Bug#236882: apache-perl: I can reproduce this segfault as well

2004-03-08 Thread Dave Rolsky
Package: apache-perl Version: 1.3.29.0.2-1 Severity: normal Followup-For: Bug #236882 I've tracked it down to this bit of code in HTML::Mason::ApacheHandler my $c = $r ? $r : Apache-server; @val = HAS_TABLE_API ? $c-dir_config-get($p) : $c-dir_config($p); It's the call to -dir_config

Bug#236882: apache-perl: The patch from bug 226131 is the culprit!

2004-03-08 Thread Dave Rolsky
Package: apache-perl Version: 1.3.29.0.2-1 Severity: normal Followup-For: Bug #236882 I rebuilt the apache packages from the Debian source without the patch provided in bug report 226131, and it starts up fine. Looking at the code being patched, I have absolutely no idea why it would cause a

Bug#236882: apache-perl: The patch from bug 226131 is the culprit!

2004-03-08 Thread Dave Rolsky
On Mon, 8 Mar 2004, Dave Rolsky wrote: patch may just be incorrect. It's testing if if RETVAL is SvOK, but _all_ SVs are SvOK, whether they are the Nullsv global, or a new SV created via newSV! I take it back. Nullsv should not be SvOK, and I'm not so sure about newSV, looking at the core

Bug#236882: apache-perl: New patch from mod_perl maintainer

2004-03-09 Thread Dave Rolsky
Package: apache-perl Version: 1.3.29.0.2-1 Severity: normal Followup-For: Bug #236882 I brought this up on the mod_perl list and it turns out the upstream patch was indeed a mistake. Here's the message from Geoffrey Young, one of the MP dev folks: Geoffrey Young wrote: -if (!SvTRUE(RETVAL))