnace left	10:14
jla	thXs ikke , i solved it with 'for remote in `git branch -r`; do git branch --track $remote; done' prior to git fetch + pull (--all) ^^	10:17
ikke	just git fetch is enough	10:18
	← jonosterman left	10:18
ikke	pull is fetch and merge into local branch	10:18
	→ jonosterman joined	10:20
	← shalok left	10:21
	→ shalok joined	10:21
	→ hqdruxn08_ joined	10:21
jla	ah	10:24
	← hqdruxn08 left	10:25
	← dhch left	10:30
	← wernerfred left	10:38
	→ cousteau joined	10:38
	← ConstantDesc left	10:38
	→ jwillikers joined	10:38
	← nightstrike left	10:39
	→ reset joined	10:46
	→ ConstantDesc joined	10:51
	← xiongxin left	10:54
	→ gnoo joined	10:54
	→ momomo joined	10:55
	→ dhch joined	10:55
	→ wernerf joined	10:57
	wernerf → wernerfred	10:57
	← ConstantDesc left	11:01
	→ daoudr joined	11:03
	← jla left	11:04
	→ ConstantDesc joined	11:07
	← ConstantDesc left	11:20
	← betelgeuse left	11:20
	→ betelgeuse joined	11:21
	→ lgc joined	11:21
	→ ConstantDesc joined	11:21
	← causasui left	11:22
	→ Swahili joined	11:29
	← mrfhitz left	11:30
	→ hoochmonger joined	11:31
	→ jla joined	11:35
	← alkino left	11:36
	→ KaitoDaumoto joined	11:36
	← KaitoDaumoto left	11:37
	→ KaitoDaumoto joined	11:38
	← KaitoDaumoto left	11:38
	→ KaitoDaumoto joined	11:38
	← KaitoDaumoto left	11:38
	→ alkino joined	11:40
	← jla left	11:44
	← LuKaRo left	11:49
	→ LuKaRo_ joined	11:50
	→ onizu joined	12:01
	→ christall joined	12:05
	→ alzgh joined	12:11
alzgh	hello, I'm relatively new to git. I use `git add --interactive` a lot and sometimes when I manually edit the hunk, I get errors that the patch couldn't be applied. I don't know the reason tho.	12:11
	Other times it works.	12:12
	I can show you an example if it helps to get to the bottom of the problem.	12:12
	← computeiro left	12:13
ikke	alzgh: manipulating diffs is kind of tricy	12:14
	tricky*	12:14
	→ xiongxin8802 joined	12:14
ikke	things like context matter	12:14
alzgh	ikke: do you mean that nothing can be done about it?	12:15
ikke	It means you need to be familiar with the diff format and how patches are applied.	12:16
	There are other tools which make it easier to selectively stage changes	12:17
	for example, tig allows you to select per line to stage it	12:17
alzgh	my current problem concerns a hunk that contains the tail of my document. I want to add some lines from the beginning of that hunk and leave the rest of the tail untouched. This isn't surely very complicated, I assue?	12:19
	*assume	12:19
	→ momomo_ joined	12:20
	← momomo left	12:22
alzgh	Also, if you know of a good read(s) about how I can overcome this problem by learning how diffs work, I'll be happy to educate myself :D ikke	12:23
ikke	alzgh: the format is called unified diff	12:23
	→ tirnanog joined	12:24
	← ConstantDesc left	12:27
	← christall left	12:28
	→ otisolsen70 joined	12:29
osse	alzgh: there's a guide at the bottom of the editor	12:30
	alzgh: fro mwhat you said it sounds like you want to delete the lines you don't want to add	12:31
	← chexum left	12:32
	→ chexum joined	12:35
	→ ConstantDesc joined	12:36
alzgh	osse: I know that guide. I use it regularly. My problem is that git throws (in my ignorant view) random errors for some patches where I edit a hunk and does not do this in others.	12:39
	This should have been an easy edit.	12:39
ikke	yes, but sometimes you hit edge cases (literally and figuratively)	12:39
alzgh	Just remove the lines at the end or comment them out using `#`. But it throws an error and I can't understand why.	12:40
ikke	alzgh: can you show the diff?	12:40
alzgh	yes ikke, that must be it and me not knowing all the inner workings leads to not being able to debug.	12:40
	ikke: yes	12:40
ikke	alzgh: not sure if it matters, but one thing that git tries to help you with is to adjust the hunk header numbers	12:42
	@ -x,y +x,y @	12:42
	It could be failing there	12:43
alzgh	https://paste.centos.org/view/c60f8f09	12:50
	if you look at the last 4/5 hunk, I want only the first 2 lines of that hunk	12:50
