[Catalyst] RFC: C::P::Alarm (was Re: timeouts)

[Nilson Santos Figueiredo Junior]

On 5/4/06, Peter Karman <peter at peknet.com> wrote:
> It's not alarm() that's the issue, iirc. C::P::Alarm uses the CORE
> alarm() function. I see that I wasn't clear enough about that in the
> BUGS section of the pod.

alarm() has some restrictions under Win32, since it's emulated (as fork() is).
I couldn't find these specific restrictions right now, but I vaguely
remember it having something related to fork() (maybe because both are
emulated). So there are alarm() specific issues besides the whole
signal issues under Win32.

> I would be willing to include a win32-specific check to avoid using
> Sys::SigAction on those systems. Instead, I could use the straight up
> $SIG{ALRM} syntax (which is what Yuval originally posted on this topic).
>
> Of course, I would need a win32 user like yourself to test. :)

Well. When it's ready I'll gladly report wether it works or not.

> What's the currently preferred way of testing for OS platform these
> days? Is it still just $^O  ?

Hmm... I guess so. $^O eq 'MSWin32' should suffice.

-Nilson Santos F. Jr.