Bug #89538 for MCE: __DIE__ handler is not always correct

> I haven't tried the new MCE 1.5, but I've found the previous versions > to be very useful. It's a very nice interface for running pool of > workers. Thanks for that! > > However, I think the way MCE handles exceptions from the children > isn't entirely correct. It seems to rely on $^S to decide if the > exception is going to be trapped or not. But that isn't very > reliable. Observe: > > $SIG{__DIE__} = sub { print $^S }; > eval 'die'; # Prints 1 > eval 'use BogusModule'; # Prints nothing > > So if the MCE workers try to load a module that doesn't exist, then > the child exits completely even if I wrap it all in eval. > > perlvar advises against using $^S and suggests using an END block > instead. I think that might be a more reliable (and natural) solution > here.

> Tue Oct 15 16:40:22 2013: Request 89538 was acted upon. > Transaction: Ticket created by THALJEF > Queue: MCE > Subject: __DIE__ handler is not always correct > Broken in: 1.415 > Severity: (no value) > Owner: Nobody > Requestors: thaljef@cpan.org > Status: new > Ticket <URL: https://rt.cpan.org/Ticket/Display.html?id=89538 > > > > I haven't tried the new MCE 1.5, but I've found the previous versions to be very useful. It's a very nice interface for running pool of workers. Thanks for that! > > However, I think the way MCE handles exceptions from the children isn't entirely correct. It seems to rely on $^S to decide if the exception is going to be trapped or not. But that isn't very reliable. Observe: > > $SIG{__DIE__} = sub { print $^S }; > eval 'die'; # Prints 1 > eval 'use BogusModule'; # Prints nothing > > So if the MCE workers try to load a module that doesn't exist, then the child exits completely even if I wrap it all in eval. > > perlvar advises against using $^S and suggests using an END block instead. I think that might be a more reliable (and natural) solution here.