The new version includes performance improvements for common cases (i.e. text without special compositing), and XML documentation for almost everything.
A new version numbered 126.96.36.199 from January 4th 2009 is out.
This version includes a bug fix relating text starting with numbers, followed by RTL text (which I consider a serious bug).
A new version was just released with a major bug fix regarding BiDi control characters (LRO, RLO, LRE, RLE and PDF) not doing their job.
Also in this release - Arabic shaping support (Thanks to Yasser Markam for his help) and full support for .net 1.1 out of the box.
Well, due to a new request the algorithm is now finished, as I implemented the part of if regarding splitting a string into paragraphs.
Also I changed the project state from alpha to beta, until I get some real results from people who use it with approval or with bug reports.
Have fun using it, and please, report bugs and performance issues.
This is the first news post, and for a good reason: Today was the first time a unit test actually passed, and correctly reformatted a string containing both English and Hebrew text.
The road is still long, though, as there are two missing rules implemented (L3 and L4, responsible for combining marks and character mirroring), and many more tests have to be written.
In addition, no real attempts to make things efficient had been made, so after all tests are written and pass, performance is the next step.
However, you are welcome to try and use it, write tests, etc.
I'll be glad to hear from you if you had tried it and found it somewhat helpful, even in its current state.
I won't make any releases until more tests are made, but for now you can get the code from SVN.... read more