[cgiapp] Data::Session/CGI::Session

[Mark Stosberg]

On 12/18/2010 06:37 PM, Nicholas Bamber wrote:
> I notice that a 4.43 o CGI::Session has been released. That should get 
> into Debian in due course. Reviewing the Debian bug list I noticed this:
> 
> "Session file not being written for driver:File; serializer:Storable" 
> (http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=315669)
> 
> That appears to be related to #482355
> "Storable.pm segfaults when called during global destruction"
> 
> And from my brief reading of the bug reports this seems to be related to 
> design/backwards compatibility issues
> in both CGI::Session and Storable. Can I conclude from that, that the 
> best thing to do with #315669
> would be to mark it "wontfix" and to suggest that people bother by the 
> bug try Data::Session (which would of course will need to be packaged)?

I would say that people bothered by the issue should explicitly call
flush(). If the issue is triggered by call to flush() in a DESTROY
subroutine, than unfortunately Data::Session won't help-- Data::Session
kept that part of the design:

    sub DESTROY {
        my $self  = @_;
        $self->flush;
    }

I would eliminate the DESTROY method in my own version of re-write. As a
young module, Data::Session could still reasonable choose to do that,
but currently I don't think Data::Session can seen as "fix" for this
issue. A head-to-head to test with both of them could confirm this.

    Mark