Re: mailing list stuff in progress

Miles Egan <[email protected]> writes:

> All the message parsing stuff is pretty hacky.  I'm hoping to dump it
> all in favor of a standard library.  Maybe once yours is ready?

Last night I finished adding unit tests for the date parsing.  It works
pretty ok.  My realization last night was that my approach will not
scale very well.  I decided to use CLOS to represent the various fields
and parts of a message and the regex scanning plus the overhead of
creating CLOS instances (as opposed to keeping just lists of strings)
makes this approach not ideal for very-high volume use-cases.

That said, I'll probably end up providing an alist-type implementation
too where there are no CLOS instances and a message represation is
basically as simple as:

  ((:header
    (:subject "Hey!")
    (:from "Erik Enge <[email protected]"))
   (:body "How's life?"))

Which I bet would be much more efficient.

/That/ said, my regexes can be optimized a thousand times by someone
more proficient with them than me.

I need to add code for the Keywords, Return and Received headers and the
message class needs to be told about how many of the various fields it
can legally keep.  The reason this has taken me a bit of time is that I
decided to do it with regexes and I'm only half-way into Jeffrey
Friedl's book, being the regex-newbie that I am.

Erik.