fix: propagate monitor errors to the frontend

[kittaakos]

Motivation

This PR enables clear communication of the problem when starting a monitor process fails.

Shows monitor connection status in the <input>

• not-connected (board or port is missing),
• connecting (when trying to connect to the monitor port),
• connected ✅, and
• error (the connection has failed for whatever reason, and the user should select another board or close + reopen the widget after fixing any port issues outside of IDE2)

1508.mp4

Change description

• Propagate error from backend to frontend via the monitor ws.

• Open the ws connection before creating the monitor duplex. Otherwise, the status update or error cannot reach the frontend.

• Fixed monitor widget DI.

• Fixed monitor service lifecycle inside the widget. onBeforeAttach happened multiple times on IDE2 start if the widget was previously opened.

• Fixed connect when creating the ws connection to the monitor. Previously, the promise might have been resolved before the ws OPEN event.

• Moved monitor state to common.

• Monitor client proxy waits for reconciled boards after the app startup, then starts the monitor.

• Replaced the connected: boolean with a fine-grained connectionStatus and kept the code backward compatible with the plotter app.

• Add translation for baud rate.

• Smoother monitor widget re-render on monitor reset. (On monitor reset, the monitor widget is forcefully re-rendered causing the <input> and <select>s to disappear for a short time #1985)

1985.mp4

• When there is no monitor connection, the <input> is readOnly and keeps the user input. (The monitor <input> is not readOnly when there is no connection to the monitor #1984)

1984.mp4

• The monitor gracefully handles when the board's platform is not installed. (Serial monitor throws error when trying to retrieve the enumerateMonitorPortSettings and the platform is not installed #1974)

1974.mp4

• Set the monitor widget header color and shifted a few pixels here and there to provide a better monitor UI until a final design is ready as part of Maintain border on Serial Monitor input field when it loses focus #682.

682.mp4

• Aligned HC warning colors with the error colors:

  2.0.4:
  

  PR:
  

  

Other information

Closes #1508
Closes #1985
Closes #1984
Closes #1974
Ref #682

Reviewer checklist

• PR addresses a single concern.
• The PR has no duplicates (please search among the Pull Requests before creating one)
• PR title and description are properly filled.
• Docs have been added / updated (for bug fixes / features)

[kittaakos]

I have verified the followings:

• macOS
  • upload to the same board while two connected monitors are running (with Uno and 33 BLE),
  • start monitor when another process is using the port, such as Arduino IDE 1.x.
• Windows
  • upload to the same board while two connected monitors are running (with Uno and Micro),
  • start monitor when another process is using the port, such as Arduino IDE 1.x.

I could not upload to 33 BLE on Windows. I restarted my notebook and ran the CMD.exe as an Administrator, but it did not help:

c:\Users\kittaakos\Desktop\cli 0.31.0>arduino-cli.exe version
arduino-cli.exe  Version: 0.31.0 Commit: 940c9457 Date: 2023-02-22T16:15:54Z

c:\Users\kittaakos\Desktop\cli 0.31.0>arduino-cli.exe core list
ID                Installed Latest Name
arduino:avr       1.8.6     1.8.6  Arduino AVR Boards
arduino:mbed_nano 4.0.2     4.0.2  Arduino Mbed OS Nano Boards


c:\Users\kittaakos\Desktop\cli 0.31.0>arduino-cli.exe board list
Port Protocol Type              Board Name          FQBN                        Core
COM3 serial   Serial Port (USB) Arduino Nano 33 BLE arduino:mbed_nano:nano33ble arduino:mbed_nano
COM4 serial   Serial Port (USB) Unknown


c:\Users\kittaakos\Desktop\cli 0.31.0>arduino-cli.exe compile -b arduino:mbed_nano:nano33ble "..\\..\\Desktop\\sketchbooks\\test_sketchbook__2\\monitor__1_1508"
Sketch uses 84680 bytes (8%) of program storage space. Maximum is 983040 bytes.
Global variables use 44560 bytes (16%) of dynamic memory, leaving 217584 bytes for local variables. Maximum is 262144 bytes.

Used platform     Version Path
arduino:mbed_nano 4.0.2   C:\Users\kittaakos\AppData\Local\Arduino15\packages\arduino\hardware\mbed_nano\4.0.2

c:\Users\kittaakos\Desktop\cli 0.31.0>arduino-cli.exe upload -b arduino:mbed_nano:nano33ble -p COM3 "..\\..\\Desktop\\sketchbooks\\test_sketchbook__2\\monitor__1_1508"
TOUCH: error during reset: opening port at 1200bps: Serial port busy
No device found on COM3
Failed uploading: uploading error: exit status 1

c:\Users\kittaakos\Desktop\cli 0.31.0>arduino-cli.exe upload -b arduino:mbed_nano:nano33ble -p COM3 "..\\..\\Desktop\\sketchbooks\\test_sketchbook__2\\monitor__1_1508" -v
Performing 1200-bps touch reset on serial port COM3
TOUCH: error during reset: opening port at 1200bps: Serial port busy
Waiting for upload port...
No upload port found, using COM3 as fallback
"C:\Users\kittaakos\AppData\Local\Arduino15\packages\arduino\tools\bossac\1.9.1-arduino2/bossac.exe" -d --port=COM3 -U -i -e -w "C:\Users\kittaakos\AppData\Local\Temp\arduino\sketches\9374601C8719AF4D46679576E3DD0DC6/monitor__1_1508.ino.bin" -R
No device found on COM3
Failed uploading: uploading error: exit status 1

c:\Users\kittaakos\Desktop\cli 0.31.0>

[per1234]

