Re: [PATCH 0/4] Fix crash in "set python" and small cleanups

[Philippe Waroquiers <philippe.waroquiers@skynet.be>]

On Tue, 2021-05-18 at 08:50 -0600, Tom Tromey wrote:
> > > I didn't add a specific test because what's the chance of "set python"
> > > regressing?
> 
> Andrew> Don't we know the answer to that is 100% - it's currently regressed,
> Andrew> right?
> 
> I think he meant the chances of it happening again.
> 
> Anyway, Marco, in gdb it is normal to add tests for things like this.
> It can just test "set python" and have a comment explaining that, once
> upon a time, this caused a crash.
> 
> thanks,
> Tom
IIRC, it has happened already for a few other commands that launching a command
without argument caused a SEGV or similar.

So, maybe what could be done is a test that calls all set and show commands 
with something like
  (gdb) | help show | grep -e -- | sed -e 's/--.*$//' -e 's/,.*//' > launch_all_show
  # the second sed expression cleans up  'show check, show ch, show c'
  (gdb) source launch_all_show
and similar for the set.

However, the launch_all_set quickly stops on an error
as the set commands need some arguments
but that can be solved as soon as we have
  with ignore-errors -- source launch_all_set
or
  ignore-errors source launch_all_set
:)

With the above, we can verify automatically that all set and all show commands are not
crashing gdb.

Philippe