	ikke:	12:51
	when I go into edit mode and delete the rest of the lines, it doesn't work	12:51
	when I comment out the rest of lines it also doesn't work	12:51
	→ R2robot joined	12:52
alzgh	It's actually the 5/5 hunk, the last	12:52
	→ christall joined	12:52
ikke	Do you remove '\ No newline at end of file'?	12:52
alzgh	should I or should I not?	12:52
ikke	I'm not certain. I don't think it's official part of the diff format	12:53
alzgh	I removed it know and it was accepted with a warning saying:	12:54
	← alkino left	12:57
alzgh	warning: 1 line adds whitespace errors.	12:58
	sorry for the delay	12:58
osse	missing newline at the end of the file is concidered a whitespace error by git i think	12:59
	you can just ignore it	12:59
alzgh	nice	12:59
	In some previous projects, I would get a similar error when trying to edit hunks	12:59
	In those cases the hunks always involved `#`. They were MD files and I had to use `#` for H1	13:00
	I can see why git would be confused, but I don't know what to do about it.	13:01
	→ gh34 joined	13:02
	← jonosterman left	13:04
	→ jonosterman joined	13:05
	← Swahili left	13:05
	← dviola left	13:08
	→ dviola joined	13:08
	← wernerfred left	13:11
	→ pulse joined	13:14
	← Xenguy left	13:17
	← chexum left	13:22
	→ chexum joined	13:22
	← jonosterman left	13:22
roxlu	would it be possible to find/retrieve a file from a git repository which has been removed and all I know is the name of the file?	13:23
	→ jonosterman joined	13:24
osse	try git log -- '**/file.txt'	13:25
	the first listed commit should be the one that removed it	13:25
	→ zebrag joined	13:25
	← ThomasD13 left	13:27
	← Murr left	13:27
roxlu	hmm can't find it .. maybe wasn't part of the repos in the first place. I'm not sure how ffmpeg (which is the repos) handles this. https://ffmpeg-devel.ffmpeg.narkive.com/ZLifxdzv/patch-v2-examples-vaapi-dec-scaling-add-a-vaapi-decoding-scaling-sample	13:27
	→ Murr joined	13:28
	← TJ- left	13:28
	→ TJ- joined	13:29
	← lpapp_ left	13:30
	→ kek_ joined	13:30
kek_	hi	13:30
	← jonosterman left	13:33
	← elibrokeit left	13:33
	→ jonosterman joined	13:34
	→ elibrokeit joined	13:35
	← jonosterman left	13:39
alzgh	hi kek_	13:41
	→ wernerf joined	13:41
	→ jonosterman joined	13:41
kek_	I made branch A. I checked out branch A. I commited stuff and pushed branch A and made a merge request. I now want to work on something else and make a branch B for it. But to work on B I need A. I thought I could just merge A into B and start working but then I basically already merged it, surpassing the merge request I just did. Now the reason I need A to work on B is simple: It won't compile without A but it does not depend on A. There's a	13:42
	dependency X of version 3.3 and I upgraded it locally to 3.4 and in A I solve some simple type issues which are backwards compatible. So I only need A in B because I can't compile it locally otherwise. Ofc the solution would be to have proper dependency management but that's not gonna happen (not in my power), I don't even think they fixed the version of the dependency. People just hope it works basically (and we get a lot of issues from it, but	13:42
	whatever). What should I do? Can I somehow "temporarily merge A into B, work on B, ditch A, and push B"? (that would not break anything).	13:42
	wernerf → wernerfred	13:45
	← Dotz0cat left	13:45
kek_	I think I just downgrade the dependency and work on it and try to convince the "boss" to implement better dependency management.	13:47
	→ fulljab joined	13:48
osse	You can start B from the tip of A, and after A is accepted you can rebase B onto the main branch	13:48
alzgh	based and git pilled :D	13:49
	→ Brumaire joined	13:49
ikke	You could also try to merge updating the dependency firs, and then both A and B can be based on that	13:49
	(if it makes sense)	13:49
	→ Dotz0cat joined	13:49
kek_	osse, thanks, might do that alzgh ikke didn't get what you were saying but I think what osse said makes sense and seems. At least to my limited git knowledge.	13:53
