Re: Glib-1.071/t/c.t fails under Test::More 0.53

[Rafael Garcia-Suarez <rgarciasuarez gmail com>]

On Thu, 16 Dec 2004 17:34:11 +0100, Torsten Schoenfeld
<kaffeetisch gmx de> wrote:
> On Mon, 2004-12-13 at 06:57 +0300, Alexey Tourbin wrote:
>
> > Details: eq_array() calls _deep_check() on each pair of elements.
> > _deep_check used to test ($e1 eq $e2) in the first place and return
> > true if so, until recently.  Now _deep_check drops stringification
> > overload in the first place (with Test::Builder::_unoverload()) and
> > tests that both elements are references or both not.
>
> Hrm, that sucks.  That pretty much means they break public API, because
> I think our assumption that eq will be used is correct, the function's
> called eq_array after all.  Do you know if they have a good reason for
> that change and if they would consider reverting it?

If I understand correctly, eq_array and other is_deeply functions are
supposed to check for identity, and to work even in front of
overloading bugs :) so I think this change is actually a feature. 
This kind of discussion actually belongs to perl-qa. You may also
consider a replacement implementation of eq_array ; see
http://qa.perl.org/test-modules.html for ideas.