No I don't think it would be that slow really... not compared to everything else that has to go on - so that sweeping could work, as it is line based... (I was planning on doing that in the near future, using a fast parser which has lexer rules only for that stage). So in summary, you can ignore DSLs for the moment.
On 5/5/06, Russ Egan <[EMAIL PROTECTED]> wrote:
Would it be too slow to do the dsl replacements before anything else? What I mean is, first sweep the whole drl, replacing all dsl statements...nevermind. That wouldn't work, since strings might appear in comments or java blocks that might look like a dsl token but isn't. Yeah, I like the "don't worry about it right now" idea. :) On Thu, 04 May 2006 20:02:07 -0400, Michael Neale <[EMAIL PROTECTED]> wrote: > yes, that was a challenge... basically had to "reparse" things when a DSL > expression is happened upon, and track line numbers etc... took a while. > What I am thinking is to have the DSL step as a nested parser using > simple > rules - this means it can very quickly convert the rules into the native > one > (this has the added debug advantage of being able to switch between DSL > view > and normal view possibly), whereby the usual parser does its thing - in > short, I would say don't worry about the DSLs to start with, they can be > added on later. > > On 5/4/06, Russ Egan <[EMAIL PROTECTED]> wrote: >> >> I just discovered that there's a groovy plugin for idea that uses antlr >> as >> the scanner and parser, so I'm hoping I can use that as a model for drl >> files. Not sure how I'll do the dsl though *shiver*. >> >> On Wed, 03 May 2006 21:07:43 -0400, Michael Neale >> <[EMAIL PROTECTED]> wrote: >> >> > yes eclipse works by having partitioners, and scanners... its not >> really >> > parsing, but its meant to be fast and fault tolerant, so that the >> editor >> -- Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
