On Sun, Aug 8, 2010 at 1:09 PM, David Gamey <[email protected]> wrote:

> Beyond the acknowledgment that part of the code is a kludge and the desire
> to better integrate with the rest of the language.  Has the
> problem/challenge been defined in a bit more detail?
>

I don't actually acknowledge or accept the notion that Sudarshan's thesis
work is a kludge. But it was subject to the same criticism that patterns in
Snobol4 received: it was a separate sublanguage residing within a host
language, something which Icon "fixed" via string scanning. In any case, we
have an Snobol4-like pattern type, where the patterns are data values, are
composable to form larger patterns, and can evaluate certain expressions at
certain points (mainly, assigning variables to hold substring pieces that
were matched) during a pattern match. It is at present independent of the
string scanning control structure which was invented to expose and
generalize certain internal mechanisms of pattern matching, namely
generators and goal-directed evaluation.

What we want is a integration of these two mechanisms. Patterns should be
usable in the middle of string scanning, rather than be a separate
mechanism. For reference, Sudarshan's thesis is at
http://www.cs.nmsu.edu/~sgaikaiw/Thesis.pdf. Some interesting comments by
Steve Wampler, with responses from me, are at
http://www2.cs.uidaho.edu/~jeffery/unicon/pattern-sbw.html. That
conversation is now somewhat dated, but still may provide some useful
context for the conversation.

Clint
------------------------------------------------------------------------------
This SF.net email is sponsored by 

Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev 
_______________________________________________
Unicon-group mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/unicon-group

Reply via email to