Karsten Bräckelmann wrote: > Mark, how positive are you, that the Perl bug you filed is the real, > underlying issue? Seeing all these reports about effected systems, and > seemingly identical systems not effected...
I'm certain this bug is the underlying issue. I picked debugging from where Matt drove it, and debugged it down to the perl level. I haven't noticed the problem because we are running perl 5.13.10 on our mailer, which is unaffacted. The perl bug report turns out to be a duplicate of #8262, #27344 and #5475. Other unaffected systems apparently did not have any tflags=multiple rules compiled through re2c which would require a re-check on a lossy pattern match. Or maybe they did not have compiled rules enabled at all (loadplugin Mail::SpamAssassin::Plugin::Rule2XSBody). Mark
