At 2003-11-05 09:59 +0100, Michael Walter wrote:

Very cool.

How about supporting .. syntax, btw. as in [1..3] or ["a".."z"]? Might no be the worth, just thinking out loud ;)

I'm also in favor of a shorter notation for array() and list().

In fact this also helps to ease the problem I have
with 'foreach ($A as $I=>$d)', because it would
then be possible to write: while ([$i,$d]=each($A)).
It's still not as short as I'd like it to be:
'while ($i=>$d in $A)', but it's already a lot
better.

I'd also like to see ranges supported and propose
to add notations for inclusive and non-inclusive
ranges as:
1..3 inclusive at both ends so equal to 1,2,3
1.<3 non inclusive at the end so equal to 1,2
1>.3 non inclusive at the beginning so equal to 2,3
1><3 non inclusive at both ends so equal to 2

Of course generally only the first two variations
are being used.

(Please note that this syntax conflicts slightly
with the syntax of floating point numbers and
compares. Perhaps it's solvable by scanning '..',
'.<', '>.' and '><' as tokens first and not
allowing ranges with floating point numbers
anyway.)

By the way, Python has three 'group' notations:
[] for a list
{} for a dictionary
() for a tuple

Very confusing (and one of the reasons I prefer PHP
over Python ;-), and I agree with Rasmus that at least
here the symbols no longer make clear what is being
meant, but I think that just [] for the values in an
array is very intuitive because the same symbols are
being used to add the index of an array. (But {}
from the set notion in mathematics would be fine too,
I think).

By the way, don't forget to allow a comma at the end
of a list: 1,2,3,

This especially makes sense when the elements are
listed each on a seperate line:
1,
2,
3,

Greetings,
Jaap

Christian Schneider wrote:

I propose to add an alternative (backward compatible) short array creation syntax:
$a = [ 1, 2, 3 ]; and $a = [ 'a' => 42, 'b' => "foo" ];
It can also be used in function calls:
img(['src' => "logo.gif", 'alt' => "Logo"]);
Reason behind this change: Arrays are used a lot and should therefore have as little syntactic overhead as possible. And I think the short syntax is also easier to read and write.
A patch for the parser is trivial and is attached for Zend2.
Note: I checked the newsgroup archive but couldn't find a discussion about this. After not hearing back about my proposed enhancement to debug_backtrace() and the dangling comma for function call parameters being rejected I wonder if I'm using the right mailing list for this :-)

• Chris

---

Index: Zend/zend_language_parser.y

RCS file: /repository/ZendEngine2/zend_language_parser.y,v
retrieving revision 1.127
diff -u -r1.127 zend_language_parser.y
--- Zend/zend_language_parser.y 19 Oct 2003 08:38:48 -0000 1.127
+++ Zend/zend_language_parser.y 4 Nov 2003 23:32:12 -0000
@@ -581,6 +581,7 @@
| '@' { zend_do_begin_silence(&$1 TSRMLS_CC); } expr { zend_do_end_silence(&$1 TSRMLS_CC); $$ = $3; }
| scalar { $$ = $1; }
| T_ARRAY '(' array_pair_list ')' { $$ = $3; }

•   |       '[' array_pair_list ']' { $$ = $2; }
   |       '`' encaps_list '`'             { zend_do_shell_exec(&$$, &$2 TSRMLS_CC); }
   |       `T_PRINT` expr  { zend_do_print(&$$, &$2 TSRMLS_CC); }
  

;