Requesting a release

[chutten]

Hi! We're hoping to use this to resolve an infrequent and irritating crash in Firefox (see bug) now that we've updated our stdlib to a fixed version. Are there any plans for a release in the near term?

[josephlr]

Cutting a v0.2.9 release seems reasonable to me. It seem like the only API change is in #291. However, I'm not sure if free functions are the right API for future-proofing extensions to this crate.

@newpavlov would you be OK:

• temporiliy reverting Add getrandom_uninit(dest: &mut [MaybeUninit<u8>]) -> .... #291,
• Having a bug fix only releae v0.2.9
• Add back Add getrandom_uninit(dest: &mut [MaybeUninit<u8>]) -> .... #291
• Have further discussion about the API

[newpavlov]

@josephlr

However, I'm not sure if free functions are the right API for future-proofing extensions to this crate.

Could you expand on that? Note that we may release getrandom v0.3 in future without causing too much trouble downstream. Most of the ecosystem does not use getrandom directly and instead relies on rand_core and it does not expose any types from getrandom in its public API.

Right now I do not see the need for reverting getrandom_uninit and think that we can do release with it.

[josephlr]

Could you expand on that? Note that we may release getrandom v0.3 in future without causing too much trouble downstream. Most of the ecosystem does not use getrandom directly and instead relies on rand_core and it does not expose any types from getrandom in its public API.

Right now I do not see the need for reverting getrandom_uninit and think that we can do release with it.

See #353 for the basic idea. If we think something like that is maybe worth doing, we can revert #291 and release v0.2.9 while we discuss it.

[newpavlov]

I don't think your proposal conflicts with existence of getrandom_uninit. You left the getrandom function as a safe default which simply wraps the default option, so getrandom_uninit can be handled in the same way.

But if you feel more comfortable releasing v0.2.9 without getrandom_unit and feel strongly about it... I am fine with its temporary removal, though I personally would prefer to keep it in the new release.

[josephlr]

I don't think your proposal conflicts with existence of getrandom_uninit. You left the getrandom function as a safe default which simply wraps the default option, so getrandom_uninit can be handled in the same way.

That seems fair. I'm good to do a release with getrandom_uninit.

Final question, in #293 (comment) we talked about fill_uninit being a better name than getrandom_uninit, do we want to change the name to fill_uninit now?

[newpavlov]

I think getrandom_uninit will be a more consistent name for now. In future we may rename it together with getrandom.

[josephlr]

Sounds good, let's cut a release. I can do it Monday, but I'm OOO until then.

[josephlr]

Published: https://crates.io/crates/getrandom/0.2.9