On Wed, 2003-07-30 at 15:13, Chris Santerre wrote: > 3) This seems to be the topic recently. People would like to get a little > better then simple rule writing. I haven't looked into writing eval tests at > all. I think there is a good reason why there is little to no info on this. > Devs don't need people messing with SA code ;) > I'm already in there and messing :)) Expect an automatic listwashing token (rot13 etc) eval from me soon as long as nobody makes me do any real work.
> That said, is there a way to get local eval tests? Is there a good source of > explanation (eval test for complete bumbling iditots.) on how to write > these? hmm, eval tests for idiots, kinda sounds like loaded shotguns for idiots, should reduce the idiot count but is that really what is desired? :) I managed to figure how to write my own eval test by reading the source code. Anyone who isn't at a level where they can do that really shouldn't be trying to, or they're going to shoot themselves in the foot. I'm not trying to dissuade anyone from trying to learn, but there's a lot more involved in writing an eval than there is in writing a rule, a lot more ways to get it wrong. > This is pretty much a question for the devs. > > Before anyone gets crazy with the eval tests idea, I have seen many rules > (meta) that will do exactly what you want. However in David's case, I can't > think of a series of rules to do this. > There are some things you can only do inside spamassassin with an eval test. However, there is another way. Write a test as a recipe for procmail or even call an external test prog from procmail (do this before spamassassin is called), add an X-MyTestName header if the test scores positive. Use a test in spamassassin to match up with the header you added earlier and score as appropriate. Don't forget to bayes_ignore_header the header you add. Removing the header after the mail has been through spamassassin is the final touch. (note formail -I won't remove it if you're using report_safe 1 as it ends up in the body of anything marked as spam, use sed -e). I've used this technique to add a rule which catches my own usenet+date@ email addresses and adds points for older addresses. I also have the same technique running on a customer's server with an early version of my listwashing token identification code. > -Chris > -- Yorkshire Dave -- Scanned by MailScanner at wot.no-ip.com ------------------------------------------------------- This SF.Net email sponsored by: Free pre-built ASP.NET sites including Data Reports, E-commerce, Portals, and Forums are available now. Download today and enter to win an XBOX or Visual Studio .NET. http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
