Kill this thread, it shouldn't have been one ( was: Re: CYRUS = ....)

Rob Siemborski rjs3 at andrew.cmu.edu
Tue Feb 3 12:02:15 EST 2004


On Tue, 3 Feb 2004, Brian wrote:

> Incidentally, even a cursory search through the archives would show that
> this has been debated to death and the developers have stated time and
> again that the benefits to a SQL backend haven't been worthwhile enough
> for them to want to implement it.

Note that "backend Cyrus in MySQL" is not at all the same thing as "have a
cyrusdb backend that supports mysql"

The former just isn't really feasable.  However the later is feasable, it
just isn't efficient for mailboxes and seen state.  It might be reasonable
for deliver.db (with the exception that expiration is likely to be slow),
and if the database was local it might be useful for tls sessions (same
problem with expiration).

As it turns out, with the CVS additions that allow quota to be backended
by cyrusdb, there is actually a situation where backending the quota
database in mysql is a win -- that is, being able to have a quota root
that is split between murder backends (still only likely to be useful for
shared folders).

-Rob

-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Rob Siemborski * Andrew Systems Group * Cyert Hall 207 * 412-268-7456
Research Systems Programmer * /usr/contributed Gatekeeper





More information about the Info-cyrus mailing list