2009/7/20 Karsten Bräckelmann <[email protected]>: > I'm slightly confused, but that may be just me... > > First, I noticed that *prior* to my sa-update, the open-whois rules > already have been removed from my stock update rule-set, and that the > last update was 10 days old. Alas, didn't retain a copy. > > Then there is the revision, as mentioned in bug 6157, comment 2. > http://svn.apache.org/viewvc?view=rev&revision=795855
oh great! no wonder I got an svn "G" indicator when I svn up'd that branch ;) > That's the latest updates branch for 3.2, and it wondered about the > note. It claims to have been copied from the old and deprecated > spamassassin/rules/branches/3.2 -- is this just a bad SVN note, or is > this still the way to go for the stable branch? ugh. it's complicated. There were no less than 4 places where we had rules: rulesrc: sandboxes trunk rules dir: distro /rules, as packaged 3.2.x rules dir: 3.2.x branch /rules, as packaged 3.2.x rule updates dir: confusingly ANOTHER 3.2.x dir, for producing 3.2.x updates from we've just gotten rid of the rulesrc sandboxes place. the other 3 are still intact, including the 3.2.x updates dir. When we do 3.3.0 we should _definitely not_ use that approach, it hasn't helped at all, and is quite complex. > Looking at the sa-update tags, I guess I see the reason for the initial > issue. Already back on Mar 25, the WHOIS_ rules have been removed, > though probably never pushed as an update (bug 6090), and published > later with another sa-update. > > Would that be correct? What exactly did you change for today's sa-update > then? And can bug 6090 be closed, too? I'm not sure about the bug 6090 changes, since mine at least included score changes too: http://svn.apache.org/viewvc/spamassassin/branches/3.2/rules/50_scores.cf?r1=795850&r2=795849&pathrev=795850 I suspect bug 6090's diff was a partial remove of those rules. -- --j.
