Hallo Michael,
>> but I'd advise that you split it in (1) progress and (2) signal
>> toggling so that the first part is more likely to make it before 12
>> freeze.
>
> Ok, done so in the attached.
Fine.
I think that it is good to show the overall impact of the signal stuff, in
particular the fact that the size must always be computed if the progress
may be activated.
Also, I'd suggest to add "volatile" on the show_progress variable in the
second patch.
--
Fabien.