[Mailman-Developers] thinking out loud...

[Fil]

> Or perhaps a better way would be to store messages static, 
> meta-data in a database, and build indexes on the fly, since the 
> messages won't change....

Why shouldn't the messages change? People have the right to send me an email
saying "please delete my name from the archives" - usually it's their
telephone number, or their employer's domain name ;)

So, suppose the master file is the mbox: you just need a script to scan it
and output whatever is needed. What's wrong with that approach? CPU? Then
keep the metadata in a cache (pickle) file, and check its mtime against the
mbox mtime. The webscript should issue the correct expiration dates (ie not
the mbox mtime). I don't think reading an mbox will take that long on modern
servers, especially if you know where to get the contents you need.

Then you have it: no database to maintain, minimal disk footprint, no big
script. Just one tiny script that echoes your mailbox content. Like IMAP: it
just works. If it becomes too heavy, add a cache. It could well prove less
heavy than to regenerate all indexes for each incoming mail. And much much
more handy...

I'd try a go at it, but that would have to be php or perl ;-(

-- Fil