Better idea: rewrite the whole thing in Rust

Posted Dec 5, 2025 20:03 UTC (Fri) by ssokolow (guest, #94568)
In reply to: Better idea: rewrite the whole thing in Rust by cyperpunks
Parent article: Eventual Rust in CPython

That's what librsvg did. While maintaining the same external API, it was incrementally converted to Rust.

Here's the Rust tag on the maintainer's blog: https://viruta.org/tag/rust.html

Google has a more typical approach, where it's paid employees maintaining things like Android, not volunteers, and they recognize that most memory-safety bugs are in young code, and that there's a risk of re-introducing logic bugs during a rewrite (differential fuzzing is your friend), so their approach with Android has been to move the new work to Rust but to not rewrite stuff in C for its own sake.

They blogged about that here: https://security.googleblog.com/2024/09/eliminating-memor...

Better idea: rewrite the whole thing in Rust

Posted Dec 6, 2025 6:54 UTC (Sat) by mirabilos (subscriber, #84359) [Link] (1 responses)

It is also entirely unnecessary as there already *is* a RustPython, so kindly stick to C for CPython.

Better idea: rewrite the whole thing in Rust

Posted Dec 7, 2025 19:14 UTC (Sun) by ballombe (subscriber, #9523) [Link]

At least avoid a situation where a C implementation of the python interpreter cannot be called CPython because this name is used by the Rust version. This would no do anybody a favor.

Better idea: rewrite the whole thing in Rust

Posted Dec 6, 2025 9:59 UTC (Sat) by kunitz (subscriber, #3965) [Link]

A rewrite of a software component you need to assess whether the cost of the rewrite and the future maintenance cost is lower than the maintenance cost of the existing software component.

Google made such an assessment, but I have yet to see the "rewrite everything in X" apostles to provide such an assessment.