Re: Bug in crm114 parser? Or is it me again?

   From: "Ger Hobbelt" <[email protected]>
   >
   > buggaFARRR

   Whoopsky. Yep, you are correct. Forgot the regex subexpression braces in there.

   <evil grin> Something for crm114 v3.0 to have a la GCC and other
   compilers which warn you when your arg count does not match the %
   items in your 'C' printf: "did the crm script programmer provide as
   many subexpressions as (s)he did provide variables to store them in?"
   if not, warn.

Not quite.  

    match /a(bad)*(clock)+/   (:1: :2:)

is perfectly legal, and reasonable (you might want repeated "clock"s,
but have no desire to bind to them).

The other way is an error:

    match  /this is (wrong)/   (:1: :2: :3:)

which I don't recall if we check for or not.

   Of course, that would not work for run-time generated regexes, but it
   would help script novices like me. :-P   Alas, that's for someone else
   to introduce in the compiler.

This is why the current compiler is a minimal as it is (that, and
lazyness.)  

If you pre-parse too deeply, and there's an embedded :*: or :+:
redirection, or any EVALs (since EVAL is repeated evaluation) then
all of the pre-compiled substring pointers *give you the wrong answers*. 

In the general case, finding the start and end of each chunk 
*correctly* is as far as you can do at "compile" time.

	    - Bill

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/