buff_t code [was: [cvs] ... lexer.c ...]

[David Relson]

On Sat, 31 Dec 2005 17:36:11 +0100
Matthias Andree wrote:

> On Sat, 31 Dec 2005, David Relson wrote:
> 
> > Confession time:->  I don't fully understand the buffering code,
> > myself.  When developing it, I started with a vision of what was
> > needed.  However as the implementation progressed, various conditions
> > cropped up that needed special handling.  The end product is the result
> > of tweaking and "debugging into submission".
> 
> OK, let's fork the 1.1 branch in February 2006 or so, and then clean up
> all the buffer handling.
> 
> > The "if (c == '\0')" check that you mention is an example of "doing
> > what works".  Without this "early out" from the loop, there are
> > problems. Possibly this loop can be written as
> > 
> >   for (....)
> >    if ( !isalnum(c))
> >       return false
> > 
> > or similar.
> 
> OK. We'll leave it as it is now, since we know that fixes the problem,
> but we'll have to do some more debugging.

Cleaning up the buffer handling code sounds like a good idea.  Having
time to think about it and plan is a good thing.  The cleanup won't be
easy :-<  Are you volunteering?

In the meantime, I've got the lexer.c test running and will let you
know where it goes.