Contrary to Perl 6 my approach is still compatible, because you can transform them back into standard Perl 5.

      ... until the first time you run into a precedence problem, for example with hash or list slices.

    Don't know what you mean ... all my examples were parenthesized so precedence problems shouldn't occur !?!

I mean exactly:

    Actually it's practically impossible to implement this with a simple code-filter,...

My point is that the proposed transformation rules are well-defined and backwards-compatible within perl5, contrary to perl6. One can always take code with new sigils and transform it back into normal sigils and parentheses to run it with an old perl 5 version.

The implementation of these rules with static code filters can of course only be - like I already said - an ersatz to demonstrate the concept. This is certainly not meant for production.

For the realisation: If it's not possible to extend the perl parser accordingly, a real macro system is needed (which is N° 8 on my wishlist) That means changing the code after compilation but before execution!

>In other words, you're arguing to add more syntax to Perl, syntax that's difficult to type, to make the simple cases easy

well I think most languages, including perl6, consider it easier that arrays are primarily references and only transformed into a list if it's indicated by a modifier.

Maybe I should express these transformation rules with plain words instead of code, to make them clearer:

1. €arr is a scalar $arr which holds a array_ref
2. €arr[][] can be dereferenced without any arrow-operator, that means is identical to $arr->[]->[].
3. Dereferencing of €arr has higher precedence than transformation to an array:  @€arr[][] == @{€arr[][]}
4. In situations where the parser wants an array to follow (*), e.g. forced by prototype (\@) , €arr is passed directly. That means push €arr,expr is identical to push @$arr,expr

(*) more precisely in these situations an @ has to follow but only the reference is passed.

To help you to get a feeling for what I mean, try this uncompleted(!) codefilter, which is only a hack(!) under construction ...

XSigils.pm

#!/usr/bin/perl

sub ar2list ($) {
    return @{ $_[0] };
}

sub push_aref ($@) {
    CORE::push( @{+shift} , @_ );
}

package XSigils;

use Filter::Simple;



FILTER {
    warn "\n\n\n --- Before Transformation: \n$_";
    s{ push (\s|\() (\s*) €}
     {push_aref$1$2€}xg;
    s{ \@ \s* €}
     {ar2list €}xg; # 2array
    s{ €(\w+)\[ }
     {\$$1->\[}xg;   # dereference
    s{ €(\w+) }
     {\$$1}xg;       # simple scalar
    warn "\n\n\n --- After Transformation: \n$_";
    warn "\n\n\n --- Code execution!\n";
};

1;
[download]

tstXSigils.pl

#!/usr/bin/perl

use XSigils;

$\="\n";

€arr=[1..3];

push €arr,4,5;

print @€arr;
[download]

OUTPUT

--- Before Transformation: 

$\="\n";

€arr=[1..3];

push €arr,4,5;

print @€arr; # (*)



 --- After Transformation: 

$\="\n";

$arr=[1..3];

push_aref $arr,4,5;

print ar2list $arr;



 --- Code execution!
12345
[download]

Cheers Rolf

UPDATE: (*) of course that's more complicated than print @arr, but I wonder how perl6 intends to handle this case ... ?

I think it's time for you to let it go. It's not just your own time you're wasting.