[antlr-interest] Question about lexer/parser boundaries

Phil Oliver antlr at olivercomputing.com
Tue Jun 5 09:21:21 PDT 2007


Jim - thanks for your last reply. I decided to drop trying to make it 
LL(1) for now and do more in the parser and I'm slowly making 
progress that way - at least I have it compiling and debugging even 
though it's not quite right yet. ANTLR (or perhaps I should say 
ANTLRworks since that's the way I use ANTLR) is great tool but I see 
it as rough around the edges when it comes to helping you identify 
grammar problems or handling its own errors. For example, it took me 
a while to realize that when I get a "Cannot connect" error in the 
debugger in Antlrworks that it was the compiled grammar silently 
blowing up (out of memory because of an unidentified grammar problem) 
without the exception being trapped and reporting back. I would say 
that's a bug in the debugger setup.

There are also serious bugs in the ANTLRworks editor, it's mangled 
find/replace more than once. I think that's a case where re-using an 
existing mature editor (maybe something from Eclipse) would be a 
better approach. In that respect (and others), the commercial ANTLR 
Studio Eclipse plugin is making better use of the framework of a very 
powerful development tool, though unfortunately I'm not sure if the 
developer plans to update it for ANTLR v3 (anybody know?)



More information about the antlr-interest mailing list