Wikipedia in iSilo (or Plucker) format can't be based on the same code that is available now. A lot of the (broken) Perl scripts that are out there that try to convert it to something useful, don't, and are not really scalable.
The actual
MediaWiki software that drives the wiki has to be updated.. and I've already done this in my copy. This includes quite a few changes to make the output of the actual wiki more "compatible" with handhelds.
I've been working with the mediawiki developers to work this out. I have an approach that might work, but it still requires quite a bit of work to the database, content, and schema. The wiki tags used in WikiPedia are not exactly "valid" in most cases. A set of rules has to be developed to handle this conversion.
Importing the SQL dump of the wikipedia + images into a MySQL database and then parsing that with Perl to convert it, is the wrong approach. Spidering the actual visible, modified mediawiki code on a per-language basis, is the right way to do this.
It'll happen, but a lot of other pieces have to be fixed up first. See
this message I wrote discussing my progress in this, and other areas.