[tapx-dev] Broken coverage testing

[Andy Armstrong]

On 9 Aug 2007, at 13:18, Leif Eriksen wrote:
> I'll try a D::C around 0.50 or so but I feel that D::C isn't
> the problem, I feel (with zero evidence) that T::P is doing
> 'something' - I use D::C _all_ the time and nothing else is
> having this issue. I apologise if that gets people's hackles
> up, I am new here after all, but its just a feeling.

Hackles fully retracted :)

When I chatted to Paul Johnson about it he seemed pretty convinced  
that it's a D::C bug. In a sense it must be a D::C bug - we're  
certainly not doing anything purposely to upset D::C - and D::C's job  
is to handle whatever we throw at it - within reason.

For clarification have you ever been able to get coverage for the  
whole test suite? I didn't attempt it until a couple of weeks ago and  
it was broken then.

If we can't establish that it has ever worked then the search for a  
magic combination of versions that play nicely together is a wild  
goose chase.

> I'm going to try to make a few mod's to D::C, I plan
> to tie() the Devel::Cover::DB::File blessed hash to a
> class to trace changes to that hash, to see if it does
> get populated like we expect and gets crushed later by
> something in T::P, or if the problem is completely in D::C.
> I'm hoping to have that done over the weekend...the wife's
> away, so I've got to do _something_ for the weekend! Its that
> or fix the car...

:)

Paul waved in the general direction of Devel::Cover::DB->merge as the  
likely culprit.

-- 
Andy Armstrong, hexten.net