Yaki

Yaki (which stands for “grill” in ) was the provisional name for my project, which was completed in May 2007.

It is the CMS that ran this site from to , and can be described as a full-blown, heavily filesystem-oriented wiki engine.

At the moment, there are three independent versions of it:

Since people like lists of features, here are a few (for the production version):

  • 100% pure , with extensive UTF-8 support1.
  • Entirely self-hosting, running atop a modified (1.51-rcarmo) application framework2.
  • Completely filesystem-based (pages are stored in a directory tree, not a database).
  • Heavily optimized processing:
    • Pages are pre-processed to .
    • HTML and other internal information are stored in a single-file cache, a la Haystack.
    • Everything is served via sendfile(2) calls whenever possible.
    • Uses every HTTP caching trick in the book to minimize actual page hits.
  • Completely markup-agnostic–all the internal processing relies on , and it ships with support for:
    • Raw HTML.
    • Textile.
    • Markdown.
  • Any markup engine that generates HTML can be added, and markup can be defined on a site-wide or page-by-page basis.
  • Has all the usual features, like:
  • Has a number of unusual Bliki features, like a blog-like home page, linkblog support, and the SeeAlso table at the bottom of each page.
  • Supports full-text indexing and search thanks to Whoosh.
  • The core code is open source under the MIT License.

Similar Projects

There are a number of similar endeavours out there you might find at least as interesting. You can check out for a vastly more comprehensive list I compiled during Yaki’s development, but these are fairly modern:


  1. As much as possible within OS and filesystem constraints. ↩︎

  2. Forget about massive sprawling toolkits; Snakelets is stable, tweakable and largely platform-agnostic. ↩︎

This page is referenced in: