[Spambayes] It gets funnier all the time....

[Tim Peters]

[Skip]
> The performance penalty should be reasonable.  One split() and length
> check for each part in the common case.  The try-to-fix function is only
> called if it's long enough, which should be rare.

I'd hate to see the code bloat with gimmicks that don't prove themselves via
testing (as TESTING.txt warned from the start, such stuff at best gets in
the way later of testing things that might truly help a little, by adding a
quasi-random component to testing results).  Has this form of spam been an
actual problem for anyone here?  Note that in Rob's original post, the
classification line was

X-Spambayes-Classification: unsure; 0.98

Adding code that moves this example's score from 0.98 to, say, 0.995,
doesn't appear worth any effort.