Neil Hodgson wrote:
Eric Promislow:
I assume the technical details would be of interest here -- just ping
me and I'll write something up. The blog entry is geared towards
building a lexer.
That would be interesting as I expect other people are interested
in extending Scintilla's lexing capabilities. In the post you say that
"regular expressions alone are too weak" which was my belief when
starting Scintilla so lexers are code which is more tedious but more
likely to be capable of being complete. Transitions are important in
these composite languages and there's a good chance the current
SinkWorld code is too simplistic.
Definitely full C++ lexers are more complete than any VHLL-approach[1]
can give you. But on the other hand, if you can get 90-95% of your needed
functionality with a few minutes of work, that might be a worthwhile
trade-off.
The post is a little scarifying to a potential language definer:
they'd need to be quite motivated to start work on one. A lexically
simple language (possibly Eiffel or BCPL) may make a good
introduction.
Neil
Good point -- they're very easy to implement. The lexer for Luddite
itself is only 77 commented, white-space-ridden lines. The point I
wanted to make in that article was that most of the work in implementing
a multi-language lexer was already done, and showed how to implement
the rest.
- Eric
[1] Very-High-Level Language
_______________________________________________
Scintilla-interest mailing list
[email protected]
http://mailman.lyra.org/mailman/listinfo/scintilla-interest
