Performance issue and websocket error

I just upgraded to version 1.0.0 and immediately noticed that my test suite takes much longer to run, now about 18 seconds when it finished in about 10 seconds before. My guess is there is something slow going on during startup because the first test result takes much longer to show up as done.

Even worse, every other time or so that I run my tests I get the following error during startup:

C:\path_to_project\node_modules\playwright-chromium\lib\server\chromium.js:211
            session.queue.push(parsedMessage);
                          ^

TypeError: Cannot read property 'push' of undefined
    at WebSocket.<anonymous> (C:\path_to_project\node_modules\playwright-chromium\lib\server\chromium.js:211:27)
    at WebSocket.emit (events.js:315:20)
    at Receiver.receiverOnMessage (C:\path_to_project\node_modules\ws\lib\websocket.js:789:20)
    at Receiver.emit (events.js:315:20)
    at Receiver.dataMessage (C:\path_to_project\node_modules\ws\lib\receiver.js:422:14)
    at Receiver.getData (C:\path_to_project\node_modules\ws\lib\receiver.js:352:17)
    at Receiver.startLoop (C:\path_to_project\node_modules\ws\lib\receiver.js:138:22)
    at Receiver._write (C:\path_to_project\node_modules\ws\lib\receiver.js:74:10)
    at writeOrBuffer (_stream_writable.js:352:12)
    at Receiver.Writable.write (_stream_writable.js:303:10)

This is not an error that I have seen before and when I revert to version 0.2.5 these issues go away.

[olsch]

@mmarkelov I can verify that the next version of jest-playwright-preset (currently 1.3.0-rc1) together with next version of playwright-chromium (1.2.0-next.1594082323189) solves the performance issue that @DavidZidar reported above (#158) as long as you use launchType: 'LAUNCH'. Also, the crash issue appears to be gone. When I run my tests (8 test suites / 58 tests) with launchType: 'PERSISTENT' the behaviour is erratic. Only one of 5+ testruns finished successfully. The others were either hanging after the last testrun or crashing with the message A worker process has failed to exit gracefully and has been force exited. This is likely caused by tests leaking due to improper teardown.. When I tried launchType: 'PERSISTENT' with v1.2.1 I didn’t have any of these problems. Running with launchType: 'SERVER' works fine but is very slow. BTW, I’m a colleague of @DavidZidar.

[DavidZidar]

Just tried upgrading to playwright-chromium@next and I can confirm the crashing issues I’ve had are gone! Great success.

It is still running much slower than before though, but that’s not as bad as outright crashing.