Remove `generational-arena` from project

[comphead]

Is your feature request related to a problem or challenge?

Followup on #20862

Crate:     generational-arena
Version:   0.2.9
Warning:   unmaintained
Title:     `generational-arena` is unmaintained
Date:      2024-02-11
ID:        RUSTSEC-2024-0014
URL:       https://rustsec.org/advisories/RUSTSEC-2024-0014

Describe the solution you'd like

No response

Describe alternatives you've considered

No response

Additional context

No response

[Dandandan]

This is from abi_stable which also seems to be unmaintained.

[comphead]

@coderfender this might be a good followup after #20946

[comphead]

This is from abi_stable which also seems to be unmaintained.

we use it heavily in FFI, @timsaucer FYI as FFI expert

[coderfender]

Yes I did some investigation and this seems like a transitive dependency

[coderfender]

@comphead There is an existing github issue but seems like there is no recent activity : rodrimati1992/abi_stable_crates#124 . As @Dandandan , rightly pointed, we might want to deprecate usage of the abi_stable crate from DF itself ? Thoughts @timsaucer ?

[comphead]

@comphead There is an existing github issue but seems like there is no recent activity : rodrimati1992/abi_stable_crates#124

The project doesn't seem to be active for quite a while as @Dandandan mentioned, the last commit was 12 October 2023 https://github.com/rodrimati1992/abi_stable_crates/commits/master/

removing abi_stable looks requiring significant efforts, this is what genAI suggests

Project / Approach	Type	Activity	Rust Type Support	Cross-Language	Performance	Best Use Case
abi_stable	Stable Rust ABI framework	Low / slowing	High	No	High	Rust dynamic plugins with strict ABI guarantees
stabby-abi	Modern Rust ABI framework	Active	High	No	High	Replacement for abi_stable, Rust plugin systems
CGlue	FFI trait bridge	Active	Medium–High	Yes (C/C++)	High	Rust traits across FFI boundaries
savefile-abi	Adaptive ABI via serialization fallback	Experimental	Medium	Possible	Medium	Version-tolerant plugin interfaces
WASM plugins (Extism / Wasmtime)	WebAssembly runtime	Very active	Medium	Yes (many languages)	Medium	Safe sandboxed plugin systems
C ABI (repr(C) + extern "C")	Manual FFI	Universal	Low	Yes	Very High	Maximum compatibility and stability
Same-toolchain dynamic linking	Native Rust linking	Common practice	High	No	Very High	Dev tools, game engines, internal plugins

[comphead]

https://github.com/ZettaScaleLabs/stabby sounds like a replacement

[coderfender]

Yeah , I believe that is a good replacement . Perhaps worth a try to see if we can deprecate abi_stable in favor of stabby ? (Happy to take a stab at it but this requires significant rewrite AFAIK)

[timsaucer]

We could reach out to the maintainer of abi_stable as they haven't moved on my PR to handle this issue: rodrimati1992/abi_stable_crates#123

I'm very concerned about how much churn we have in the FFI crates, to the point that I even opened this issue about it. Changing from abi_stable to stabby would be a major breaking change.

It looks like stabby does have async support, so that does at least make it compelling as an alternative.

Another project to throw in the mix is https://cxx.rs/ as mentioned by @davisp but it lacks proper async support: https://cxx.rs/async.html?highlight=async#async-functions

Bottom line for me: If we could reach out to abi_stable and have them make a release with the dependency updates that would by far be my recommended way to go. If they've abandoned the project then I suppose we should consider one of these replacements. It would be helpful to have another person working deeply on the FFI crate, and this would definitely be a way into that if you're willing to take on the work load @coderfender

[davisp]

The big warning in stabby's README is a bit concerning. I haven't looked deep enough to know whether that just applies to making stable traits or if it applies to all stable types since they appear to include an autogenerated trait as well.

I agree with @timsaucer that asking for an updated release would be the easiest approach for now.

[coderfender]

I am not very confident if we would get any traction on getting changes done in abi_stable crate (unless we port it to be a part of DF ecosystem ?) .There seems to be no update on any of the GitHub issues and the last commit is around October 2023 . However, given that stabby seems to be a more promising package makes it worth a try at least and happy to be an additional dev on FFI side of things as Tim pointed out

[kumarUjjawal]

I was curios, can we fork and maintain the abi_stable crate. If this is something that would a desirable direction, I would be happy to maintain the fork. What do you all think?

[coderfender]

Yup , I proposed that we maintain the fork in the wider DF ecosystem to not have a single point failure as well. But yeah I would like to see what others in the community think