This also fixes some failures due to `broadcast` and `update_contacts_for_users`
being fallible. As part of this commit, these two functions don't return `Result`
anymore: the reason for this change is that we don't want a request to fail only
because a peer disconnected while we were trying to broadcast a message to them.
We forgot to observe the footprint of the reconnecting replica's prior
undos into the local clock. This could cause the replica to generate
edits with a version strictly smaller than what other peers may have
observed. As such, those peers would think they had already seen those
edits and skip them.
Over time, I think we may end up having multiple services, so it seems like a good opportunity to name this one more specifically while the cost is low. It just seems like naming it "zed" and "zed-server" leaves it a bit open ended.
We have an hypothesis that the server gets stuck while processing
an incoming message, either because the buffer fills up or because
a handler never completes. This should mitigate that and, once we
add logging, give us some clue as to what is causing it exactly.
Co-Authored-By: Nathan Sobo <nathan@zed.dev>
...in contrast to doing so during worktree registration. This fixes a
randomized test failure which would panic because store invariants would be
violated. This would happen when a peer disconnected before it had a chance
to register a worktree because, when removing all the state associated with
that peer upon disconnection, we would notice the registered project without
however finding it in the peer's connection state.
This fixes an error in the randomized test that would cause the future
returned from `Worktree::share` to never finish due to a bug in `postage`
that causes its waker to not be notified upon drop.
Also, allow arbitrary types to be used as Actions via the impl_actions macro
Co-authored-by: Nathan Sobo <nathan@zed.dev>
Co-authored-by: Keith Simmons <keith@zed.dev>
This is in contrast to not dispatching the event altogether in `Flex` when
the event is not contained in the flex element bounds. That approach was
problematic because it didn't give an opportunity to `MouseEventHandler`s
to handle mouse move events when they didn't intersect with the element bounds,
causing elements to never clear their hover state, cursor style, etc.
