On 11/9/07, David Gerard dgerard@gmail.com wrote:
Before this, step 5.5) A bunch of bots run over all the wikis fixing corner-case wikitext that's fixable - changing stuff that works only in the old parser to stuff that works in old and new.
For some reason, I was expecting "7) hop on the back of a passing pig and fly away". But maybe you were actually being serious.
I thought my suggestion of this slow migration to a new parser was a bit fanciful. But is it a good idea? What do people think? Given that:
a) The parser is extremely complex, and difficult to make significant improvements to b) Writing a new parser that can implement *all* of the current parser's idiosyncracies while actually being an improvement, is a massive challenge c) Breaking everyone's wikitext overnight is not going to endear anybody to anybody.
is there any other way to end up with a new parser based on fully specifiable formal grammar?
I'd really like to hear from Brion on this.
Steve