Those are definitely valid concerns. Has anyone made a wiki-like site with Yesod? I hadn't heard of Yesod until I joined this mailing list, but I've seen quite a bit of buzz around it since then. If a large enough chunk of the community is backing a framework and focusing on making it secure and reliable, then it should be possible to build applications with it (wikis, blogs, etc.) that draw on the framework's strength and security. You may still have security issues, but if they're continually addressed and maintained at the framework level it benefits everyone building applications on top of that framework. I'm still relatively new to the Haskell community so I apologize if much of this has been addressed before!


On Fri, Jun 3, 2011 at 3:11 PM, Gwern Branwen <gwern0@gmail.com> wrote:
On Fri, Jun 3, 2011 at 4:17 PM, Eric Rasmussen <ericrasmussen@gmail.com> wrote:
> This is a bit of a tangent, but has anyone developed wiki software in
> Haskell?

Gitit is the most developed one, and it's been suggested in the past
that hawiki move over. It's not a good idea for a couple reasons,
which I've said before but I'll repeat here:

1. Performance; there have been major issues with the Darcs backend,
though mostly resolved, and we don't know how well the Git backend
would scale either. Gitit has mostly been used with single-users (how
I use it) or projects with light traffic (wiki.darcs.net). I don't
know why hawiki is slow, but whatever it is is probably either
hardware or configuration related - MediaWiki after all powers one of
the most popular websites in the world.
2. Security; there have been big holes in Gitit. Some of it is simple
immaturity, some of it due to the DVCS backends. Where there is one
hole, there are probably more - if there aren't holes in the Gitit
code proper, there probably are some in Happstack. There's no reason
to think there aren't: security is extremely hard. And in that
respect, Mediawiki is simply much more battle-tested. (Most popular
websites in the world, again, and one that particularly invites abuse
and attack.)
3. The existing hawiki content is Mediawiki centric, relying on
templates and MW syntax etc. Templates alone would have to be
implemented somehow, and Pandoc's MW parser is, last I heard, pretty
limited.

Gitit is great for what it is, and I like using it - but it's not
something I would rely on for anything vital, and especially not for
something which might be attacked. (This isn't paranoia; I deal with
spammers every day on hawiki, and c.h.o was rooted recently enough
that the memory should still be fresh in our collective minds.)

--
gwern
http://www.gwern.net