Re: [PATCH v3 02/15] xdiff: introduce rust

From: Patrick Steinhardt <ps@pks.im>
Date: Sep 4, 2025, 11:10 UTC
Message-ID: <aLlzj-FxXCmBXTQz@pks.im>
In-Reply-To: <xmqqms7bchln.fsf@gitster.g>

On Wed, Sep 03, 2025 at 03:10:44PM -0700, Junio C Hamano wrote:

Show 12 quoted lines

> Patrick Steinhardt <ps@pks.im> writes:
> > Note that I'm not saying that we need to have both a C and Rust
> > implementation for everything written in Rust. I don't think that's
> > sustainable in any way. But any feature written in Rust should be a
> > _new_ feature that can be disabled and that users can live without for
> > the time being.
> 
> Yes, if we can find such modular niche, it would be ideal.  But how
> many areas that we can cleanly plug an optional thing in without
> disrupting existing codebase are there?  Offhand, all I'd think of
> are a new merge backend, a new rebase backend, a transport helper,
> or perhaps a new diff-algorithm?

Not too many, I guess.

If we cannot find anything, an alternative could also be to take a very simple subsystem that doesn't see a lot of changes and convert that to Rust. We'd retain both implementations in that case, which I mentioned is painful because we now have to keep both in sync. But if we say that this is a testballoon, only, and that we don't continue to convert other code until Git 3.0, then that might be fine.

"varint.c" could be a good match. It's trivial, only 30 lines of code, and completely standalone.

We could still build new and optional functionality via Rust, but I guess it also doesn't hurt to have a test balloon that is part of libgit.a to test interoperability.

I'll send patches later today.

Patrick