Closed jnhyatt closed 3 weeks ago
Investigating the CI failures. Added a changelog entry but apparently there's a future that's not sync in linux
After some investigation, I found that adding sync bounds is not only non-trivial, it's probably totally unnecessary. Closing this out
Trying to use async
save_file
and co. in a context that requires my futures to beSend
andSync
and getting errors. Turns out theFuture
returned bysave_file
is alreadySend
, but the signature ofsave_file
only guaranteesimpl Future<...>
. We can get theSend
guarantee for free by specifying a specific type for the future returned bysave_file
(and I did the others for consistency).The change I'm less certain on is adding
Sync
toDialogFutureType
's inner boxed type. It all builds, so I don't think there's a problem with that (haven't tested on wasm) and I don't even think anything I did is a breaking change. If you have a suite of tests to put this through, though, that probably wouldn't be a bad idea.PS Just looked and I don't think anything I did should affect wasm since
DialogFutureType
isn'tSend
on wasm, and I didn't add any trait bounds for that config. So unless I'm way more tired than I'm realizing, this PR shouldn't break anything! 🎉