On Thursday, January 27, 2011 20:44:57 Alexander Aparzev wrote: > using Apache/2.2.9 (itk MPM)
Just a guess, has anyone ever tested itk-MPM? However, perhaps it's only a matter of bad timing. pnotes2.t is one of mine, I think. It looks as follows: for my $i (1..12) { t_client_log_warn_is_expected(); t_start_error_log_watch; $u="$url?$i"; $ok=GET_BODY $u; select undef, undef, undef, 0.2; # give it time to write the logfile ok t_cmp scalar(grep { /pnotes are destroyed after cleanup passed/; } t_finish_error_log_watch), 1, $u; } You see the 200ms pause? Maybe that's too short for itk. Try the following: # clean the log >t/logs/error_log # run the test t/TEST -verbose t/modperl/pnotes2.t # now check the log and see if it contains 12 lines of 'pnotes are ...' ((12==$(grep 'pnotes are destroyed after cleanup passed' t/logs/error_log | wc -l))) && echo ok The reason for the test is a bug in a former mp2 version where the refcount of the pnotes hash might become wrong: $ svn log -r584380 t/response/TestModperl/pnotes2.pm ------------------------------------------------------------------------ r584380 | gozer | 2007-10-13 08:48:14 +0200 (Sat, 13 Oct 2007) | 8 lines Don't increase the refcnt of the pnotes HV* twice, we will leak it. Submitted-By: Torsten Foertsch <torsten.foert...@gmx.net> Message-Id: <200710111518.18560.torsten.foert...@gmx.net> Reviewed-By: gozer Torsten Förtsch -- Need professional modperl support? Hire me! (http://foertsch.name) Like fantasy? http://kabatinte.net