Currently supported

• Optimized for the iPhone screen
• Multiple concurrent languages (with interlanguage links for languages installed)
• Full-text and exact search on article titles
• Case insensitive and diacritic-insensitive search (searching for internationalisation will match articles containing IñtërnâtïôÑàlizãtiøn in their title)
• Images Including thumbs, inline, and galleries! More info in ImageSupport
• UTF-8 support including CJK (Chinese, Japanese, Korean) - as long as the font is present on the iPhone (so far I've only seen Arabic missing among major languages)
• NEW: Categories (and subcategories)
• Great parser (reuses a former semi-official wikimedia parser) Most of the wiki grammar supported:
  • References
  • Lists
  • TOC
  • Tables
  • Cite templates
  • Infobox templates
  • Footnotes, cross-linked with their location in the article text
  • Math notations (more info on MathSupport)
  • IPA (international phonetic alphabet)
• Filters out irrelevant article namespaces (Wikipedia:, Help:, Special:... language-dependent namespaces are detected automatically) to not pollute the search results page
• For more convenient browsing, a [top] link at every chapter header to come back to the top of the page
• Search form always handy
• Auto-Hide Safari address bar at page load
• Very decent index size (180Mb for the English version)
• Redirects (up to 4 levels)
• Can run on your desktop or any device that support PHP and Perl
• Easy for anyone to contribute (Perl/PHP)
• Runs in Safari: Zoom, landscape mode, back button etc coming out of the box
• Images and math files are concatenated in fewer large files to facilitate transfer to iPod
• Fully automated indexer
  • Archive downloading, splitting, title listing and full-text indexing, image download, math image generation in one command line

Check out the ScreenShots to see how it all looks.

Known limitations

• Images (even links). Oddly, image urls cannot be built directly out of their reference in the article text
• Math
• Timeline
• Categories
• Most templates aren't supported
• It's not very fast. On very long articles the Wiki parsing takes up to 10 seconds
• No search suggestions yet
• There's no easy way to know if a link will exist or not, so I can't print them in red...
• The search engine doesn't support translating accented charaters to their latin form. i.e. searching for evidence will not match the French évidence. I'm working on that but it's not so simple...

Semi-planned improvements

• Still I could probably improve the performance by indexing not only the archive file number but also the offset of the article within the archive - although I think wikitext processing is far more heavy Done. Indeed the improvement is small compared to the wiki processing.
• I have to try if it works with other wikimedia projects (should be but they may have templates that wikipedia doesn't which are important for page rendering) Tried wikibooks. Working fine, but there are a lot of interwiki links to wikipedia, which I don't support.
• Support for redirects (trivial to do, the only thing is I'd like to notify the user he's been redirected)
• Im having a small issue with articles with & in their title... shouldn't be too hard to fix, but I need some sleep right now...