Re: RFC: Extracting the action transformation from the scanner

[Hans Aberg]

On 26 Aug 2005, at 01:54, Paul Eggert wrote:

> > > Perhaps it'd be better to drop the postprocessor phase entirely;  
> > > it's
> > > nice in some respects, but I'm afraid it's turning out to be more
> > > trouble than it's worth.
> >
> > What do you have in your mind here?
>
> Nothing concrete yet.  Perhaps just come up with a minimal macro
> language and implement it in C.  I hate to reinvent the wheel, but M4
> isn't working out well.

I used that approach in a parser that generates C++ code, but I  
thought it was a hassle. Generating C++ code is tricky, because  
output tied to one component should often be put in several different  
places. One alternative might be to use the program CodeWorker  
<http://codeworker.free.fr/>; I haven't picked it down and checked it  
out though.

However, the above is for a parser, generating code output. Probably  
much less is required for writing a parser. I think a program like  
Scheme sounds interesting, because then one has access to full  
computing power. But I do not know how good it is at the specific  
task at generating text output, specifically, if it is convenient to  
enter large chunks of text, and if it chokes on large text bodies.  
One can probably only get to know what works by trying.

  Hans Aberg