	→ Swahili joined	13:56
	← Brumaire left	13:58
	→ Brumaire joined	13:58
	← fulljab left	13:58
	← deltab left	14:03
	→ AnAverageHuman joined	14:04
	→ deltab joined	14:10
	← jwillikers left	14:10
	→ hnOsmium0001 joined	14:12
	→ fulljab joined	14:14
	→ geirha1 joined	14:14
	← elkalamar left	14:15
	geirha1 → geirha	14:15
	← hoochmonger left	14:21
	→ sadtaco joined	14:26
	→ arcatech joined	14:26
	← natrys left	14:28
	← Brumaire left	14:29
	← jonosterman left	14:29
	→ sudoforge joined	14:30
	→ jonosterman joined	14:31
	← jonosterman left	14:31
	→ jonosterman joined	14:33
	← christall left	14:34
	→ thiago joined	14:39
	→ christall joined	14:39
	→ gast0n joined	14:42
	← cousteau left	14:43
	← wernerfred left	14:43
	→ Xaldafax joined	14:53
	→ AbleBacon joined	14:57
	← fulljab left	15:00
	→ ChmEarl joined	15:07
	← average left	15:11
	← kenanmarasli left	15:11
	→ odoood joined	15:11
	gordonfish- → gordonfish	15:17
	→ elkalamar joined	15:18
	← jonosterman left	15:18
	→ jonosterman joined	15:22
	← jonosterman left	15:24
	→ jonosterman joined	15:25
	→ thekingofbandit joined	15:25
	← jonosterman left	15:26
	← kek_ left	15:26
	→ jonosterman joined	15:26
	← Murr left	15:27
	→ Murr joined	15:28
	← jonosterman left	15:31
	→ jmartin joined	15:32
	← jmartin left	15:32
	→ jonosterman joined	15:32
	← jonosterman left	15:34
	→ jonosterman joined	15:34
	← jonosterman left	15:35
	→ jonosterman joined	15:35
	← madewokherd left	15:36
	→ Null_A joined	15:37
	← jonosterman left	15:38
	→ jonosterman joined	15:39
	→ subopt joined	15:43
	subopt → _subopt	15:43
	← pulse left	15:43
	→ jazzy2 joined	15:44
	jazzy2 → jazzy	15:46
	→ Guest63 joined	15:47
	← srinidhi left	15:47
Guest63	Hello there. This is my first post. What would be the set of git commands any newbie should master?	15:48
	→ Brumaire joined	15:49
	→ average joined	15:49
ikke	clone, status, add, commit, log, diff, show, branch, fetch, push, merge, checkout (or switch + restore). A bit further rebase and pull	15:51
cbreak	Guest63: man git-xxx is a great command to start with for any xxx subcommand :)	15:55
	git has very good man pages in my opinion	15:55
tang^	!book	15:56
gitinfo	There are several good books available about git; 'Pro Git' is probably the best: https://git-scm.com/book but also look at !bottomup !cs !gcs !designers !gitt !vcbe and !parable	15:56
tang^	lots of good reading material, too	15:57
Guest63	thanks guys!	15:57
	_subopt → subopt	16:00
	→ mokulus joined	16:01
	← cyberpear left	16:01
	→ fulljab joined	16:02
	← fulljab left	16:03
	→ madewokherd joined	16:04
	→ dionys joined	16:04
	← tejr left	16:04
	→ tejr joined	16:05
	→ Guest17 joined	16:05
	← rsx left	16:05
	→ jimklimov1 joined	16:07
	→ durham joined	16:07
	→ m0viefreak joined	16:08
	← jimklimov left	16:08
	← jonosterman left	16:08
	← xiongxin8802 left	16:08
	→ jonosterman joined	16:10
	→ meator joined	16:12
	→ kenanmarasli joined	16:14
	→ junktext joined	16:15
	← axisys left	16:16
	→ axisys joined	16:17
	← austin987 left	16:19
	→ maroloccio joined	16:22
	← jonosterman left	16:22
	← Null_A left	16:23
	→ Null_A joined	16:23
	→ jonosterman joined	16:23
	→ Null_A_ joined	16:26
	→ Null_A__ joined	16:28
	← alzgh left	16:29
	← Null_A left	16:29
	→ Null_A joined	16:30
	→ austin987 joined	16:30
	← Null_A_ left	16:31
	→ perrierjouet joined	16:32
perrierjouet	hi all	16:32
	can we do this git submodule add --recursive ?	16:32
	← Null_A__ left	16:33
	← Guest63 left	16:33
cbreak	perrierjouet: you can git clone --recursive and then submodule add	16:35
	← TomyWork left	16:35
	← odoood left	16:36
	← Null_A left	16:36
	← meator left	16:38
	→ Null_A joined	16:41
	→ LuxuryMode joined	16:51
	← john_johnk left	16:52
	→ z3r0 joined	16:52
	← Brumaire left	16:53
	→ meator joined	16:53
	← AnAverageHuman left	16:55
	→ natrys joined	16:55
	→ Null_A_ joined	16:55
	← Null_A left	16:58
	← jonosterman left	16:58
	→ jonosterman joined	17:00
	← meator left	17:00
	← gnoo left	17:01
	→ Null_A joined	17:01
