Repository navigation
chore: port errors, connection, events and waiter from the Node.js client - #2055
Conversation
Connection and ChannelOwner now follow the member order of the upstream files: object factories, sendMessageToServer with CallOptions, close(cause) rejecting the pending calls, wrapApiCall with an ApiZone, dispose(reason), setEventToSubscriptionMapping and updateSubscription.
…ol import lost in the restack
|
One issue worth fixing here; the rest of the port looks sound.
Not a regression (on main the same path threw from CI: the only red job is |
A closed connection made off() throw from the dispose of a wait and replaced the error of the wait.
|
Fixed in f6d1847: |
On Firefox the pipe closes before the pages report their close, and the pump threw 'Browser closed' from the wait instead of the close error of the waiter.
connect() waited forever once the websocket was open. The three fixmes for it passed whenever the websocket took longer than the timeout to open, which failed the run.
8ffd0b5
into
microsoft:main
Summary
Errors,Connection,ChannelOwner,Events,EventEmitter,TimeoutSettingsandWaiternow followerrors.ts,connection.ts,channelOwner.ts,events.ts,eventEmitter.ts,timeoutSettings.tsandwaiter.ts;// java:comments mark the Java-specific spots.waitFor*methods go throughWaiter, and events are emitted under the upstream event names.PlaywrightException, and a navigation error fails the wait regardless of the URL matcher. Fivejava:fixmes in the ported tests are dropped.