`Either` uses `ValidateNapiValue` + `TypeName` to validate and report error on value not being matched. So there's no way to remove these super traits from it. So I implemented these types to `Tuple` types.
As of before this commit, there was a lock in the codegen preventing Reference
from being used as a function argument outside of a Reference<Self>.
This changes it, allowing Reference of any class to be added as a class method
argument anywhere. It has the same limitations as reference, as in it requires
the class to have been created with a factory or constructor. This change
implements FromNapiValue on Reference, which will unwrap the class and call the
existing from_value_ptr method. It also updated typegen so that we only emit
the reference type if we're in an impl block that doesn't match the Reference
we're getting. This ensures that typegen works as expected with the previous
behaviour.
For many reasons, in an app we can end up in a scenario where the thread / task that initially called the ThreadsafeFunction gets interrupted. This means that the receiver would get collected, should a ThreadsafeFunction be awaited on the said thread. This will create an error when the ThreadsafeFunction's callback will be called, and call a napi_fatal_error.
This change makes it so that if the send errors, this error is hidden and thus prevents any hard failure. If a ThreadsafeFunction is called in such a case, its output won't be used, but we'll still run all the logic to ensure it ran properly.
Fixes https://github.com/napi-rs/napi-rs/issues/1665 - cc @Brooooooklyn
There is a piece of custom logic that has been added a while back to ensure
that Buffers can be sent across threads, and be dropped properly. This involves
a custom GC that runs on NodeJS's current thread (per my understanding). The
logic to drop the buffer on that custom GC differed from the one in the Drop
impl. This meant that everytime Node sent a buffer back to a napi-rs function,
the reference wouldn't be cleaned up properly, and it would leak (96 bytes per
Reference on an ARM MacOS machine).
This commit updates the logic in the custom GC so that it matches the one in
the Drop impl. This worked locally, and fixed any occurence of the leak I could
find.
This seems to be a relic of when napi-rs used channels to send data to the
tokio runtime. This removes the outdated doc, as this could be misleading if
one wants to understand what those channels are about.