	→ cousteau joined	17:04
	← Null_A_ left	17:04
cousteau	Hi	17:04
	→ ash_worksi joined	17:05
	→ odoood joined	17:05
	← christall left	17:05
	→ christall joined	17:06
	→ cek joined	17:06
cousteau	Is it possible to make git worktree use relative paths? I just `cat ../the_worktree/.git` and it says `gitdir: /home/cousteau/the_project/the_git_repo/.git/worktrees/the_worktree` instead of the more convenient `gitdir: ../the_git_repo/.git/worktrees/the_worktree`	17:07
	Like, why isn't this the default behavior? Or at least the default behavior when I add the worktree using a relative path (which I did)	17:08
	...Or is it the default behavior on newer versions?	17:09
	← christall left	17:11
	← Swahili left	17:11
	→ AnAverageHuman joined	17:14
cousteau	In fact, do .git files (gitdir: ...) always use absolute paths (e.g. on submodules)? If so, shouldn't they default to relative paths? I remember having trouble or uncertainty when moving projects around because of absolute paths being used somewhere	17:17
	OK, submodules are smart enough to use relative paths	17:22
	→ srinidhi joined	17:22
	→ christall joined	17:22
cousteau	Both in the submodule's folder .git and in the .gitmodules file	17:23
	← AnAverageHuman left	17:24
	→ fulljab joined	17:25
cousteau	BUT, the config option submodule.the_submodule.url uses absolute paths	17:25
	Weird	17:25
	← fulljab left	17:26
cousteau	Why would .git/config and .gitmodules use different values for the url field?	17:27
	→ fulljab joined	17:27
	← Murr left	17:27
	← christall left	17:27
	→ Murr joined	17:27
	← fulljab left	17:28
	→ christall joined	17:29
	→ AnAverageHuman joined	17:34
cousteau	I guess I'll write to the mailing list to complain :)	17:35
	← sudoforge left	17:38
	→ pulse joined	17:38
	← AnAverageHuman left	17:40
	← ash_worksi left	17:41
	→ whytwulf joined	17:44
	← whytwulf left	17:45
	← ConstantDesc left	17:45
	→ whytwulf joined	17:46
	→ fulljab joined	17:46
	← whytwulf left	17:46
	→ Brumaire joined	17:47
	← fulljab left	17:48
	→ fulljab joined	17:48
	← jonosterman left	17:48
	→ ash_worksi joined	17:51
	→ jonosterman joined	17:51
imMute	cousteau: I think config URLs need to be absolute so that git can actually use them. and .gitmodules URLs can be absolute or relative depending on your needs.	17:53
	→ clime joined	17:53
	← durham left	17:54
	→ skapata joined	17:54
	← skapata left	17:54
	→ skapata joined	17:54
	→ whytwulf joined	17:56
	← whytwulf left	17:56
cousteau	imMute: I don't think they NEED to be absolute. They could be "relative to the project's top dir", and let git do the relativification	17:57
imMute	cousteau: URLs refer to remotes, not the local path...	17:57
ikke	I think the difference is that worktrees are more external to the git repository	17:57
	← z3r0 left	17:57
ikke	If they were relative, and you moved the git repository, the worktrees would break	17:57
cousteau	Otherwise you get errors when moving projects around. One should be able to move entire projects to other paths without fear of errors due to absolute paths	17:58
	→ hornhack joined	17:58
cousteau	ikke: true, but if they're in the same super-directory, and you move the whole super-directory, they wouldn't break (if they were relative)	17:59
	→ sudoforge joined	17:59
	← hackinghorn left	18:00
cousteau	And if you specified the path of the worktree as a relative path, chances are that it is in the same super-directory as the git project. That's why I was thinking on making it absolute or relative depending on how you specified it when creating the worktree	18:00
imMute	that would make sense to me. let the user decide which mode to use.	18:01
cousteau	Alternatively, an --absolute-path/--relative-path switch (which may default to whichever style the user used when specifying the path)	18:04
	s/may/probably should/	18:05
	→ con3 joined	18:05
	→ durham joined	18:05
cousteau	Gotta go, bye! I'll see if I can send an email to the mailing list later	18:06
	← cousteau left	18:06
	← nvmd left	18:08
