Meet “Cthulhu” and “Katy Freeway”, two examples from our Git history that I picked for a discussion on what •not• to do.

We luckily have a very different mindset nowadays and monstrosities like these are a thing of the past

Disappointment is seeing a SAAS claim to have effortless #git integration but only supporting Github or Gitlab.

New Git diff alternative just dropped!

**riff** — A diff filter highlighting which line parts have changed.

Supports highlighting conflict markers, merge commits & more!

Written in Rust!

GitHub: https://github.com/walles/riff

Git-hago delenda est

$ commit -m "hack: too lazy to rsync with laptop. commits go brrr.”

Someone that also works on a project I worked on ( and created) tried to do a quick one and close a pr with comments on from me, and blockers on the code review, open a new pr with the same changes (and a release tag) and quickly merge it in to main, since that new pr didn't have any blockers.

WTF. If you try to do that at any job, you would be fired.

#git #git-annex

Honestly, for large binary files, git-annex sounds like heaven. However, Codeberg/Forgejo don't seem to know about it

#git-annex #codeberg

@Codeberg Do you support git-annex?

https://git-annex.branchable.com/tips/centralized_git_repository_tutorial/

#git #GameDev

Dreaming aloud. How can git & game dev cooperate?

In project, have path/to/bigfiles where the blend files (for eg) live.

In .gitattributes:
track *.blend -> path/to/bigfiles/

Use git as normal. All blend files become little text pointers to the path. (Like in LFS.)

Push etc. sends the little files, and then rsyncs the large files to the server.

So git as usual, but it also saves your big files via rsync *without* ever-growing file sizes on the server!

Any thoughts?

Ich habe auf der Projektseite von ubuntuusers.de ein paar meiner Programme mal kurz vorgestellt. Vielleicht kann das eine oder andere Programm jemand von Euch ja gebrauchen.

https://forum.ubuntuusers.de/topic/kleine-helferlein/

#git #lfs
I was told, a few days ago, that git lfs only saves space on my local drive but that the server (origin, i guess) grows full of file versions.

E.g i push 3d.blend, it's 10mb.
I edit it, push again, server now reports 20mb used. Then 30mb etc.

Is this the expected lfs behaviour? Can someone point me to where this is documented?

I'd much rather there be one 3d.blend file locally AND remotely. Is that even possible to do?

I don't think people realize this issue about lfs.

Linus Torvalds has proper motivated reasons for really disliking file systems without a case sensitivity.

Read this with me from the kernel lkml, regarding bcachefs.

Re: [GIT PULL] bcachefs fixes for 6.15-rc4 - Linus Torvalds
https://lore.kernel.org/lkml/CAHk-=wjajMJyoTv2KZdpVRoPn0LFZ94Loci37WLVXmMxDbLOjg@mail.gmail.com/

#bash #sh #zsh #ksh #csh #tcsh #fish #git #Linux #lkml #POSIX #FOSS #100daysofCode #640DaysOfCode #coding #1024DaysOfCode #github #programming #Torvalds #filesystem

Sometimes I'm happy to tinker with #HomeLab stuff, and other times I just want it to work when I thought it'd be a quick job.

Related: If anyone knows why I'm getting a 500 error running git-http-backend in a #FreeBSD jail with #nginx that'd be great.

I've set up directory browsing as well, and that all works fine. Nginx logs show no errors. But cloning always results in a 500 response.

Have a GitFlow diagram I made using DrawIO

Dal momento che sono stufo delle minchiate dell bigtech a proposito dell'IA, ho deciso che tutto ciò che ho su #GitHub sarà trasferito su #Gitea , e questo grazie ai @devol che tengono in piedi *anche* questo servizio.
Approfitto per mostrare una schermata di #LazyGit , un client #git fantastico!

Today in #git:

Didn't realize I still had a bisect going, and when I tried to delete a branch, it said it was being used in another workspace (i'm not using workspaces).

When #git talks about a "detached HEAD", it sometimes gives me nasty mental images.

Code at work is stored in #SVN, with a highly irregular structure - no standard trunk/tags/branches, a lot of externals, a bespoke tool to do sparse checkouts. Nobody I spoke about it to likes it, but apparently it's just too entrenched into various workflows to easily migrate away from.

A lot of devs prefer #git over svn though. But because of the weird repo structure, they can't just use git svn. So the next best thing they (and I) came to is to just put it all in a local git repo - including the .svn folder. This adds a lot of overhead, sure, but it's the simplest way to get the amenities you're used to.

I came a little bit further, though. I have tried to write a script that would import a (subset of) svn history into this local git repo - by essentially parsing svn log --xml, then doing svn up && git commit on every revision, setting up authors, dates and messages nicely. It is slow, of course, but I ran it overnight. Feeling quite proud of myself now, being able to, e.g. use git blame fully locally.

Probably won't share the script here though, since it's tied to our specific repo quirks.

Pierre-Étienne Meunier and Florent Becker developed pijul [1], a modern version control system. I praised it in 2022 [2] and still use it for rust projects.

Tomorrow at Grazer Linuxtage #GLT25, I am going to present #pijul [3] in a 5-minute lightning talk and show which problem it solves (compared to git).

#SCM #VCS #git #rust

[1] https://pijul.org/
[2] https://lukas-prokop.at/articles/2022-07-26-trying-out-pijul
[3] https://lukas-prokop.at/talks/glt25-pijul/slides.pdf