[Mediawiki-l] Debugging spam blacklist extension

[Rick DeNatale]

With the recent rash of wikispam attacks, I decided to install the
SpamBlacklist extension on my mediawiki.

It seems to be working, but I can't seem to tailor it.  I downloaded
the blacklist from meta to wikimedia_blacklist, and I set up the file
list with:

$wgSpamBlacklistFiles = array (
	"$IP/extensions/SpamBlacklist/wikimedia_blacklist", // Wikimedia's list
//	database	title
	"DB: wikidb Spam_blacklist",
);

I created an article in the main namespace called "Spam blacklist" and
put the following regex fragment in it:

p21\.info

I tried to resave an article from one of the spammers with a link to
this site and it wasn't caught.

I messed around and while trying to isolate the problem I put that
line into the beginning of the wikimedia_blacklist file, and it
worked!

So I took it out and tried again, and sure enough it failed to trap
it.  So question 1 is

What's wrong with my setup in pointing to the local Spam blacklist
article, should it be in a different namespace perhaps?

Then I put the line back into the file so that at least I'd be
blocking p21.info until I refreshed the master list, but it then again
failed to work. I tried again by picking another domain in the master
list, and it got blocked.

So question #2 is why changes to the master list aren't effective? 
Looking at the extension code, it seems to be talking about somehow
caching the reg exp list, but how that works is cryptic to me. Any
advice or insight?  Perhaps this is also why the database article
isn't working either.