	← ash_worksi left	18:10
	← sudoforge left	18:11
	← elkalamar left	18:11
	← odoood left	18:13
Gustavo6046	How do I changelog?	18:13
imMute	Gustavo6046: nano CHANGELOG	18:13
Gustavo6046	:<	18:13
	how do I changelog with git?	18:14
	→ cyberpear joined	18:14
ikke	Gustavo6046: if you have good commit hygene, you could use git shortlog	18:14
imMute	you mean how do you ask git for a change log? 'git log ...' with appropriate options for the commit ranges and data you want.	18:14
	→ mrfhitz joined	18:14
Gustavo6046	ikke: My commit hygiene is mehh, but I can try tags	18:14
	→ nvmd joined	18:15
Gustavo6046	can I sign and annotate a tag at the same time?	18:15
	just -as?	18:16
	-a's description almost implies that it doesn't like -s	18:16
ikke	Gustavo6046: A signed tag is annotated by default	18:17
	so it's implied	18:17
Gustavo6046	Ah	18:17
ikke	Well, not only by default, it must be annotated	18:17
	← unluckyshrubbery left	18:17
Gustavo6046	ahhh, I see	18:17
	→ unluckyshrubbery joined	18:17
	← durham left	18:18
	→ durham joined	18:19
Gustavo6046	and how do I one-time view a specific file from another commit?	18:21
ikke	git show <commit>:<path/to/file>	18:22
Gustavo6046	oh, ooh	18:22
	→ sudoforge joined	18:22
Gustavo6046	thank you!	18:24
	← nvmd left	18:26
	ferr_ → fercell	18:27
	← saroy left	18:27
	← average left	18:31
	← koolazer left	18:35
	→ snedd joined	18:35
	← jonosterman left	18:38
	← imMute left	18:38
	→ imMute joined	18:38
	← imMute left	18:41
	→ imMute joined	18:41
	→ jonosterman joined	18:41
	← Guest17 left	18:42
	→ Guest17 joined	18:42
	→ voltron joined	18:43
	→ irrgit joined	18:44
	← vdamewood left	18:44
	← fulljab left	18:49
	→ sadako joined	18:52
	← kenanmarasli left	18:53
	← sadtaco left	18:53
	→ meator joined	18:57
	← shailangsa left	19:01
	→ vdamewood joined	19:11
lantech19446	I created a second branch on github to avoid my changes going to production prematurely, it's populated, making the branch on my local repo and pushing from it were no problem but i can't seem to figure out how to pull from it	19:20
osse	lantech19446: how did you push to it?	19:22
lantech19446	first I did git push --set-upstream origin changes	19:23
	then I just did git push	19:23
osse	then git pull should work out of the box	19:23
lantech19446	from a different computer?	19:23
osse	no	19:23
	from a different computer you can git pull origin changes	19:24
lantech19446	because when I try git pull --set-upstream origin changes it tells me it's not valid	19:24
ikke	lantech19446: you want to use git fetch for new branches	19:24
	and then checkout	19:24
	lantech19446: git pull = fetch + merge	19:24
	I assume you don't want to immediately merge that branch in your currently checked out branch?	19:25
lantech19446	it's ok to do that on this computer i just can't do it on the production server	19:25
	→ nvmd joined	19:25
lantech19446	at some point i'll need to merge the two branches on github and then just pull from my master branch	19:25
	← pizdets left	19:25
osse	what is the error message	19:26
lantech19446	I'm not getting one now I apparently just can't set my upstream to that branch	19:26
	← EvilDMP left	19:26
osse	set-upstream is not a valid git pull flag	19:26
	it is for push, though	19:26
lantech19446	ok	19:26
	→ EvilDMP joined	19:27
	→ pizdets joined	19:28
	← jonosterman left	19:28
	← yates left	19:29
	→ jonosterman joined	19:30
	→ ash_worksi joined	19:31
	← LuxuryMode left	19:31
lantech19446	well that's working now, now i just gotta figure out what on earth this person was thinking when they coded this	19:32
	→ Gustavo6046_ joined	19:32
	← Gustavo6046 left	19:33
	← mokulus left	19:33
	Gustavo6046_ → Gustavo6046	19:35
	→ hackinghorn joined	19:37
	← hornhack left	19:39
	→ Betal joined	19:40
	→ sniperwolf joined	19:46
	→ alzgh joined	19:48
alzgh	good bot gitinfo	19:48
	I can only recommend 'Pro Git'	19:48
Gustavo6046	Is it common practice to have code reformatters (like Python's black) run by a "CI" tool with results auto-amended to your commit and pushed upstream?	19:50
