Re: DEMOLISH and bad call to new()

Ernesto,

Very odd, when I turned on all your tests, I got one error, which was  
the second test saying it didn't have anything inside $@. This was  
fixed by localizeing $@ inside the DEMOLISHALL routine.

Look at line 50 of Moose::Object here:

   http://code2.0beta.co.uk/moose/svn/Moose/trunk/lib/Moose/Object.pm

And you will see the change i made. I also cleaned up your test a  
little to make sure that each test was not polluting the previous  
one, and then added it to the test suite here:

   http://code2.0beta.co.uk/moose/svn/Moose/trunk/t/100_bugs/ 
011_DEMOLISH_eats_exceptions.t

This should take care of your issues.

(FYI - the Moose svn repo is public readable, so you can grab the  
latest version of that and Class::MOP if you like, they should be  
released in the not too distant future).

- Stevan


On Apr 5, 2008, at 4:03 PM, Ernesto wrote:
> Hi,
>
> encountering strange errors à la 'sometimes it works, sometimes not'
> from my objects I found, that this maybe a bug deep in Moose's bowels.
>
> If you define a class using DEMOLISH and a required typed attribute,
> then the very first call to new() without any parameters returns only
> 'undef' and does not die, as expected, with the error message
> 'Attribute ... is required'.
>
> Subsequent calls to new with undefined or empty parameters die with  
> the
> correct error.
>
> Moreover, if you create another object with no DEMOLISH defined first,
> than the subsequent call to new for the class with DEMOLISH will  
> work as
> expected.
>
> I boiled this down to a repeatable test - please have a look at
>
> http://dienstleistung-kultur.de/moose/02_demolish_required.t.txt
>
> Xanx!
>
> Yours,
>
>     Ernesto
>
>
>
>