On Wed, 10 May 2006, Larry Rosenman wrote:
> Josh Berkus wrote:
>> Marc,
>>
>>> Why? Personally, I think it falls under 'the right tool for the
>>> job' ... there are two things I use MySQL for: phpAdsNew and
>>> Spamassassin/Bayes DB ... neither of which I care if there is data
>>> lost for, and for which using Berkeley DB would work just as well,
>>> if I cared to learn its API ...
>>
>> Well, problem #1 is that you never discussed this on pgsql-www that I
>> can find in the archives. It's quite possible that the other WWW
>> team members would not share your evaluation, and would have decided
>> to take another course. This looks like a case of you making a
>> unilateral decision which other members of the team could (will?) be
>> required to do damage control on.
>>
>> BTW, the reason Spamassassin requires MySQL is that it's badly
>> designed -- on my server we couldn't even use MySQL, we were running
>> it in RAM because of the I/O intensive design. If you switched to
>> dspam, you could use Postgres properly.
>
> I also believe that SpamAssassin 3.1.1 has a PostgreSQL friendly bayes
> storage algorithm.
>
> I'm likely to try it on mine here shortly.
Please let me know how it turns out ... we're running MAIA Mailguard here,
so switching to dspam isn't an option, but would be interested in moving
off of MySQL for this is the Bayes stuff actually was improved in 3.1.1
... the one thing I hate with MySQL is that I can't use a simple 'ps' to
see what the connections are doing ;(
----
Marc G. Fournier Hub.Org Networking Services (http://www.hub.org)
Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