imMute	Gustavo6046: I'd rather it run locally, so I can decide when to run it and when to commit the results.	19:50
Gustavo6046	Ah.	19:51
	Good point.	19:51
ikke	yeah, running these things locally is a lot better	19:51
imMute	Gustavo6046: at most, a CI-run formatter should simply block PRs until the formatter doesn't change anything (ie, you've run it locally)	19:51
Gustavo6046	Yeah	19:51
ikke	CI could warn when things are not according to style	19:51
Gustavo6046	Yah	19:51
	Ohhh, that	19:51
	that's what flake8 does, not reformatting. I'm stupid	19:51
	Thanks!	19:51
	→ teut joined	19:54
	→ cousteau joined	19:54
cousteau	Sup	19:54
	← meator left	19:54
teut	so I pushed some changes on origin/master . now on local I reset head to head~4 --hard . now I did git pull origin master. With a fix of little 4 line conflict , it merged but all those commits I got back	19:55
cousteau	Is there a way to see the complete diff3 of a merge commit? `git diff HEAD HEAD^1 HEAD^2` only seems to show the conflict resolution changes	19:55
teut	I did not want that but to push local to origin but pulled origin to resolve cnonflicts	19:56
	what should I have used?	19:56
osse	i don't follow completely, but it sounds like you wanted to push --force	19:57
teut	I should ve done just a force push I think in this situation	19:57
	ya, ok so I will do it next time	19:58
ikke	yes	19:58
	You can still undo the merge and still force push	19:58
teut	how?	19:58
	git revert ??	19:58
	or reflog?	19:58
ikke	reset --hard HEAD~ if the merge (pull) is the last thing you did	19:58
cousteau	teut: I'm a bit confused. You pushed master to origin/master, now master = origin/master, then rewinded master 4 commits, now origin/master is ahead of master by 4 commits, then you pulled origin/master, which fast-forwarded master those 4 commits so that master = origin/master	19:59
teut	ya you are correct	19:59
	I got merge conflicts though because I merged my collegue's branch	20:00
cousteau	So pulling origin/master pretty much undoes the rewinding	20:00
	→ travaldo joined	20:00
cousteau	Oh	20:00
teut	he has 1 little change	20:00
	had*	20:00
	1 commit I guess so	20:00
	but it was a merge so shouldnt matter	20:00
	← clime left	20:01
cousteau	You can avoid guessing by opening `gitk --all&` and seeing the commit graph in all its glory :)	20:01
	← Null_A left	20:01
cousteau	I find gitk very helpful	20:02
	→ odoood joined	20:02
teut	https://dpaste.org/Qn7g	20:02
	The ampersand (&) character is not allowed. The & operator is reserved for future use;	20:03
cousteau	Huh?	20:03
ikke	I think they use windows / powershell	20:03
teut	ikke I created a commit after git add instead of git merge --continue by mistake	20:03
	ya	20:03
	I m on windows	20:04
cousteau	Oh	20:04
teut	I used git bash now	20:04
	it worked	20:04
	→ elkalamar joined	20:04
cousteau	Then gitk --all with whatever powershell uses to open programs in the background	20:04
teut	ya there is a tree now in the git gui	20:04
	← jonosterman left	20:05
	← pizdets left	20:05
teut	I already have extensions in vscode	20:05
	but both are tough to understand	20:05
	→ pizdets joined	20:05
teut	yukk, the tree looks horrible	20:05
	I did too much merging	20:05
cousteau	Not really. Each ball is a commit. It descends from another commit just below it, so that whenever you commit you add a ball on top of another.	20:06
teut	ikke so what can I do to revert it seeing the reflog?	20:06
	ya thats fine cousteau	20:06
ikke	You don't need to use the reflog to undo a merge	20:06
teut	I showed reflog so that you can understand better what I did	20:07
	I forgot what I did, often I do	20:07
	← srinidhi left	20:07
	→ jonosterman joined	20:08
teut	I think reset to 2c8b05f should do it	20:08
	→ shailangsa joined	20:08
	→ Null_A joined	20:08
ikke	Yes, most likely	20:08
	← arcatech left	20:10
cousteau	Re: three-way diff. Apparently `git diff HEAD HEAD^1 HEAD^2` (merge diff) isn't supported in 2.18, which is the newest version I could get my hands on	20:11
