Dave, > > [Bug 6624] BayesStore/MySQL.pm fails to update tokens due to > > MySQL server bug (wrong count of rows affected) > > https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6624 > > > > [Bug 6625] Bayes SQL schema treats bayes_token.token as char > > instead of binary, fails chset checks > > https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6625 > > > > [Bug 6626] Newer MySQL chokes on TYPE=MyISAM syntax > > https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6626 > > Shouldn't one of these patches include the change to InnoDB in the > schema itself instead of just a recommendation in the documentation?
It does, the Bug 6626 (attached patch and the committed change to trunk) changes engine in the schema, then adjusts sql/README.bayes accordingly. > Combined with the other patch, maybe Benny's should be the new preferred > schema? There are several small differences there, some good, some not so good, some irrelevant, some unnecessarily overspecified. Possibly a matter for a separate topic. > > Dave, could you please revert my previous patch and apply instead > > the patch attached to Bug 6624 (applicable to 3.3, or just install > > trunk). > > > > Also consider the tiny fix to the schema as suggested by Benny Pedersen > > and found as an attachment to Bug 6625. This should avoid a > > problem if your SQL server defaults to UTF-8 or other character set > > with stricter checking than a plain Latin-1. > > Okay, both patches applied, and confirmed that it fixes the problem. Great, thanks for testing. The [Bug 6624] is worth remembering, the subtle mysql server bug is likely already affecting many MySQL users, especially because it does not complain loudly when using an existing database, just learns poorly, yielding poor bayes results. Thank you for pointing out the problem! Mark
