[parallel-queries] Make CrateMetadata::cnum_map immutable

[michaelwoerister]

The CrateMetadata::cnum_map field is wrapped in a Lock but it does not seem to be mutated anywhere. Less mutable state is always better. To solve this task, remove the Lock and make things compile again.

cc @rust-lang/wg-compiler-performance

[rleungx]

It seems that cnum_map field is mutated here, right?

[michaelwoerister]

@rleungx Oh, you are right of course. In that case, I'll remove the E-easy tag. It's still worth looking into though.

[michaelwoerister]

#50532 implements a partial solution. We still have a lock, but its around data that is accessed a lot less.

[wesleywiser]

Is this still a partial solution?

…

On Thu, May 10, 2018, 10:04 PM bors ***@***.***> wrote: Closed #50502 <#50502> via #50532 <#50532>. — You are receiving this because you are on a team that was mentioned. Reply to this email directly, view it on GitHub <#50502 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAyu2OScGgVW0_UzuzJmTDaTQ2CSNRt4ks5txPGkgaJpZM4T1EXS> .

[michaelwoerister]

It's partial (because it does not reduce the amount of shared mutable state) but it addresses my main concern by not requiring to keep the cnum_map in a mutex. So it's fine to close this issue, I think.

Removing more mutable state from CrateMetadata would probably require a bigger refactoring.