On Thu, Feb 10, 2005 at 06:32:30PM -0500, Stas Bekman wrote:
That most likely means that someone redefines sub lib::import. Could be Time::Piece or something else used by that module. It's their fault, not A-T's one. You could easily verify that by adding something like:
I love hanging out with you guys; always something new to learn. Thanks for the hints. Unfortunately the $SIG{__DIE__} trick didn't work. Perhaps that's due to the fact that this bug lies within the UNIVERSAL module. Basically, Time::Piece has the following line which is causing the problems:
use UNIVERSAL qw(isa);
From what I read about UNIVERSAL in its source, it causes the importmethod to be redefined. I was able to create a script that reproduces the error. The error happens under Linux as well. This begs the question as to why A::T works without error on Linux but not on Windows.
Further testing has shown that order is important. If I place a 'use lib' line *after* the use UNIVERSAL (or any other module which uses UNIVERSAL), all is well. However, I still don't get the reason why these tests would fail under Windows but work under Linux. There must be something different about my Perl environment between the two.
Now, since we know that using the UNIVERSAL module is not uncommon, should A::T take measures to defend against lib->import getting redefined? Adding a 'use lib;' line just before the call to lib->import($top_dir) works for me under WinXP and does not generate redefine warnings.
Since it's not A-T's fault that someone has redefined the lib::import, I don't think it's for us to fix it. At least for now I believe it's the best that you apply the workaround in your code. i.e. after you load the module that redefines lib::import force the reload of lib.pm which should restore it back. alternatively you could remember the original sub and restore it w/o reloading lib.pm:
my $orig_sub = *lib::import{CODE};
require ModuleBreakinglibpm;
*lib::import = $orig_sub;notice that it uses 'require' and not 'use'.
if later on we start getting more similar reports we will consider adding a workaround in A-T.
Sounds OK?
-- __________________________________________________________________ Stas Bekman JAm_pH ------> Just Another mod_perl Hacker http://stason.org/ mod_perl Guide ---> http://perl.apache.org mailto:[EMAIL PROTECTED] http://use.perl.org http://apacheweek.com http://modperlbook.org http://apache.org http://ticketmaster.com
