In regards to http://www.nntp.perl.org/group/perl.perl6.language/21120 which discusses character class syntax in Perl 6, I have some comments to make.

First, I've been very interested in seeing proper set notation for char classes in Perl 5. I was pretty vocal about it during TPC in 2002, I think, and have since added some features that are in Perl 5 now that allow you to define your own Unicode properties with not only + and - and ! but & as well.


If we want to treat character classes as sets, then we should try to use notation that reads properly. I don't see how '+' and '|' are any different in this case: <+Foo +Bar> and <Foo | Bar> should produce the same results always. I suppose the + is helpful in distinguishing a character class assertion from any other, though. To *complement* a character class, I think the character ~ is appropriate. Intersection should be done with &. Subtraction can be provided with -, although it's really just a shorthand: A - B is really A & ~B... but I suppose huffman encoding tells us we should provide the - sign.

Here are some examples, then:

  <+alpha -vowels>        all alphabetic characters except vowels
  <+alpha & ~vowels>  same thing
  <[a..z] -[aeiou]>       all characters 'a' through 'z' minus vowels
  <[a..z] & ~[aeiou]> same thing
  <~(X & Y) | Z>      all characters not in X-and-Y, or in Z

The last example shows <~ which is currently unclaimed as far as assertions go. Since I'd be advocating the removal of a unary - in character classes (to be replaced by ~), I think this would be ok. The allowance for a unary + in character classes has already been justified.

For the people who are really going to use it, the notation won't be foreign. And I'd expect most people who'd use it would actually abstract a good portion of it away into their own property definitions, so that

  <~(X & Y) | Z>

would actually just be

  <+My_XYZ_Property>

which would be defined elsewhere.

What say you?

--
Jeff "japhy" Pinyan         %  How can we ever be the sold short or
RPI Acacia Brother #734     %  the cheated, we who for every service
http://japhy.perlmonk.org/  %  have long ago been overpaid?
http://www.perlmonks.org/   %    -- Meister Eckhart

Reply via email to