Re: [PATCH 3/3] Fix control-c handling on Windows

[Hannes Domani <ssbssa@yahoo.de>]

 Am Montag, 12. Dezember 2022 um 18:06:23 MEZ hat Tom Tromey <tromey@adacore.com> Folgendes geschrieben:

> Tom> I used RDP and started powershell, then ran gdb there.  I did 6 tests: a
> Tom> C-c and a C-break test of "run", "run" with "set new-console 1", and
> Tom> "attach" -- and in all these cases, it worked.
>
> I found out how to reproduce the crash you saw.  In my tests, I was
> doing everything in a single gdb instance.  However, if I start a new
> gdb, then "set new-console 1", and then "run", I can make it crash.
>
> Changing sigint_ours to a gdb::optional fixes this problem.
>
> I couldn't reproduce your problem #3 ... my python-less build seems to
> work fine in other situations.
>
> I'll send v2 of this series shortly.

I also can't reproduce problem #3 any longer with the v2 series.
So I think that this was actually the same problem that you fixed with
gdb::optional, and it was just that the different signal() calls confused me.

So, I also have no problems any more with v2 applied.


Thanks, Hannes