RE: patch for PoCo::IKC::ClientLite.pm
Philip Gwyn <[email protected]>
| Newsgroups | gmane.comp.lang.perl.poe |
|---|---|
| Message-ID | <[email protected]> |
Thank you for the patch.
I do have a question though : could the alarm(0) be moved into the if($@) {
block? Would this be cleaner?
On 14-Oct-2009 flw wrote:
> When it working with a server which used Non-IKC protocol,
> alarm( $self->{connect_timeout} ) will break user process via
> SIGALRM(default behaviour)
>
> Following patch is based on:
> $Id: ClientLite.pm 495 2009-05-08 19:46:42Z fil $
>
>
> --- ClientLite.pm.orig 2009-10-14 11:35:38.000000000 +0800
> +++ ClientLite.pm 2009-10-14 11:36:06.000000000 +0800
> @@ -136,11 +136,11 @@
> alarm( $self->{connect_timeout} );
> $sock->print('WORLD'); # phase 003
> chomp($resp=$sock->getline);
> - alarm( 0 );
> die "Phase 003: $!\n" unless defined $resp;
> die "Didn't get UP from $name\n" unless $resp eq 'UP';
> $DONE = 1;
> };
> + alarm( 0 );
> if($@)
> {
> $self->{error}=$error=$@;