Show quoted text ajt> Andreas,
ajt> On Mon, 07 Jul 2008 at 08:08:40AM +0200, Andreas J. Koenig wrote:
Show quoted text >> >>>>> On Sun, 6 Jul 2008 13:58:40 +0100, Dr Adam Trickett
<adam.trickett@iredale.net> said:
Show quoted text >>
>> > Could you possibly run the tests manually (Build and Make) to see
if there is
Show quoted text >> > any difference with your automated tests.
>>
>> Interesting, I sent you three fails but I also sent you 5 passes. So
>> it is nothing special for my testing box.
ajt> Most peculiar.
Show quoted text >> I just tried the fail 1831687 manually and could reproduce it.
Show quoted text ajt> That's good, as long as it's reporiducable there is a chance it
ajt> can be fixed. If I could reporduce it on one of my systems I'd be
ajt> a lot happier, as I'll be able to fix it without disturbing you.
Show quoted text >> I've now run out of time to further investigate, will continue in the
>> (European) evening. Let me know if you have more concrete suggestions
>> how to find out more.
Show quoted text ajt> Thanks, I'll have a think and send you some ideas...
Found it: Sub-Uplevel-0.19_03 was the cause. Downgrading to 0.1901 let
your tests pass.
I found the same issue with the same three perls for Archive::Any
which is also a user of Test::Warn. And then I looked at the
dependencies of Test::Warn and Sub::Uplevel jumped into my eye because
the error message contained the word "level".
I wholeheartedly recommend that module writers add as many
dependencies as they can get hold of to their listed dependencies.
Including indirect dependencies. That way the test reports will
contain better statistical data that can lead us to the culprit.
In your case only instinct could help us.
@David: Adam is the author of XML::RSS::Tools which use Test::Warn
which FAILed on some of my perls due to Sub::Uplevel 0.19_03.
I've replaced Sub::Uplevel in all my perls now.
--
It's Not Magic, It's Work!
