Re: How to require a "Re:" for subject-based threads?

1999-03-09 18:13:15
On March 9, 1999 at 13:34, Jamey Wood wrote:

In the interim, is there a resource to tell mhonarc to compute ordering of 
threads on the index page based on the date of the most recent member of the 
thread (not just the beginning of the thread)?  So in my case, even if a 
message was placed into some ancient thread it would at least be noticed, 
since that thread would jump to the head of the index page...

There is no resource to do this.  Supporting this would cause reworking
on how threads are computed and could have significant performance
impacts (plus the end-user usuability impact issues [ie. thread listing
locations moving around frequently]).  I can elaborate more about the
issues if there is interest.

IMHO, this is a case where the date index is useful.  The date index
can be used by readers to see what are the latest messages.  From
there, they can always jump to the thread index from the message
pages (using TSLICE in message pages may add extra benefit).

Also, my (probably obvious) comment on the changes you mentioned to require a
"Re:" to be included in a thread...  You'll start running into cases where an
incoming message's subject could place it in one of multiple threads, and it 
should probably go into the most recent thread...

Or maybe not ...  After some more thought, the "Re:" requirement is
looking less trivial to implement.  I am thinking that computing time
deltas between messages would have to be done to deal with the
problem having more than one possible parent message in subject-based
detection.  BTW, just having to handle multiple possible parents adds
a little twist during the initial pass of finding first occurances of

Although maybe one day you
could implement something advanced like comparing body text quotes of the new
message with the contents of the messages in each thread to see if it might 
really belong to an older thread...  But I know that's starting to ask a lot.

Yep.  Here we start getting into more complicated heuristics and major
performance issues.


             Earl Hood              | University of California: Irvine
      ehood(_at_)medusa(_dot_)acs(_dot_)uci(_dot_)edu      |      Electronic 
Loiterer | Dabbler of SGML/WWW/Perl/MIME

<Prev in Thread] Current Thread [Next in Thread>