[antlr-interest] Lexer speed comparison

Alex Sedow alexsedow at mail.ru
Sat Feb 28 01:03:50 PST 2004


Hello Terence!

> There are many companies that use
> ANTLR for real work; just check the showcase. :)

I know that, for example many companies use ANTLR java grammar to create its
commercial projects (RefactorIt, CodeWizard, etc.). But the question is how
these companies use ANTLR? As part of parser code generation, debug/research
grammar, or as code generator at start project (without using ANTLR for ocde
generation later) and than hand-fixing generated code on whole project
development trying speed-up it?

Turning back to rc-parsers (how I write it). On first stage write and debug
grammar using ANTLR (this is very simple and fast process). On second stage
write lexer (from the beginning), hand-overwrite parser and add tree
structures and debug it. In this "process" I use ANTLR (to save time), and I
may/must say that I use ANTLR. But end user that run program will not have
dealings with ANTLR generated code, because it absent in exe-file.

Finally. ANTLR is good, well-documented and useful tool. At the current time
main thing that it need is speed.

Two words about ANTLR3 DFA-like lexer. I hope it will be based on switch
statement (like re2c (for example re2c used in Wave Preprocessor)), because
table-driven lexers too big (when I try use PCCTS lexer for rc-files it
create is about 200K exe-file (of course I use table compression)).

Alex.



 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
     http://groups.yahoo.com/group/antlr-interest/

<*> To unsubscribe from this group, send an email to:
     antlr-interest-unsubscribe at yahoogroups.com

<*> Your use of Yahoo! Groups is subject to:
     http://docs.yahoo.com/info/terms/
 



More information about the antlr-interest mailing list