[Nmh-workers] Thoughts on the UTF-8'ing of nmh, external database searches

[Jon Steinhart]

Years ago I added the hooks interface to nmh so that I could sync an external
database to my nmh messages for fast searching.  It was an ugly system that
used Berkeley DB.  I just redid it by writing a couple of hundred lines of
scripts and using ElasticSearch as the back end.  Would be happy to share it
if anybody cares.

My original system predates iconv, and is also somewhat out of date as far as
the new RFCs on character sets.  Mostly still works as I generally don't care
about mail in languages that I can't read.  My system works with a set of
external helper programs that convert mime parts to text for indexing; for
example using pdftotext to extract text from pdf attachments, html2test for
html attachments, etc.

It would be really really cool if the proposed UTF-8 version of nmh was
structured in a manner that would allow me to more easily do this stuff.

It would be nice to be able to have a way to say "parse a message and give
me the results as UTF-8 including running mime attachments through specific
filter programs".

So, don't wanna create extra work for those doing the work on this, but
please keep this in mind 'cause it would be cool.

Thanks,
        Jon