I started working on such a transform about one month ago.
I can share a first implementation which solved succesfully the few
sudokus I tested:
http://eric.brechemier.name/project/sudoku/2005-02-14-sudoku-xslt.zip
The transform annotates sudoku squares with allowed and forbidden
values, using three "human-like" reasonings, which can help to
determine the level of the puzzle. The result of the transform is an
updated sudoku state.
It is still a little rough at the edges, so any feedback will be much welcome.
I am especially interested in any suggestion on how to allow the
transform to loop without getting inconsistent values in xsl:keys.
Eric
On 2/13/06, andrew welch <andrew(_dot_)j(_dot_)welch(_at_)gmail(_dot_)com>
wrote:
Does one exist?
I've been asked if it's possible to solve a Sudoku puzzle with XSLT,
which I replied "of course"...
I've started to write one myself - its gets pretty hard pretty
quickly, has anyone does this before?