Skip Menu |

This queue is for tickets about the Sys-SigAction CPAN distribution.

Report information
The Basics
Id: 39998
Status: resolved
Priority: 0/
Queue: Sys-SigAction

People
Owner: lab [...] lincolnbaxter.com.MAKE.ME.VALID
Requestors: lab [...] lincolnbaxter.com
Cc:
AdminCc:

Bug Information
Severity: (no value)
Broken in: (no value)
Fixed in: (no value)



CC: bug-Sys-SigAction [...] rt.cpan.org
Subject: Re: Sys::SigAction and "safe" attribute.
Date: Sun, 12 Oct 2008 22:21:04 -0400
To: krivenok [...] orangesystem.ru
From: "Lincoln A. Baxter" <lab [...] lincolnbaxter.com>
Hi Dmitru, See below. On Fri, 2008-10-10 at 16:44 +0400, Dmitry V. Krivenok wrote: Show quoted text
> Hello! > > Doc (http://search.cpan.org/~lbaxter/Sys-SigAction-0.10/lib/Sys/SigAction.pm) says that > it's possible to pass "safe" attribute in set_sig_handler: > > my $h = set_sig_handler( 'ALRM' ,\&mysubname ,{ mask=>[ 'ALRM' ] ,safe=>1 } ); > > However, I found (in SigAction.pm version 0.10) that this attribute is > ignored (it is just checked in _attrs_warning subroutine). > > Why not to add > > $act->safe( $attrs->{safe} ); > > after > > my $act = POSIX::SigAction->new( $handler ,$mask ,$attrs->{flags} ); > > if perl >= 5.8.2 is used? >
See the note at the bottom of the test, t/safe.t I spent some time on it, and I could not come up with a way to prove it works. I believe it doesn't, but I would be happy to be shown where my confusion is, if that is the case. That test is an invitation to anyone who wants it badly enough, to show me how, and I will happily release a new version. Lincoln Show quoted text
> > Currently we always use unsafe signals model (despite of safe attribute). > > > Thank you beforehand! > >
This was my email response to bug 39599... On looking closer and testing with a clone of mask.t in safe.t I believe the suggested patch works... I am working on a release of 0.11 now. :-)