async-signature: remove mandatory Send + Sync bounds

[QuinnWilton]

See: #1373

I've confirmed that with these changes I'm able to produce signatures from non-extractable keys in a browser, using an implementation of AsyncSigner<S> that calls out to the SubtleCrypto APIs.

I have not tested the AsyncDigestSigner traits in a wasm project, however I have confirmed that I'm able to build the project with the digest feature enabled.

[tarcieri]

Did you mean to remove the 'static bound here?

[tarcieri]

Going to go ahead and merge this as we'd like to do some repo reorganization which impacts this crate

[QuinnWilton]

Thanks for merging it, and apologies for not getting back to you on the static bounds yesterday! I did intentionally remove them, but I probably should have split that into a separate commit.

I believe that the 'static bounds in the old code were actually superfluous, because all Send types are already 'static, and without the Send bound, requiring 'static would overconstrain the generic: I don't know whether that'd end up being a problem in practice though.

[tarcieri]

Oof, well it's out with the 'static bound, though it's an uncommonly used crate and we could probably yank it and cut a replacement release, or just bump the minor version again

[QuinnWilton]

No worries at all! I'll verify it still works with my code today or tomorrow, and let you know if the 'static bound gives any issues :)

It's just on the type of the signature itself, which is probably fine, since that'll mostly just be first-party types from rust-crypto anyway, which shouldn't really be holding any references to the stack anyway.

[tarcieri]

When Rust 1.75 drops we can transition away from the async-trait crate to Rust's native support for async traits, at which time the 'static bound can potentially be removed

[QuinnWilton]

See: #1373

I've confirmed that with these changes I'm able to produce signatures from non-extractable keys in a browser, using an implementation of AsyncSigner<S> that calls out to the SubtleCrypto APIs.

I have not tested the AsyncDigestSigner traits in a wasm project, however I have confirmed that I'm able to build the project with the digest feature enabled.

[tarcieri]

Did you mean to remove the 'static bound here?

[tarcieri]

Going to go ahead and merge this as we'd like to do some repo reorganization which impacts this crate

[QuinnWilton]

Thanks for merging it, and apologies for not getting back to you on the static bounds yesterday! I did intentionally remove them, but I probably should have split that into a separate commit.

I believe that the 'static bounds in the old code were actually superfluous, because all Send types are already 'static, and without the Send bound, requiring 'static would overconstrain the generic: I don't know whether that'd end up being a problem in practice though.

[tarcieri]

Oof, well it's out with the 'static bound, though it's an uncommonly used crate and we could probably yank it and cut a replacement release, or just bump the minor version again

[QuinnWilton]

No worries at all! I'll verify it still works with my code today or tomorrow, and let you know if the 'static bound gives any issues :)

It's just on the type of the signature itself, which is probably fine, since that'll mostly just be first-party types from rust-crypto anyway, which shouldn't really be holding any references to the stack anyway.

[tarcieri]

When Rust 1.75 drops we can transition away from the async-trait crate to Rust's native support for async traits, at which time the 'static bound can potentially be removed

[QuinnWilton]

See: #1373

I've confirmed that with these changes I'm able to produce signatures from non-extractable keys in a browser, using an implementation of AsyncSigner<S> that calls out to the SubtleCrypto APIs.

I have not tested the AsyncDigestSigner traits in a wasm project, however I have confirmed that I'm able to build the project with the digest feature enabled.

[tarcieri]

Did you mean to remove the 'static bound here?

[tarcieri]

Going to go ahead and merge this as we'd like to do some repo reorganization which impacts this crate

[QuinnWilton]

Thanks for merging it, and apologies for not getting back to you on the static bounds yesterday! I did intentionally remove them, but I probably should have split that into a separate commit.

I believe that the 'static bounds in the old code were actually superfluous, because all Send types are already 'static, and without the Send bound, requiring 'static would overconstrain the generic: I don't know whether that'd end up being a problem in practice though.

