Re: [Phpwiki-talk] Wiki spam and the future of phpwiki

[Philip J. H. <ph...@po...>]

On Wed, 30 Mar 2005 08:56:02 -0600, "Dan Frankowski"
<dfr...@cs...> said:
> Finally, I believe this goal would be aided if there were much less
> code to maintain. For example, support only one backend. I see so much
> work going into different database ports, needlessly. I vote for
> MySQL, since that is what I am using. However, I might even go with a
> CVS backend (*shudder*) if that were the community's will, as long as
> it supported query-able relationships between 2 things (exactly like
> backlinks, or we've expanded also to ratings, categories, and in the
> future perhaps list items), and multiple instances from the same code
> base (currently with multiple MySQL DBs).

I heartily second that.  I suspect deciding one one backend might be
politcally extremely difficult.  It sure seems like something that needs
to be done.  One thing I really like about PmWiki is it avoids the whole
database question entirely by just storing pages as textfiles.  Have you
ever tried to configure mysql on a hosted system?  It can be extremely
annoying (or extremely easy, depending on the provider).

But there's no doubt that lots of the bug fixes have been for the
various database engines.  Perhaps the real core of the problem with
phpwiki is _too much_ choice.  All these choices result in a
combinatorial explosion of bugs.  For example, I use the Crao theme on
my website.  It turns out that there are a number of phpwiki
functionality bugs (like you can't delete pages) that happen only with
that theme.  Ugh.

> >I realize that the spam problem is not unique to phpwiki.  However,
> >I'm not seeing anything being done to address it.  For example, ip
> >address blocking is not implemented in the phpwiki version I am
> >running. Obviously that's not a particularly effective way to deal
> >with spam, but it is a start.  What about blacklists such as the ones
> >from chongqed.org or Movable Type?  Captchas?  Typekey?  I'm guessing
> >that if I'm having these problems, lots of other people are too.
> >What is being done in phpwiki to deal with the spam problem?
>
> To be fair, Reini put in two things, and I believe they are in RC1
> (aiming at 1.3.11 if all goes well):
>
> 1. Don't allow saving a page that has more than 20 external ("
>    http://") links. In our code, I modified "20" to be a configurable
>    parameter SPAM_MAX_EXTERNAL_LINKS. We've been completely spammed as
>    well, and I believe this will help us a lot. We have a wiki where
>    each legitimate page only contains a few external links, but spam
>    pages contain tons (>50 for sure) external links.

Obviously the spammers will adapt to that by creating lots of pages,
each with only a couple of links.  Then we will need a mechanism to
rate-limit the creation of pages, etc.  The never-ending spam battle.

Actually, I'd like to see rate-limiting for page creation now.  Maybe no
more than one page every 30 seconds, by default?

> 2. Hook up with babyspam, which is a hookup to Spam Assassin. In other
>    words, an evolving tool for detecting spam.
>
> Might not be a bad idea to also have IP filtering, but this can also
> be done at the .htaccess file in your dir .. IF your Apache allows
> overrides.

Those sound like very positive steps.  At the same time, they highlight
the brokenness of the phpwiki development model.  I had no idea these
features were being integrated.  If we had some more regular releases,
then these things would show up in the release notes (hopefully) instead
of being buried in cvs commit messages, etc.

Right now I kind of feel like each upgrade to phpwiki is like Christmas.
 Lots of exciting presents, but you never have any idea what you are
getting.

Thanks,
P.
--  
Philip J. Hollenback
ph...@po...
www.hollenback.net