#git

[cool-retro-term-official]

Github has it for you. It's literally on gitlab. Just go on sourceforge and download the source code. Just git clone it from my gitea ! Right. On. Gitbucket ! Just click your way to my Kallithea and download it ! It's on sourcehut now !

[nixcraft]

[tamlin12]

kobold git fork

kobold git fork for Avarios on FA I had to commit to that pun

[lavendercowboyart]

he thinks u should go outside ✨

[cubbihue]

When you said that the unluckiest person is most hopeful, it reminded me of Nagito Komaeda. If the universe were combined, do you think he would fair well having Anti Cosmo as his anti fairy?

Funnily enough, I feel like it would be Hajime who would have an anti-fairy companion. I mean, he’s surrounded by extremely skilled people! That would attract bad luck in some way for him. Nagito wouldn’t be good food for anti-fairies. Or Fairies. Or Pixies.

...Honestly, Fairies aren’t quite sure what food he produces. It’s like a weird amalgamation of luck, desire, greed, and all sorts of emotions they can’t disentangle. Like a crockpot gone horribly horribly wrong. Eating any bit of that would make any fairy sick.

Eugh. The hell is wrong with that one.

Bitties Series: [Start] > [Previous] > [Next]

[ogrepoppenangleaksofficial]

[sexy-dance-fighting]

My copy of the Danny Phantom graphic novel hasn't arrived yet, but based on isolated images I've seen without context, this is basically what happens right ???

I know they obtain cowboy hats and trauma

[essential-randomness]

I can't believe this is real and I'm physically holding it

(digital preview of @fujowebdev Issue 1 is shipping soon to our Kickstarter supporters 👀 )

[andmaybegayer]

me watching the QA guy copy and paste functions between his own branches

[kernelbroken]

[sharkinatimeloop]

James is a clumsy git who hit a growth spurt. He trips all over those moving stairs.

[backjustforberena]

[...] I was made a prisoner in my quarters...

[animentality]

I've been learning to code at my new job and I work with a senior software engineer, who's a genius at this sort of thing.

But like an hour ago, he was in my office angrily murmuring, "what?! what?! what the hell?" over and over again as he furiously typed on my keyboard, trying to fix something in git.

and it's honestly very reassuring to know.

that whether you're a novice or an expert at coding... sometimes you just find yourself angrily swearing at the program for showing you the same error message a dozen times, asking why are you doing that, almost begging, really.

it was actually kinda funny, he sounded like he was about to start crying, which was literally me, two hours ago, when I initially asked him for help.

it's part of the human condition, I think, wrestling with computer programs and furiously typing in commands, only to be met with defiance.

[zram-official]

looking to migrate my stuff from github to another service, and i know lots of people have been recommending to switch to gitlab for years, but is it still even that good? i see their front page advertise AI like crazy.

feature-wise what im looking for honestly is to have some private repos, and markdown to look similar enough to github so i don't have to spend too much time on the README files.

[serinemisc]

So I came across this recently.

It's funny, because I think I exactly half agree with it. I do rebase-heavy workflows in Git mostly because every single Git client makes merge-based workflows ugly and hard to use. If GitHub simply displayed merges the way it displayed squash-merges, that would eliminate so much of the need for squash-merges.

But I don't think this covers everything. So let me go through every use-case for rebase separately:

git merge --squash

The squash-merge is one of the most popular ways to merge pull requests on GitHub, and it's an abject failure of the Git ecosystem that it's so popular.

When you do a regular merge on a pull request, you are essentially taking a bundle of commits from somewhere else, and putting it on top of your own main branch. It's an extremely linear thing to do.

But if you do that, GitHub's commit log just gets a bunch of commits interspersed throughout, with zero indication where they're from. And the nicer clients, if they do, visualize it as a tree (pronounced "DAG") (pronounced "a huge tangle of curvy lines"):

This pic is from an article telling you to rebase, and, like, sure, rebasing sure is one way to work around a UI that displays your merges as a huge tangle. But Fossil makes a really good point. Why not instead display your merges as, like, not a huge tangle? git log --first-parent does this (and that's clearly an option in that Git UI), but it should be the default everywhere. And even when expanding the "bundle", the bundled commits should still be grouped together, not interspersed with other commits at essentially random.

The other issue is that, when showing the "tangle of commits", the reason it's so tangled is because it's showing the commits in chronological order of when the commits were made. Which is a completely useless sort order, compared to, say, chronological order of when they arrived in the current branch (i.e. grouping the merged-in commits together). This is why GitHub's rebase-merge is also such a popular alternative to merges.

git pull --rebase

Okay, so. Now you've fixed commit log visualization of merged pull requests. But that's not the only use of rebase! Here's another one: if you're working on some code, and constantly keeping it synced with remote, you'll generate tons of merges that are complete useless noise. Unlike a merged PR, these should ideally be hidden completely, or at least nearly-completely.

Anti-rebase people say that these merges serve the functionality of, like, preserving history. You made one commit when the remote was in this state, and another commit when the remote was in that state, and this is sometimes important history to preserve.

I think they are way overestimating how important that history is (judging by how many people use pull-rebase). I'm fine preserving that history if you can declutter the UIs, but it does require your UI to be able to distinguish between "important" merges (of new features from feature branches) and "unimportant" merges (keeping branches in sync with remotes).

The linked post doesn't talk about this problem at all, so I don't know how well Fossil handles this.

git commit --fixup

That leaves the amend/fixup commit. The link does mention that Fossil supports editing past metadata (e.g. commit message). But sometimes you want to edit the actual changes of a commit.

Now, for a sufficiently published commit, this is a bad idea. But if you have a habit of "commit early, commit often", having 50 bugfix commits makes a commit log really cluttered.

I frequently, like, have to weigh stuff like "is it worth cluttering the commit log to fix one typo in one comment?" for old code. And it would really suck to also have to do that for unpublished code, instead of going in with my trusty rebase scalpel.

git that's all I wanted to say

In conclusion. git rebase is a solution to a number of things that could also be viewed as UI problems, and fixed in other, better ways, and Fossil sure sounds like it's fixed some of them. But some of those UI problems are legitimately hard, and I'm not convinced Fossil fixes all of them, and GitHub extremely has not, so I'm gonna keep rebasing.