Description:
------------
When I "upgraded" to multibyte char encoding and functions, I discovered that I could no longer rely on $s{strlen($s)-1} to access the last character in a string. Strlen correctly counted only whole chars, but the indexing came up short by the number of double-wide chars in the string and hit the wrong char. I could (and did) patch my scripts with rtrim, but others may not be so lucky. I had an idea...

Suggestion:
Some time ago, square brackets were deprecated in favor of curly braces for individual char access in a string. Could both be employed, each for a different purpose? Could curly braces to be used for multibyte-sensitive indexing? Could square brackets, as always, offer simple byte-wise access?

Reproduce code:
---------------
// After a loop chains field names + commas...

$sql{strlen($sql)-1} = ' '; // Replace last trailing comma with a space
$sql.= $where_clause;

// Because my aim was at the very end of a string, rtrim could solve
// my problem. Less specialized cases would not be so easy.


Expected result:
----------------
// There should be two chars in quotes and no comma between the quoted
// value and WHERE

... x="**" WHERE ...


Actual result:
--------------
// Two mb chars elsewhere in string displaced the substitution
// by two places to the left, wiping out a quoted char and leaving
// the comma:

... x="* ",WHERE ...

// Could it be possible for $s{} to index by mb chars? If so, then
// $s[] could be resurrected to index by bytes as both do currently.