On Monday October 21 2013 9:09:47 AM Marc Lehmann wrote:
> On Sun, Oct 20, 2013 at 01:45:53PM -0600, Darin McBride 
<[email protected]> wrote:
> > I'm having a hard time trying to get AE/EV/Coro working under Perl 5.18.1,
> > not
> I haven't had time to port my modules to 5.18 - the new speed and the
> amount of breakage introduced makes it very hard for me to follow every
> release.
> 
> It's quite possible that something has changed incompatibly that causes
> trouble - I assume this does not happen when the program is about to exit
> (in that case, it's likely the longstanding data corruption bug during
> global destruction).

Ah, that's it for AE::DBI - global destruction.  Put some braces in, and that 
particular problem went away.  That should make it easier to replicate the 
original Coro problem, which definitely was not during global destruction.

And then, once I do that, I might want to move back to 5.16.3, just so this 
all works until you've had a chance to get it all going again.  Which is 
unfortunate, but that's just the way things go.

Note that this all passes all your .t files, it all installs fine.  I'm just 
hitting some sort of odd issue in my own tests that cause a segfault in 
Perl_pad_push when called by Coro's State.xs file somewhere (I don't recall 
where offhand but can produce a backtrace in gdb).

_______________________________________________
anyevent mailing list
[email protected]
http://lists.schmorp.de/mailman/listinfo/anyevent

Reply via email to