It turns out that
IF you have a chunk of code that sends stuff through the Xojo Firewall
AND you open the firewall via a subroutine where the FW is foolishly declared locally
THEN the FW may indeed open for a little while (time not specified, probably some few seconds)
…
Your code may have a short time window in which to send stuff through the FW, leading you to think your port is properly open
…
AND when you eventually close the firewall, via another subroutine where the FW is (again) foolishly declared locally
THEN the Linux console app at exit dies with a segmentation fault instead of exiting normally.
The Firewall port opening behaviour is documented; i.e. that the FWP must remain in scope for the duration of interest, or the port will self close and cause grief.
The segmentation fault on exit is not documented as far as I can see. However, it is evident that there are many ways to kill things, and only a few to allow them to live and work in peace. A list of things not to do would be very large, and perhaps not be of use to anyone except the morbidly curious.
Thanks to Greg O’Lone, Jason Parsley, Christian Schmitz, and others for suggestions.
[quote=300882:@Tony Barry]
THEN the Linux console app at exit dies with a segmentation fault instead of exiting normally.[/quote]
I’d like to see an example of this in a feedback case because this should not be happening.