[tarcieri]

Oof, well it's out with the 'static bound, though it's an uncommonly used crate and we could probably yank it and cut a replacement release, or just bump the minor version again

[QuinnWilton]

No worries at all! I'll verify it still works with my code today or tomorrow, and let you know if the 'static bound gives any issues :)

It's just on the type of the signature itself, which is probably fine, since that'll mostly just be first-party types from rust-crypto anyway, which shouldn't really be holding any references to the stack anyway.

[tarcieri]

When Rust 1.75 drops we can transition away from the async-trait crate to Rust's native support for async traits, at which time the 'static bound can potentially be removed

[QuinnWilton]

See: #1373

I've confirmed that with these changes I'm able to produce signatures from non-extractable keys in a browser, using an implementation of AsyncSigner<S> that calls out to the SubtleCrypto APIs.

I have not tested the AsyncDigestSigner traits in a wasm project, however I have confirmed that I'm able to build the project with the digest feature enabled.

[tarcieri]

Did you mean to remove the 'static bound here?

[tarcieri]

Going to go ahead and merge this as we'd like to do some repo reorganization which impacts this crate

[QuinnWilton]

Thanks for merging it, and apologies for not getting back to you on the static bounds yesterday! I did intentionally remove them, but I probably should have split that into a separate commit.

I believe that the 'static bounds in the old code were actually superfluous, because all Send types are already 'static, and without the Send bound, requiring 'static would overconstrain the generic: I don't know whether that'd end up being a problem in practice though.

[tarcieri]

Oof, well it's out with the 'static bound, though it's an uncommonly used crate and we could probably yank it and cut a replacement release, or just bump the minor version again

[QuinnWilton]

No worries at all! I'll verify it still works with my code today or tomorrow, and let you know if the 'static bound gives any issues :)

It's just on the type of the signature itself, which is probably fine, since that'll mostly just be first-party types from rust-crypto anyway, which shouldn't really be holding any references to the stack anyway.

[tarcieri]

When Rust 1.75 drops we can transition away from the async-trait crate to Rust's native support for async traits, at which time the 'static bound can potentially be removed

[QuinnWilton]

See: #1373

I've confirmed that with these changes I'm able to produce signatures from non-extractable keys in a browser, using an implementation of AsyncSigner<S> that calls out to the SubtleCrypto APIs.

I have not tested the AsyncDigestSigner traits in a wasm project, however I have confirmed that I'm able to build the project with the digest feature enabled.

[tarcieri]

Did you mean to remove the 'static bound here?

[tarcieri]

Going to go ahead and merge this as we'd like to do some repo reorganization which impacts this crate

[QuinnWilton]

Thanks for merging it, and apologies for not getting back to you on the static bounds yesterday! I did intentionally remove them, but I probably should have split that into a separate commit.

I believe that the 'static bounds in the old code were actually superfluous, because all Send types are already 'static, and without the Send bound, requiring 'static would overconstrain the generic: I don't know whether that'd end up being a problem in practice though.

[tarcieri]

Oof, well it's out with the 'static bound, though it's an uncommonly used crate and we could probably yank it and cut a replacement release, or just bump the minor version again

[QuinnWilton]

No worries at all! I'll verify it still works with my code today or tomorrow, and let you know if the 'static bound gives any issues :)

It's just on the type of the signature itself, which is probably fine, since that'll mostly just be first-party types from rust-crypto anyway, which shouldn't really be holding any references to the stack anyway.

[tarcieri]

When Rust 1.75 drops we can transition away from the async-trait crate to Rust's native support for async traits, at which time the 'static bound can potentially be removed

[QuinnWilton]

See: #1373

I've confirmed that with these changes I'm able to produce signatures from non-extractable keys in a browser, using an implementation of AsyncSigner<S> that calls out to the SubtleCrypto APIs.

I have not tested the AsyncDigestSigner traits in a wasm project, however I have confirmed that I'm able to build the project with the digest feature enabled.

[tarcieri]

Did you mean to remove the 'static bound here?

