Occasionally print "[[O" or "[[I" abnormally on the screen

[fengyichui]

Hi,

When I use tmux, occasionally print "[[O" or "[[I" abnormally on the screen. And Windows Terminal do not has this issue.

What can I do to help this?

[mintty]

These are the focus report escape sequences. They are only sent if explicitly enabled with a DECSET 1004 control sequence. If a program enables these reports and does not catch them properly, it is a bug in that program.
There may be a race condition if a program disables the reporting (and does not expect it anymore) while a report was already triggered. Please report to tmux.

[fengyichui]

I have reported to tmux, and we check the debug log, tmux just received literally "[[O". The authors of TMUX think this may be a WSL bug.
I tested with Windows Terminal for a long time and couldn't find this issue

[mintty]

So there is a tmux issue? Please provide its link.

[fengyichui]

Here: tmux/tmux#3601

[fengyichui]

I use v3.2.0 cygwin1.dll, this issue doesn't seem to happen anymore. cygwin_3.2.0.zip

[mintty]

Thank you for the information, that is very interesting indeed.
I was trying to test with mintty versions before reporting to cygwin, and it's quite bothersome to set up a reliable test case for this. What is your current test scenario? How often would it happen?
Anyway, if we can establish that the issue does not occur with cygwin 3.2.0, I will prepare a report.
Would be good also to work out which cygwin version between 3.2.0 and current 3.4.7 introduced a bug.

[fengyichui]

Today I stumbled upon a backup of my old cygwin and gave it a try.

My test environment still uses the tinytask tool to automate mouse clicks, and the frequency of these clicks is about tens or hundreds of times.

[mintty]

It does not seem to happen with the latest snapshot from http://cygwin.com/snapshots/ which is a 3.4.0 patch version.
Neither does it seem to happen with 3.5.0 which can be installed as a test version.
Please check that.

[fengyichui]

I test latest snapshot, this issue is happened.
I test some older snapshot: cygwin1-20220217.dll is bad, cygwin1-20220127.dll seems good:
cygwin1-20220127-good.zip
cygwin1-20220217_bad.zip

[mintty]

Hmm, so the newer one is good but the current one bad again? This does not appear consistent, maybe it's just the testing which did not expose the issue because it appears so rarely.
Testing 3.5.0 would be helpful.
I've tried to create an automatic test scenario to improve the process, which would get feedback from the tracing application to stop once the issue occurs, so you wouldn't have to fiddle interactively with that tool. But it's tricky to build something appropriate...

[fengyichui]

I also tested cygwin-3.5.0-0.341.dll, it is bad.
It seems that since the cygwin1-20220217.dll version, there is a problem.
Today I'll work with version cygwin1-20220127.dll to see if this version is really good.

[mintty]

The issue is fixed with the latest cygwin test release 3.5.0-0.370, to be released with 3.5.0 (later this year) and 3.4.8.

[mintty]

Cygwin 3.4.8 has been released, with this fix included.

[mintty]

Reopening as this is a wsltty issue, sorry. Next release to come soon.

[mintty]

Released 3.6.5.