experiment: Require safety comments in all of std & alloc - #160824
Conversation
rustbot
added
O-android
nia-e
removed
O-windows
nia-e
added
O-windows
|
We have CODEOWNERS? |
|
we have configs for teams to get pinged when certain files are touched. close enough |
|
I think, for the sake of this being reasonable to merge at all, we should probably split this into multiple pieces. Also, perhaps starting with alloc would be best? |
|
|
|
I feel like offering up a PR with just
|
There was a problem hiding this comment.
This file is generated from windows-bindgen so shouldn't be edited. But the good news is that, since Default is now implemented on pointers, the unsafe will be going away in the next version of windows-bindgen.
| // Safety: We only touch lock_count when we own the inner mutex. | ||
| // Additionally, we only call `self.owner.set()` while holding | ||
| // the inner mutex, so no two threads can call it concurrently. | ||
| // SAFETY: Untriaged. |
There was a problem hiding this comment.
Two safety comments here now
There was a problem hiding this comment.
fair. I did all of this with just a regex, not intending to merge as-is ^^ there's very likely a lot of cases like this
| _ => { | ||
| // SAFETY: Untriaged. | ||
| unsafe { | ||
| cvt(libc::pipe(fds.as_mut_ptr()))?; | ||
|
|
There was a problem hiding this comment.
Formatting seems a little off here
There was a problem hiding this comment.
This is an unfortunate side-effect of macro formatting; not even indenting gets done right.
|
☔ The latest upstream changes (presumably #160954) made this pull request unmergeable. Please resolve the merge conflicts by rebasing. |
6 participants
cc @workingjubilee @joshtriplett, per discussion in Zulip.
this is... awful, but not as awful as expected :D it'd also cause a lot of conflicts. i'm scared of how many
CODEOWNERSmight be triggered