Andi Gutmans wrote:

future: <?php $a = [1, 2, 3]; $a[1] = [1]; $a = [1]; [$a] = $a; $a[] = []; echo $a.[1]; // not a parse error ?>

now: <?php $a = array(1, 2, 3); $a[1] = array(1); $a = array(1); list($a) = $a; $a[] = array(); echo $a.[1]; // now a parse error ?>

For what it's worth, the examples above show why I'm a strong -1 on this idea. It introduces the possibility of weird misunderstandings that just don't exist now, and also introduces a new way to accidentally break your script that wasn't possible before.

The only way I could see solving this would be:

1) [] != list() 2) use a[] as in $a = a[1, 2, 3] but then we might as well stick with array().

Either way, [] is overloaded in the new definition:

1) array access 2) ArrayAccess object offsetGet/offsetSet 3) incremental index ($a[] = 1;) 4) array creation [5) list()]

Whereas now, it is only 3 possible meanings, all of which are consistent with a few quirks in the ArrayAccess one as of PHP 5.2.

Greg