F.e. we have @a and $a = [EMAIL PROTECTED]; the same:
push @a,1,2,3 push $a, 1,2,3
$b = @a $b = $a
(?) say "@a[]" say "$a[]"
(?) myfunc( [EMAIL PROTECTED]) myfunc ( *$a)hm.. i'm not so competent to continue that list.. could anyone kind make a comparison table?
in all other places where they work different, we could use @$a in place of @a, right?
so I can consider @ as "threat as container" specificator, and nothing more.. (in abstract, of course). And we just happen to have arrays with specificator (@) and name without sigil (a), contrary to scalars which always have to had "$" in name ($a).
(
I must admit that I like to have three different namespaces for arrays,hashes and scalars. In Perl5. the _only_ thing I miss is ability to
sub foo (@@) {
(\my @a, \my @b) = \(@_);
...
}
)
but perl6 has somewhat different data model. And we going to use references much more often and easier than in perl5. And we going to have much more different container types, not only arrays and hashes. Why to threat two containers as first class citizens, and all other as mere scalars with references to objects?
so, yeah, it's way too late to make significant changes, but could somebody tell me what was wrong with the following models, apart that they are different from perl5 (and could make it harder to write perl5 to perl6 translators)
I'm talking about unifying namespaces of arrays, hashes and scalars. I could swear i've seen some RFC about it..
1. all variables are scalars, and as scalars they include $ in their name
2. if scalar $var contains reference to array or hash, we could threat it as array or hash using @$var and %$var(or maybe just @var and %var, as shorthands).
With all associated behavior, so @var = ... makes list context to right side of assignment. We could think of it as of "Array" operator, or macro.
2. my @a is ... does right thing: it creates lexical $a, places reference to anonimous array in it, and apply all the traits as current perl6 would do (I'm not sure what differences of traits applying to container and scalars, though). So "my @a of .." is just a syntactic sugar to "my $a is Array of ..".
3. subscripts are written as $var{} and $var[], as we have now and always had before. still @var{...} could be easily recognized as slice.
as far as I can see, we have one thing to lose:
1) one namespace instead of three,
(my $x=$x[$i]; "It's a Perl!" (c))and several good points: 1) simplicity.
2) eliminating of redundancy. We could write all the code involving named arrays and hashes using _only_ named scalars with references to arrays and hashes, so why to have both ways?.
3) no logical separation between CORE and extrnal data containers (yeah, i know about tying, but why we have separate classes and tyers in perl5? is there reason to have write that (tied $var)->method.. in my proposal it would be matter of difference between $var.method and @var.method;
4) using references to containers isn't harder then using containers theyself, because it's a same.
5) one namespace instead of three.
There's plenty of other classifications of variables which would want to have separate nonclashing namespaces, (global/private, instance/class etc.) but we for whateer reason stick with classifing thousands of different container types as three major.
6) syntax would be much the same as what perl5 programmers used to see. just try to look on your perl5 code in my way, and it would work! The only thing that they should keep in mind is that $a and @a is the same..
ah, proposed model could prevent some minor optimizations.. but all thoose optimizations are gone too if you going to use references to arrays instead of arrays itself, aren't they? and if not, than there's no trouble..
P.S Yeah, I see, it looks closer to Certain Other Languages. but why not if it's simplier and maybe even better? and using @ and % as shorthand to that clumsy and verbose to "is Array" or "is Hash" is cool, I think, and it could be a distinctive feature, which could help us to withstand bitter of being less distinctive.
P.P.S I had never feel enough inner strength in myself to be a revolutioner. So I suppose I just don't see something important.. so bring me to Right Way, please..
P.P.P.S. If answer on my "why?" would be "just because!" I would take it silently.
P.P.P.P.S open $file, "filename"; print @file; ...
P.P.P.P.P.S. oooh.. sh$t! i forgot about @_ ! :) but maybe $_ is an Args object? so $_[0] is here, scalar $_ is here for little unprototyped blocks in map and others, and even $_{...} syntax for named args too, it's only matter of viewpoint?..
P.P.P.P.P.P.S I'm going to write a grammar to write such a perl dialect <strike>if</strike> when you'll decline my proposal.
