Re: Allow :: as a Fortress operator

On Apr 26, 2008, at 3:36 PM, Eric Allen wrote:

> On Apr 25, 2008, at 6:06 PM, Victor Luchangco wrote:
>
>> I think we want to hold off as much as possible on allowing  
>> multicharacter operators.  Currently, the only multicharacter  
>> operators are enclosers, as defined in the Multicharacter  
>> Enclosing Operators section of the Lexical Structure chapter,  
>> which cause enough parsing complications.  If a "multicharacter"  
>> operator really deserves special treatment, I think we should try  
>> to get the Unicode folks to give it a character of its own, as  
>> they already do for COLON EQUALS, for example.
>
> Don't forget that every token that is made up of a mixture of  
> uppercase letters and underscores (but no digits), does not begin  
> or end with an underscore, and contains at least two different  
> letters is also considered to be an operator. So we already have a  
> very large space of multicharacter operators.

Yeah, sorry--I meant multicharacter, nonword tokens.  We parse words  
and nonwords completely differently, or at least we specify the  
parsing completely differently.  The only complications with words  
(i.e., contiguous sequences of letters, digits, underscores, primes,  
etc.) that I recall had to do with handling numbers, specifically in  
dealing with the decimal/radix point and possible radix specifier and  
allowing small spaces to separate groups of digits.

In any case, as Guy points out, there are syntactic problems posed  
specifically by allowing :: as an operator.  And as Jan pointed out,  
there's already a Unicode character that looks like what you want (U