Attempt to remove YokeTraitHack

[sffc]

I get a lot further than before, but there's still an error on StructProvider.

[jira-pull-request-webhook]

Notice: the branch changed across the force-push!

• provider/core/src/data_provider.rs is different
• utils/yoke/src/yoke.rs is different

View Diff Across Force-Push

~ Your Friendly Jira-GitHub PR Checker Bot

[jira-pull-request-webhook]

Notice: the branch changed across the force-push!

• provider/core/src/struct_provider.rs is different

View Diff Across Force-Push

~ Your Friendly Jira-GitHub PR Checker Bot

[sffc]

Blocked on rust-lang/rust#89196

[sffc]

Based on @Manishearth's respose rust-lang/rust#89196 (comment), it seems unlikely that we will be able to get rid of YokeTraitHack, unless it turns out that the inference bug is solvable.

To be clear, the issue impacts virtually every call site of DataProvider::load_payload. The cost of fully qualifying every call site of DataPayload::load_payload is much higher than the cost of keeping YokeTraitHack.

Shall I close this PR and open a new bug to track the new Rust bug in the chance that it is fixed at some point?

[Manishearth]

@sffc sounds good. We can still remove the _with_capture and other hacks we had to add.

I was a bit worried about this when introducing DataMarker (because of the kind of indirection it introduces) but on balance I think having to use the trait hack is better than not having DataMarker's ergonomics

[sffc]

This PR is superseded by #1134

#1061 is the issue to track removing the workarounds for Rust bugs.

I am attempting to remove _with_capture in #1133 but I hit another issue, which is explained over there.