[antlr-interest] Possible incremental parse approach?

[Randall R Schulz]

On Sunday 14 October 2007 14:28, Rick Mann wrote:
> Maybe I could do this? Run my parser in a separate thread. The input
> character stream can block when it needs more data, at which point
> the main thread of my GUI app could put data into its queue...

There's a problem with this. Ordinarily, ANTLR treats lexical analysis 
as a distinct phase of processing which must complete before parsing 
begins. If the lexical analyzer doesn't get and end-of-file, it will 
effectively just wait until it does.

If your parser does not generate an AST, then that's fine. The action 
code does everything (triggering whatever follow-on processing is 
required). But if you're building an AST for parsgin by a tree parser, 
then you have to figure out what to do about the fact that the tree is 
never complete _and_ the fact that it grows without bound throughout 
the execution of the system.


> Does this approach sound like it would work?

It sounds problematic to me.


Randall Schulz