> This is not a theoretical concern. If someone, at some point, defines sub
> CGI {} your constructor suddenly fails. If you're using something along the
> lines of Animal::Bear->new(), and the author of Animal.pm defines a Bear()
> method, your constructor suddenly fails. This is action at a distance.
> It's a pain.
You miss my point. I'm not arguing that. I'm arguing that this:
$SIG{PIPE} = Plumber; # "Plumber" or Plumber()?
Is just as much of a pain because of action-at-a-distance. Solving this
problem with highly-specific solutions gains us nothing but more
syntactic inconsistencies and ambiguities, like these:
stat->{'mode'}
stat->mode
working massively differently. That's confusing!! It is!!! Really.
We need to step back and fix the *problem*: barewords vs. functions are
ambiguous. In *all* contexts. Always. You can screw yourself with any of
these equally:
$SIG{TERM} = IGNORE; # "IGNORE" or IGNORE()?
print output @stuff; # print(output(@stuff))
# or output->print(@stuff)?
$r = new CGI; # new(CGI) or CGI->new?
$q = CGI->new; # "CGI"->new or CGI()->new?
$name = first . middle; # "firstmiddle" or
# first() . middle()?
$total = number + 42; # number() + 42 or
# "number" + 42?
These are *all* action at a distance. The problem has nothing to do with
->. The fix shouldn't either.
The problem is this: barewords and functions are ambiguous. Always.
*ALL* contexts. What we *don't* need is a bandaid holding on a severed
limb (RFC 244). What we do need is a way to fix the actual problem,
which this RFC addresses. Perl is already too hard and inconsistent
enough.
-Nate