	Which explains why I didn't understand why I was getting the output I was getting	20:11
teut	thanks that fixed it	20:11
cousteau	Oh wait, I found a git 2.30 installed in a random place. (Who the hell maintains this machine?)	20:13
teut	if I force pull a colleague's master , will my master become exactly like his?	20:13
ikke	there is no force pull	20:14
	teut: What you want to use is fetch	20:14
	and then potentially reset --hard	20:14
teut	oh ok	20:14
	that would be dangerous to have	20:14
	yes got it thanks	20:15
	→ ZacSharp joined	20:15
	← voltron left	20:16
	← teut left	20:16
	→ fulljab joined	20:16
	← fulljab left	20:16
	← snedd left	20:16
cousteau	OK nope, using git 2.30.0 won't do it either	20:17
ikke	cousteau: I've never heard of a 3-way diff	20:17
	At least, not by specifying 3 commits	20:18
	to git diff	20:18
	← sudoforge left	20:18
	← jonosterman left	20:18
	← Null_A left	20:18
	→ jonosterman joined	20:21
	→ Null_A joined	20:22
	→ junktext_ joined	20:23
cousteau	-c. I wanted git diff -c.	20:24
	ikke: 3-way as in the merge and the two parents.	20:24
	← junktext left	20:25
	→ sudoforge joined	20:25
	← cek left	20:25
	← jimklimov1 left	20:26
	← lessless left	20:27
cousteau	Why is -c not mentioned in the git diff synopsis nor options sections?	20:28
	← ash_worksi left	20:29
	→ ash_worksi joined	20:29
	← unluckyshrubbery left	20:30
	→ unluckyshrubbery joined	20:30
cousteau	git show -c HEAD also does what I wanted. But what's this mysterious undocumented -c feature?	20:30
	← Null_A left	20:32
cousteau	And why does it generate lines saying "diff --combined" on the output but there is no `--combined` option to either diff or git diff? (it's -c, not --combined)	20:33
	← unluckyshrubbery left	20:33
	→ unluckyshrubbery joined	20:33
ZacSharp	I think -c is the short name of --combined	20:34
	but I remember two options with a c, so maybe it's the other one	20:35
cousteau	Yeah but --combined doesn't work	20:35
	There's -c and --cc	20:35
	→ jimklimov joined	20:35
	← sudoforge left	20:36
osse	man git-diff-tree mentions it	20:36
gitinfo	the git-diff-tree manpage is available at https://gitirc.eu/git-diff-tree.html	20:36
cousteau	-c generates lines saying "diff --combined", and --cc generates lines saying "diff --cc". One is consistent but the other isn't	20:36
	osse: OK, but why not git-diff?	20:36
osse	--cc is already a long flag :p	20:36
	I don't know	20:36
	→ Null_A joined	20:38
	← odoood left	20:39
cousteau	It's a rather short long flag	20:39
osse	I think git diff is sort of a wrapper around diff-files and diff-tree and friends	20:39
	at least in its implementation	20:39
cousteau	OK, so -c will only print "interesting" files (those with changes from BOTH parents), and --cc will only print "interesting" hunks	20:40
	← Brumaire left	20:40
cousteau	But none will print the complete three-way diff	20:40
osse	try -m	20:41
cousteau	In any case, I guess the interesting files are interesting enough. I just wanted to see if the merge did what I expected it to do	20:41
	← tejr left	20:42
	→ tejr_ joined	20:42
cousteau	osse: that doesn't seem to do anything in git diff	20:43
	Anyway, I guess I have the info I was looking for after all	20:44
	And now I'm going to have... dinner! Bye :D	20:44
	← cousteau left	20:45
	← Null_A left	20:46
	→ Null_A_ joined	20:53
ZacSharp	Any diff-generating command can take the -c or --cc option to produce a combined diff when showing a merge. This is the default format when showing merges with git-diff(1) or git-	20:54
	show(1). Note also that you can give the -m option to any of these commands to force generation of diffs with individual parents of a merge.	20:54
	from man git diff	20:54
gitinfo	the git-diff manpage is available at https://gitirc.eu/git-diff.html	20:54
	← Null_A_ left	20:59
	← terrorjack left	21:00
	→ terrorjack joined	21:01
	← aniruddha left	21:02
	← madewokherd left	21:03
	← otisolsen70 left	21:07
