[rt-users] RT 3.8 beta 2 now available

[Tom Lanyon]

On 20/05/2008, at 5:42 PM, Tom Lanyon wrote:

> On 20/05/2008, at 2:05 PM, Jesse Vincent wrote:
>
>>>>
>>>> Jesse
>>>
>>> Jesse,
>>>
>>> It seg faults for me on a dev box, RedHat Enterprise Linux 5 with  
>>> Perl 5.10.0. This box was successfully running 3.6.6 with no  
>>> issues and only had a few dependencies which needed upgrading to  
>>> pass a 3.7.85 'make testdeps'.
>>>
>>> It dies in HTML::Mason::Request::alter_superclass(), called from  
>>> RT::Interface::Web::Request.
>>>
>>
>> When you install HTML::Mason, does it run and pass all its tests?
>
>
> Yes, all except live tests and CHI. If it was an issue with  
> HTML::Mason I'd expect 15-subclass.t to fail (as it calls the same  
> alter_superclass method).
>

To follow up my own email, this does seem like a HTML::Mason bug as I  
can reproduce the SIGSEGV on two machines running perl 5.10.0 and  
HTML::Mason 1.39 with the following test script, yet it runs fine on  
perl 5.8.8 + HTML::Mason 1.39.


package TestRequest;
use strict;
use warnings;

use base 'HTML::Mason::Request';
use HTML::Mason::CGIHandler;

sub foo {
         my $class = shift;
         $class->alter_superclass( 'HTML::Mason::Request::CGI' );
         $class->SUPER::new(@_);
}


package main;
my $bar = TestRequest->foo();