Re: RFC 124 (v1) Sort order for any hash as an example of attributes in action

"Brust, Corwin" wrote:
> 
> <snip>
> -----Original Message-----
> From: David L. Nicol [mailto:[email protected]]
> How about
> 
>         %students : ( sort = $$students{^1}{GPA} <=> $$students{^0}{GPA} });
> 
> </snip>
> 
> Ohhhh, this is cool.
> 
> <snip>
> -----Original Message-----
>         my %students : sorted( $ME{^1}{GPA} <=> $ME{^0}{GPA} );
> </snip>
> 
> And this is even cooler!
> 
> -Corwin


I want there to be a STRICTLY SYNTACTIC way to tell if stuff that
comes after an atrtribute specifier is info for the attribute or is
initialization data for the attribute-having data element....
and, duh, I guess it's the presence of an assignment operator,
w/o one of those we're setting the properties of the attribute,
or at least passing the stuff to the attribute's methods, to do
with as it will.

Executive summary:  objects have "attributes" and "value" which
are syntacticly distinguishable.  "type" aka "package" aka "class"
may be considered a meta-attribute, as it affects which attribute
methods will be.  

Internally, the available attributes will depend on what type we're
in, and setting them may imply switching to a different type: 

Setting the :sorted on a hash switches it to a sorted-hash, with 
a variety of implications (maybe all hashes are sorted, and the
default sort value is a (hash-bucket, arrival time) pair)

So, setting an attribute in this instance will have the effect
of changing the container's internal representation, and filling
a virtual method with a coderef.


-- 
                          David Nicol 816.235.1187 [email protected]
                      Does despair.com sell a discordian calendar?