	→ Lord_of_Life_ joined	21:07
	← jonosterman left	21:08
	← ash_worksi left	21:08
	← Lord_of_Life left	21:09
	← sniperwolf left	21:09
	→ jonosterman joined	21:10
	→ ash_worksi joined	21:10
	Lord_of_Life_ → Lord_of_Life	21:10
	→ Null_A joined	21:12
	→ Brumaire joined	21:17
	← Null_A left	21:19
	→ Null_A joined	21:20
	→ lemontree joined	21:20
	← theoceaniscool left	21:25
	← Null_A left	21:26
	← Murr left	21:27
	→ Murr joined	21:27
	→ keypushe- joined	21:28
	→ ZacSharp3 joined	21:28
	← keypusher left	21:28
lemontree	after i have successfully pulled, is there a way to show me all new commits which were introduced in this pull?	21:28
	→ Null_A joined	21:28
osse	git log HEAD@{1}..HEAD for example	21:29
	← perrierjouet left	21:30
lemontree	but this is only the last commit	21:30
	what if i have more commits pulled?	21:30
	← gh34 left	21:30
osse	no it isn't	21:30
lemontree	oh	21:30
osse	i often copy and paste the range that is printed	21:31
	e.g. abcd1234...def454 master or whatever	21:31
	keypushe- → keypusher	21:31
lemontree	ah yes i was on the wrong repo	21:31
	← ZacSharp left	21:31
lemontree	thanks. :)	21:31
	← m0viefreak left	21:32
	ZacSharp3 → ZacSharp	21:32
	→ madewokherd joined	21:36
	← vdamewood left	21:36
	← ash_worksi left	21:38
	← junktext_ left	21:44
	→ junktext joined	21:44
	→ ash_worksi joined	21:45
	tejr_ → tejr	21:54
	← Samian left	21:58
	← jonosterman left	21:58
	← jsmooth left	21:58
	← natrys left	21:59
	→ jonosterman joined	22:01
	→ odoood joined	22:04
	← Null_A left	22:16
	→ jwillikers joined	22:17
	→ Null_A joined	22:19
	← nuala left	22:19
	← christall left	22:19
	← jwillikers left	22:22
	→ nuala joined	22:23
	← maroloccio left	22:27
	← mrfhitz left	22:27
	→ jwillikers joined	22:28
	← odoood left	22:28
	← Null_A left	22:28
	← junktext left	22:28
	← palasso left	22:34
	→ Xenguy joined	22:34
	→ stkrdknmibalz joined	22:36
	→ Null_A joined	22:36
	← nuala left	22:37
	← wgrant left	22:41
	← Brumaire left	22:43
	→ christall joined	22:50
	→ wgrant joined	22:54
	← ash_worksi left	23:00
	← unluckyshrubbery left	23:02
	← pizdets left	23:03
	→ pizdets_ joined	23:04
	→ unluckyshrubbery joined	23:05
	← christall left	23:07
	← unluckyshrubbery left	23:09
	→ Swahili joined	23:11
	← elkalamar left	23:15
	→ odoood joined	23:30
	← ZacSharp left	23:37
	→ Moult joined	23:44
Moult	if branch foo has extra commits x y z compared to branch bar, what is the cleanest way to merge x and y (but not z) from foo to bar without creating duplicate commits? keep in mind there are quite a few more than just x and y.	23:45
nedbat	Moult: is z later than x and y?	23:50
	→ ash_worksi joined	23:51
Moult	nedbat: hmm some will, some will not. basically i have a list of commits a b c d e, and i will want a b c, not d, and e	23:51
nedbat	Moult: you could merge c and then cherry-pick e	23:51
Moult	nedbat: one solution which works is git merge x, git merge y, but that creates lots of commits. i'm not sure how to select a range of commits. i've tried git merge x..y but it says "x..y not something we can merge"	23:51
nedbat	Moult: you merge a commit, not a range of commits. you can pick the latest commit that has only wanted commits in its history	23:52
	→ christall joined	23:52
Moult	nedbat: sorry, i am simplifying, there are about 100 commits i'm choosing from. most i want, except for a few scattered here and there	23:53
	nedbat: ah pity i can't merge a range.	23:53
nedbat	Moult: the answer doesn't change. you merge a commit, and you get all of its history	23:53
	← odoood left	23:54
Moult	nedbat: right, hence your recommendation if i want a b c, i can just merge c	23:56
nedbat	right	23:56
Moult	nedbat: i guess i could simply cherry pick ranges of commits that i want. what is the downside of doing a cherry pick?	23:58
nedbat	Moult: do you mind if i ask you about these branches? Are they meant to live for a long time? why do you need the same commits on two different branches?	23:59

Logs Search ←Prev date Next date→

Channels Documentation