[Bug] EV3 WebUSB connects but 'Run program' and 'REPL' buttons stay disabled in Chrome on Ubuntu

[BertLindeman]

Describe the bug
Chrome successfully connects to the EV3 over WebUSB WebSerial .
The connect button shows the hub as connected, and chrome://device-log shows normal USB detection:

USB device added: LEGO MINDSTORMS EV3 + Pybricks
Serial device added: /dev/ttyACM0

However, the "Run program" and "Start REPL" buttons sometimes remain disabled indefinitely.

To reproduce
Steps to reproduce the behavior:

1. Run google-chrome 150 on Ubuntu Trixie
2. Go to "beta.pybricks.com"
3. connect an EV3 (with current beta firmware)
4. See the USB connect button show there is a connection
5. select a file (at the left side)
6. expect the "run program" and the REPL button be enabled.
7. Intermittent in separate sessions both buttons are disabled. And the last test they were enabled.

Expected behavior
Buttons to be enabled on an established USB connection.

Environment

• Ubuntu Trixie
• Google Chrome 150.0.7871.181
• beta.pybricks.com Pybricks Beta v3.1.0-beta.3
• EV3 firmware v4.1.0b2

Observations

• No USB errors are shown in chrome://device-log.
• No JavaScript errors are visible in the Developer Tools Console.
• The SharedArrayBuffer warning appears in both working and failing cases and does not seem related.
• The same Chrome installation works correctly when Chrome is displayed remotely through MobaXterm X11 forwarding.
• Running Chrome locally from the Ubuntu desktop menu and from a terminal uses the same executable.
• After repeating the test without changing anything, the REPL button eventually became enabled, so the problem may be intermittent.

This suggests that WebUSB WebSerial connection itself is working, but the Pybricks web application may not always complete the hub initialization state after connecting.
Race condition?
The traffic via X11 to remote is obviously slower and that seems to help.

Is there any additional logging or debugging information that would help identify where initialization stops or fails?

[dlech]

Chrome successfully connects to the EV3 over WebUSB.

Should be Web Serial now, not WebUSB. Log looks right though, so we know what you mean.

Race condition?

Sounds like it.

Is there any additional logging or debugging information that would help identify where initialization stops or fails?

Do you have the console set to "All levels"?

We might need to add some more debug prints if nothing shows up.

[BertLindeman]

Thanks David,

Of course Web Serial. . .

I did not have All levels in console set.
Will try to get the problem situation again with "All levels" set.
Hopefully Saturday. Friday is Digital Cafe day.

[laurensvalk]

Does the play/REPL button become enabled if you switch program slots on the EV3? This will cause it to send a status message again.

This will help us determine if the connection is broken or it just missed the very first status message from the hub.

[BertLindeman]

Can confirm that the chrome session did never use block coding.
Will try to recreate the situation and add a block program.

[laurensvalk]

I just mean pressing the up and down buttons on the EV3 in the file menu.

Nothing to do with blocks or Python; it's fine if there's nothing at all.

[BertLindeman]

I just mean pressing the up and down buttons on the EV3 in the file menu.

Nothing to do with blocks or Python; it's fine if there's nothing at all.

Sorry, mis-read. Understand now.

[KWSmit]

I just mean pressing the up and down buttons on the EV3 in the file menu.

Nothing to do with blocks or Python; it's fine if there's nothing at all.

Yes, that worked for me in Chrome, 'Run program' and 'REPL' buttons become enabled after pressing up or down in EV3-menu.

I also tried this with Chromium, because there the buttons are always disabled (issue #2770), but it didn't work. So that's probably another issue.