- From: C. M. Sperberg-McQueen <cmsmcq@blackmesatech.com>
- Date: Mon, 25 Jul 2022 11:56:03 -0600
- To: Steven Pemberton <steven.pemberton@cwi.nl>
- Cc: M Joel Dubinko <micah@dubinko.info>, public-ixml@w3.org
Steven Pemberton <steven.pemberton@cwi.nl> writes:
> Until we introduced the prolog, I think I'm right in saying that ixml
> had a 1 character lookahead. It is left-to-right, but in order to make
> it top-down (i.e. LL1, which I think Pest is a subset of), you have to
> know which character starts each alternative. All the alternatives of
> term begin with a factor, so to make it LL1, you'd have to refactor
> the grammar (sorry about that) to something like
> term: factor, ("?"; "*"; "+"; "++", sep; **, sep).
> but then you'd get a different parse tree.
If memory serves, when I made an LL(1) version of the then ixml grammar
a while back, I also needed to refactor the character set rules to
handle some overlap in the prefixes of literals and ranges.
Michael
--
C. M. Sperberg-McQueen
Black Mesa Technologies LLC
http://blackmesatech.com
Received on Monday, 25 July 2022 17:59:29 UTC