Re: strictness and fully qualified global vars
On Tue, Dec 28, 2004 at 10:31:37PM -0800, Ashley Winters wrote:
: On Tue, 28 Dec 2004 22:31:47 -0700, Luke Palmer <[EMAIL PROTECTED]> wrote:
: > Ashley Winters writes:
: > > sub foo (Class $who) {
: > > my $thing := $who<$var>;
: > > my &func := $who<&func>; # how would I do this otherwise?
: > > }
: >
: > In current Perl 6:
: >
: > sub foo (Class $who) {
: > my $thing := $::($who)::var;
: > my &func := &::($who)::func;
: > }
:
: Okay, I see. S10 says ::() is the catch-all symbolic naming syntax.
: However, $who would be a reference to a class object itself. Does it
: automagically accept hard-references, or would Class objects have to
: stringify to their global ::*::ClassName?
:
: More to the point, is %::(%foo) an identity op?
We will probably make hard refs work inside ::() just to make it easier
to translate Perl 5 to Perl 6. But possibly it should require a laxity
pragma to enable it. If we end up with people simply writing $::($x)
everywhere instead of ${$x}, we haven't gained anything over Perl 5,
and we've lost strict refs. I don't want people to get in the habit
of using ::() for hard refs unless they explicitly want symbolic refs
as well. I was hoping that the mere length of $::() over ${} would
be enough to discourage that kind of thinking, but now I'm not so sure.
Larry
Re: strictness and fully qualified global vars
On Tue, 28 Dec 2004 22:31:47 -0700, Luke Palmer <[EMAIL PROTECTED]> wrote:
> Ashley Winters writes:
> > sub foo (Class $who) {
> > my $thing := $who<$var>;
> > my &func := $who<&func>; # how would I do this otherwise?
> > }
>
> In current Perl 6:
>
> sub foo (Class $who) {
> my $thing := $::($who)::var;
> my &func := &::($who)::func;
> }
Okay, I see. S10 says ::() is the catch-all symbolic naming syntax.
However, $who would be a reference to a class object itself. Does it
automagically accept hard-references, or would Class objects have to
stringify to their global ::*::ClassName?
More to the point, is %::(%foo) an identity op?
Ashley
Re: strictness and fully qualified global vars
Ashley Winters writes:
> sub foo (Class $who) {
> my $thing := $who<$var>;
> my &func := $who<&func>; # how would I do this otherwise?
> }
In current Perl 6:
sub foo (Class $who) {
my $thing := $::($who)::var;
my &func := &::($who)::func;
}
However, I agree that $Package::var is weird since the variable's name
is $var, not just var. Package::$var looks really odd to me, but maybe
it's the right thing to do. For interpolation, we could just force
curlies; package variables aren't interpolated much anyway.
Luke
Re: strictness and fully qualified global vars
On Wed, 29 Dec 2004 06:55:11 +0300, Alexey Trofimenko
<[EMAIL PROTECTED]> wrote:
> P.S. I have one (almost unrelated to topic) observation: if sigil is a
> part of a variable name, then C< Package::$var > makes more sense than
> perl5 C< $Package::var >. (AFAIK, PHP5 works this way) And this requires
> less magic from perl.
I agree... that's inconsistent.
I have another question: Are package/class/grammar namespaces valid
objects in Perl6? I would assume yes, so you can call methods on them
for meta-purposes.
Would there be a default Namespace::postcircumfix:«< >» operator,
which would make the above code look like: Package<$var> aka
Package.<$var>? This would continue to make sense even when the
namespace objects are passed to a function:
sub foo (Class $who) {
my $thing := $who<$var>;
my &func := $who<&func>; # how would I do this otherwise?
}
I assume the second line can't really be done through stringification
due to singletons. Well, on second thought, you could make $foo.meta
(or whatever) start answering to CLASS(0xDEADBEEF) style classnames.
Those are probably needed for debugging or something anyways.
Ashley Winters