When I perform the steps described in #1508 I get an incorrect message in the input field of the Serial Monitor view:

The problem that caused the monitor initialization to fail is that I had the port open in another application. I already have the correct board and port selected, so this "Select a board and a port to connect automatically." advice is wrong.

[kittaakos]

The problem that caused the monitor initialization to fail is that I had the port open in another application. I already have the correct board and port selected, so this "Select a board and a port to connect automatically." advice is wrong.

@91volt, please help with how this should work. Thank you!

[per1234]

When I perform the steps described in #1508, there is a delay of ~15 seconds before the "Port monitor error: command 'open' failed: Serial port busy." notification appears.

The error appears in the logs immediately upon opening the Serial Monitor view:

2023-03-24T07:58:20.085Z monitor-service ERROR Error: 2 UNKNOWN: Port monitor error: command 'open' failed: Serial port busy
    at Object.callErrorFromStatus (E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\@grpc\grpc-js\build\src\call.js:31:19)
    at Object.onReceiveStatus (E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\@grpc\grpc-js\build\src\client.js:413:49)
    at Object.onReceiveStatus (E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\@grpc\grpc-js\build\src\client-interceptors.js:328:181)
    at E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\@grpc\grpc-js\build\src\call-stream.js:187:78
    at processTicksAndRejections (node:internal/process/task_queues:78:11)
for call at
    at ServiceClientImpl.makeBidiStreamRequest (E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\@grpc\grpc-js\build\src\client.js:398:30)
    at ServiceClientImpl.monitor (E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\@grpc\grpc-js\build\src\make-client.js:105:19)
    at MonitorService.createDuplex (E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\arduino-ide-extension\lib\node\monitor-service.js:157:34)
    at runMicrotasks (<anonymous>)
    at processTicksAndRejections (node:internal/process/task_queues:96:5)
    at async E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\arduino-ide-extension\lib\node\monitor-service.js:220:37
    at async retry (E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\@theia\core\lib\common\promise-util.js:73:20)
    at async MonitorService.start (E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\arduino-ide-extension\lib\node\monitor-service.js:135:13)
    at async MonitorManager.startMonitor (E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\arduino-ide-extension\lib\node\monitor-manager.js:89:13)
    at async JsonRpcProxyFactory.onRequest (E:\incoming\review\1965\1-c0ca295\resources\app\node_modules\@theia\core\lib\common\messaging\proxy-factory.js:132:24)

Then it retries the monitor initialization 9 more times before finally showing the notification.

[kittaakos]

The error appears in the logs immediately upon opening the Serial Monitor view:

Nothing new. See here: #1966.

Then it retries the monitor initialization 9 more times before finally showing the notification.

This is precisely the same behavior as on the main with 9b49712. Nothing has changed:

arduino-ide/arduino-ide-extension/src/node/monitor-service.ts

Line 38 in 9b49712

const MAX_WRITE_TO_STREAM_TRIES = 10;

arduino-ide/arduino-ide-extension/src/node/monitor-service.ts

Lines 362 to 375 in 9b49712

	// if "pollingIsSuccessful" is false
	// the CLI gave us an error, lets try again
	// after waiting 2 seconds if we've not already
	// reached MAX_WRITE_TO_STREAM_TRIES
	if (pollingIsSuccessful === false) {
	attemptsRemaining -= 1;
	if (attemptsRemaining > 0) {
	setTimeout(startPolling, 2000);
	return;
	} else {
	resolve(false);
	return;
	}
	}

[per1234]

Nothing new.

I didn't intend to imply it was something new. I mentioned it because it shows that the delay does not have an external cause. Arduino IDE knows right away that the monitor initialization failed, but doesn't immediately show the notification.

This is precisely the same behavior as on the main with 9b49712. Nothing has changed:

Something has changed. Arduino IDE now shows a notification that the monitor initialization failed. There is no notification at all on main.

The problem is there is an unnecessarily long delay before the notification appears.

[kittaakos]

I made the changes. Please review. Thanks

[AlbyIanna]

Code looks good to me! I just left some remarks, but they're definitely not blocking.

P.S.: for this review I focused mainly on the code and only did some basic tests of the functionality; let me know if you want me to test it more deeply.

[AlbyIanna]

I just left some remarks, but they're definitely not blocking.

Follow up on this: all comments have been addressed. Thank you, Akos!