[tarcieri]

Going to go ahead and merge this as we'd like to do some repo reorganization which impacts this crate

[QuinnWilton]

Thanks for merging it, and apologies for not getting back to you on the static bounds yesterday! I did intentionally remove them, but I probably should have split that into a separate commit.

I believe that the 'static bounds in the old code were actually superfluous, because all Send types are already 'static, and without the Send bound, requiring 'static would overconstrain the generic: I don't know whether that'd end up being a problem in practice though.

[tarcieri]

Oof, well it's out with the 'static bound, though it's an uncommonly used crate and we could probably yank it and cut a replacement release, or just bump the minor version again

[QuinnWilton]

No worries at all! I'll verify it still works with my code today or tomorrow, and let you know if the 'static bound gives any issues :)

It's just on the type of the signature itself, which is probably fine, since that'll mostly just be first-party types from rust-crypto anyway, which shouldn't really be holding any references to the stack anyway.

[tarcieri]

When Rust 1.75 drops we can transition away from the async-trait crate to Rust's native support for async traits, at which time the 'static bound can potentially be removed

[QuinnWilton]

See: #1373

I've confirmed that with these changes I'm able to produce signatures from non-extractable keys in a browser, using an implementation of AsyncSigner<S> that calls out to the SubtleCrypto APIs.

I have not tested the AsyncDigestSigner traits in a wasm project, however I have confirmed that I'm able to build the project with the digest feature enabled.

[tarcieri]

Did you mean to remove the 'static bound here?

[tarcieri]

Going to go ahead and merge this as we'd like to do some repo reorganization which impacts this crate

[QuinnWilton]

Thanks for merging it, and apologies for not getting back to you on the static bounds yesterday! I did intentionally remove them, but I probably should have split that into a separate commit.

I believe that the 'static bounds in the old code were actually superfluous, because all Send types are already 'static, and without the Send bound, requiring 'static would overconstrain the generic: I don't know whether that'd end up being a problem in practice though.

[tarcieri]

Oof, well it's out with the 'static bound, though it's an uncommonly used crate and we could probably yank it and cut a replacement release, or just bump the minor version again

[QuinnWilton]

No worries at all! I'll verify it still works with my code today or tomorrow, and let you know if the 'static bound gives any issues :)

It's just on the type of the signature itself, which is probably fine, since that'll mostly just be first-party types from rust-crypto anyway, which shouldn't really be holding any references to the stack anyway.

[tarcieri]

When Rust 1.75 drops we can transition away from the async-trait crate to Rust's native support for async traits, at which time the 'static bound can potentially be removed

[QuinnWilton]

See: #1373

I've confirmed that with these changes I'm able to produce signatures from non-extractable keys in a browser, using an implementation of AsyncSigner<S> that calls out to the SubtleCrypto APIs.

I have not tested the AsyncDigestSigner traits in a wasm project, however I have confirmed that I'm able to build the project with the digest feature enabled.

[tarcieri]

Did you mean to remove the 'static bound here?

[tarcieri]

Going to go ahead and merge this as we'd like to do some repo reorganization which impacts this crate

[QuinnWilton]

Thanks for merging it, and apologies for not getting back to you on the static bounds yesterday! I did intentionally remove them, but I probably should have split that into a separate commit.

I believe that the 'static bounds in the old code were actually superfluous, because all Send types are already 'static, and without the Send bound, requiring 'static would overconstrain the generic: I don't know whether that'd end up being a problem in practice though.

[tarcieri]

Oof, well it's out with the 'static bound, though it's an uncommonly used crate and we could probably yank it and cut a replacement release, or just bump the minor version again

[QuinnWilton]

No worries at all! I'll verify it still works with my code today or tomorrow, and let you know if the 'static bound gives any issues :)

It's just on the type of the signature itself, which is probably fine, since that'll mostly just be first-party types from rust-crypto anyway, which shouldn't really be holding any references to the stack anyway.

[tarcieri]

When Rust 1.75 drops we can transition away from the async-trait crate to Rust's native support for async traits, at which time the 'static bound can potentially be removed