On 11/11/07, Simetrical Simetrical+wikilist@gmail.com wrote:
Not a good premise, since we don't *have* a well-defined grammar.
True.
The existing, stalled, grammar project seemed to be predicated on the notion of writing a grammar that exactly matched the behaviour of the current parser.That turned out to be hard.
Why don't we focus our attention on writing the grammar for the new parser, in a way that is X% compatible with existing wikitext? We can then build both the grammar and the parser simultaneously, after a certain point.
One attractive feature of this method is that we can leave out the "really hard" bits of the grammar at first, then eventually decide whether we want them in the X% at all.
Steve