I missed part of this thread, but I have my doubts about perl's internal crypt working correctly in a threaded environment. Using 5.8.0 threaded for the x86 platform / mod_perl 1.27 that on an uniprocessor machine it would produce correct results most of the time, but on a multiprocessor machine it would consistantly produce bad results. After battling with it for awhile and not having time to get correct data to produce a bug report - I switched to the pure perl version Crypt::UnixCrypt which works quite well across the board. *sigh*
On Thu, 10 Oct 2002, Ask Bjoern Hansen wrote: > On Fri, 6 Sep 2002, [iso-8859-2] Tomáš Procházka wrote: > > > Problem: Sometimes, although user entered correct password, is > > authentication rejected. I tried logging values of $real_pass and > > $test_pass and they differed. When I add line > > Did anyone figure this out? > > The following content handler gives a different output everytime I > restart my threaded mod_perl 2.0. Dm8yjkphWW352 is the correct > answer; I get that after about a third of the restarts. > > - ask > > > package Develooper::MT::Test; > use strict; > use warnings; > > use Apache::RequestRec (); > use Apache::RequestIO (); > > use Apache::Const -compile => 'OK'; > > my $clear = "foo"; > my $salt = "Dm"; > > sub handler { > my $r = shift; > $r->content_type('text/plain'); > my $crypted = crypt $clear, $salt; > $r->print("crypt $clear, $salt = $crypted"); > return Apache::OK; > } > > 1; > > -- //========================================================\\ || D. Hageman <[EMAIL PROTECTED]> || \\========================================================//