Move away from 2015 edition?

[eduardosm]

Right now, the libc crate is at Rust edition 2015. Is there a desire to bump to a newer edition?

That would require a significant amount of changes, such as adding crate to paths that being with :: (e.g., crate::c_int instead of ::c_int). It would also require types inside an union to be Copy or wrapped in a ManuallyDrop.

[tgross35]

Yeah, we should probably target 2021 as part of 1.0. Thanks for writing this up, I have been meaning to for a while but never got to it.

[ChrisDenton]

Note that doing this may make some backports more involved.

[eduardosm]

It looks like JohnTitor/ctest2#48 blocks this.

[tgross35]

@ChrisDenton are you just referring to maintaining both 1.0 and 0.2 at the same time like we do currently? We have a handful of planned changes like this that would make backports worse, it would probably be good to get everything else lined up then do them right before we plan to release 1.0.

[ChrisDenton]

are you just referring to maintaining both 1.0 and 0.2 at the same time like we do currently?

Yes, that's what I was referring to.

We have a handful of planned changes like this that would make backports worse, it would probably be good to get everything else lined up then do them right before we plan to release 1.0.

Makes sense.

[ismailarilik]

I have tried to migrate it to 2021 using cargo fix --edition, and cargo build is now working with the edition set to 2021 in Cargo.toml. Isn't this that simple?

[tgross35]

We cherry pick commits from main to libc-0.2 for the stable release. Doing a lot of tiny changes from the autofix means that these many cherry picks won't apply as cleanly, which would be a huge annoyance - since doing this change is so easy/automated, we can defer it until we get closer to 1.0 and are almost ready to abandon libc-0.2 (and hence the cherry picks)