One of the uglier issues in threading messages is the need for date
normalization. There is a small, but dependable percentage of messages
that have obviously bogus date fields -- years in the past, years in the
future. And there are a number of messages that use non-standard time zone
indicators and GMT offsets. I can share some pseudocode that I've written,
but it would be nice to come up with an agreed-upon way to identify bogus
dates and normalize them as best as possible. Generally speaking, there's
a good date somewhere in the headers, even if it's the date stamp from your
SMTP daemon when the message arrived.
Reading threads in proper chronological order is a great boon, so I think
this is worth the effort to make it consistent. I've even considered
grabbing quoted text and searching for it in previous messages, when all
else fails.
Nick
--
Phone/fax: (408) 733-7613 E-mail: narnett_at_mccmedia.com
Received on Mon 27 Apr 1998 04:48:29 PM GMT