Post

I don't even think git being licensed more permissively (eg, MPL2 or something) would be a bad thing! It would be a net positive for humanity probably. Just, like. Using AI to rewrite and relicense one of the single most-used pieces of GPLed software is actually a declaration of licensing war.

And git probably has contributors with standing to sue who might have the ideological motive to do so. Also, like. How are you not a derived work? What made you come to that conclusion? Or do you just really WANT to not be?

Since people aren't connecting the dots on this I feel like I need to point out that the author of this post is GitButler's CEO and the same Scott Chacon whose name is proudly displayed as a co-author at git-scm.com/book. He was also a Github cofounder.

yeah the fuck is this. also why would you not just vibe-code whatever missing features into a fork of jj if you want a good vcs written in rust

if you, personally, were to port git to rust by hand do you think the gpl would still apply? or is the agent the problem here in your view?

Not so much because I disagree (although I think I probably do) I'm curious as to why permissively licensing git would be a good thing?

1) broadly speaking, we want people to use git and support git. The GPL makes it more difficult to do so even for open-source projects. Basically everyone, open source or no, has to shell out to the git binary instead of using git as a library.

I read right up to "i skipped internationalization because not everyone needs it." Absolutely incredible argument.

TBF if I understand correctly git's i18n support is mostly "using gettext to translate the porcelain outputs", which is... not super relevant in a library context? You'd likely want to do your own i18n If you're writing an app based a git library anyways.

(it's possible git has other i18n features I am overlooking whose exclusion is more serious though.)

I sort of hope FSF/or relevant party do sue just to make the point that license washing is no bueno

I don't think the FSF itself has standing to do so? I'm not a lawyer though. We'll probably find out if they have standing because I don't a world where they don't file suit if they do. This is an existential threat for them.

Given that whatever LLM they're using under the hood is almost certainly trained on the Git source code, even if they didn't also make the Git source code available to the LLMs (which they well may have), and the test code is anyway also copyrighted and GPL-licensed... Good luck, buddies.

This is like me deciding that my AI upscale of INCEPTION which also recuts it in the order suggested by that one Reddit post "isn't a derivative work" just because I say so

Big red flag here: "especially given the pretty massive and widespread architectural changes needed to make the implementation libified and memory safe". "Magnitude of changes" does not affect copyright. If you start from the original it's always a derivative work no matter how much you change it

Similar thing happened with chardet and, well, I dunno how well that flies. "I used an LLM to re-do the same thing" sure sounds derivative to me. "Widespread architectural changes" is doing a lot of heavy lifting for what seems to be primarily an intentional re-write of the existing Git.

I can't imagine that the FSF wouldn't sue to protect the LGPL.

I hope they do get sued! As I'd love to see how this goes. What happens when everyone just points their claudes at specs and outputs and *doesn't look at* the source code? Is that still a derivative? Or a reimplementation? If latter, is license relevant? Software licensing gonna be get fun!

That's a different question. What's relevant here is that every AI has trained on Git's code. So it's not being reimplemeted without looking at the code.

Remembering that Oracle successfully affirmed in court that *APIs* can be copyrighted.