check_whitelist does not compact
Hello, Although "check_whitelist --clean" removes entries from the auto whitelist database, it does not compact the file, at least on my system (latest SpamAssassin on debian unstable, perl 5.8.8, DB_File 1.814). A minor modification of the check script that ties a new Berkley db and copies the cleaned structure over to it brought my 20MB whitelist to 600KB (after --min 3). I'd just send a patch, but I noticed there are also '.pag' and '.dir' files which I didn't know the purpose of. Are they safe to leave as is with the compacted db? Should they also be treated somehow? -- Gaal Yahas <[EMAIL PROTECTED]> http://gaal.livejournal.com/
Spellcheck plugin?
I'm wondering, would it be useful to have a plugin that penalizes messages with many spelling mistakes? This might help against all those creative ways of spelling out what the spammer wants to sell. To have any hope of working for people who expect mail in more than one language, a language recognition tool would be needed, but fortunately there are some of those around.[1] This adds quite a bit of complexity. Has anyone given it any thought? [1] E.g. <http://neugierig.org/software/langid/>. -- Gaal Yahas <[EMAIL PROTECTED]> http://gaal.livejournal.com/
Concurrent spamc via procmail
I'm using SpamAssassin 3.0.0 on Debian unstable. The docs recommend invoking spamassassin via procmail like this: :0fw: spamassassin.lock * < 256000 | spamassassin And to simply replace 'spamassassin' with 'spamc' when using the spamd setup. I just noticed that spamd preforks, so my question is whether it still makes sense to put in the lock, and if not, what limits the number of spamc instances that start up when I receive many mails in a short time. And a quickie (sorry for not asking this in a separate thread): is there a one-liner I can use to show what the Bayes score for a particular token is? I'm just curious :) Thanks a lot, Gaal -- Gaal Yahas <[EMAIL PROTECTED]> http://gaal.livejournal.com/
