Mailing List Archive

http://bugzilla.spamassassin.org/show_bug.cgi?id=3518





------- Additional Comments From parkerm@pobox.com 2004-06-17 10:36 -------
Subject: Re: Type of bayes_token::token too large (in bayes_pg.sql)

What are the implications of using bytea vs char(5)? I ask because
the PostgreSQL docs make it seem like some special quoting has to
happen for bytea strings. It looks like they use the same amount of
space between the two. I guess I'm not grasping why using bytea is a
win in this case.

Michael





------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

http://bugzilla.spamassassin.org/show_bug.cgi?id=3518





------- Additional Comments From snyder@roguewave.com 2004-06-17 11:26 -------
I _think_ it's a win (if it works) because it seems to be twice as fast as using
char(5). My test file of spam (~400K, 71 messages) took about 1.75 minutes to
learn when token was char(5), but only 40 seconds to learn as bytea.

The tokens being inserted appear to me to be (in many cases) non-printable
values, which may be the cause of the slowdown for inserting them into the
postgreSQL database. (I really don't know enough to know for sure-- I'm just
guessing at this point.)

It doesn't appear to me (yet) to have required any additional changes, but I'm
still working through my 'make test' failures.




------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

http://bugzilla.spamassassin.org/show_bug.cgi?id=3518

parkerm@pobox.com changed:

What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |RESOLVED
Resolution| |FIXED



------- Additional Comments From parkerm@pobox.com 2004-06-17 13:23 -------
Committed revision 21389.


Did not use bytea, it does not work properly. char(5) seems to work well.



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.