...
> 

> > It would also require some work with the grammar actions, actions that 
> > would they have their "reverse" counterpart, that could be undone if they 
> > maintain a symbol table for example. The same for Init and after blocks.

> Probably we'd limit it to building syntax trees.

I was wondering how you would proceed to maintain an IDE synthax highlighter 
and a symbol table, Sam Harwell has an external grammar, in this case would we 
need an external tree grammar that would parse the entire input at each 
modification (with a possible time lapse not to re-fire at each key-stroke)? Do 
you think of another way to do it (pointers?)? This is the only way I can think 
of it right now and it seems to me that it looses the interest of the 
incremental mechanism. Reverse actions could be called when branches are 
discarded/removed. Any comment on this?

> > Incremental lexer/parser is also very good publicity, it is very much 
> > visible to the "outside" world, Xtext gains a lot of traction thanks to 
> > that single feature, it helps adoption.

> Does it use the berkeley paper's algorithm for lexing?

I have no idea, I'll try to reach them and ask the question.

> > Here is a nice paper I found googling (in the even absolutely remote 
> > possibility it might be of interest):
> > http://www.google.be/search?q=General+Incremental+Lexical+Analysis+TIM+A.+WAGNER+and+SUSAN+L.+GRAHAM+University+of+California%2C+Berkeley&ie=utf-8&oe=utf-8&aq=t&rls=org.mozilla:en-US:official&client=firefox-a
> > ( http://harmonia.cs.berkeley.edu/papers/twagner-lexing.ps.gz )

> i find it extremely complicated.

it does not seem very obvious indeed, I'll try to have a look if my family life 
allows me some spare time.

Au plaisir,

Stanislas.



      
_______________________________________________
antlr-dev mailing list
[email protected]
http://www.antlr.org/mailman/listinfo/antlr-dev

Reply via email to