-
Notifications
You must be signed in to change notification settings - Fork 12.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Remove Windows function preloading #100127
Conversation
@ChrisDenton: no appropriate reviewer found, use r? to override |
Hey! It looks like you've submitted a new PR for the library teams! If this PR contains changes to any Examples of
|
r? libs |
I'd still like to review this one, then. r? @thomcc |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for doing this. I was pretty uncomfortable with relying on static init for this (and we deliberately don't do so on unix), but wasn't sure there was a good reason for that beyond users calling into std from an initializer.
Two issues:
- Orderings are wrong for lazy loading of these, and technically could result in a race between the call and the loader.
- This is a pretty leaky abstraction, and adds very tight coupling between
compat
and the windows thread parker. For now adding more comments is hopefully sufficient, but we probably should revisit it.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good with the updates.
Seems good to me! @bors r+ |
…iaskrgr Rollup of 5 pull requests Successful merges: - rust-lang#100071 (deps: dedupe `annotate-snippets` crate versions) - rust-lang#100127 (Remove Windows function preloading) - rust-lang#100130 (Avoid pointing out `return` span if it has nothing to do with type error) - rust-lang#100169 (Optimize `pointer::as_aligned_to`) - rust-lang#100175 (ascii -> ASCII in code comment) Failed merges: r? `@ghost` `@rustbot` modify labels: rollup
After @Mark-Simulacrum asked me to provide guidance for when optionally imported functions should be preloaded, I realised my justifications were now quite weak. I think the strongest argument that can be made is that it avoids some degree of nondeterminism when calling these functions (in as far as system API calls can be said to be deterministic). However, I don't think that's particularly convincing unless there's a real world use case where it matters. Further discussion with @thomcc has strengthened my feeling that preloading isn't really needed.
Note that
WaitOnAddress
needed some adjustment to work without preloading. I opted not to use a macro for this special case as it seemed silly to do so for just one thing (and I don't like macros tbh).