Performance issues....and ugly news.

[David Relson]

Nick,

It seems like your efforts to improve performance are smacking up against a 
non-ending series of flex obstacles.  It reminds me of the guy trying to 
pound the square peg into the round hole.  Maybe it just doesn't fit.

We _do_ have a performance solution - my check_alphanum() 
routine.  Depending on point of view it can be considered a quick and dirty 
hack or it can be considered an engineering solution, i.e. one that gets 
the job done, even if it's neither pretty nor elegant.

Last night I started building an all-in-one parser.  Since I've new to 
writing parsers, I'm taking baby steps at first.  At the moment, the goal 
is proof-of-concept that one parser with states, a.k.a. start conditions, 
can handle headers, plain text, html, encodings, and mime parts and that it 
can handle them without excessive underlying b.s.  Perhaps it'll work, 
perhaps not.  We shall see.

Back to the subject of performance.  Why not go with check_alphanum(), at 
least temporarily.  Nobody has evidenced a reason why it's _not_ 
usable.  Perhaps we should stop beating our heads against the wall and just 
use it.

David