Beta 4 RC 1 View as Markdown

Hi,

I rolled a preliminary beta 4 package just to make sure nothing is
seriously broken. You can still commit fixes in the next few hours and
if I don't hear of any serious show stoppers, I'll re-bundle and
release beta 4 later today.

http://snaps.php.net/~andi/php-5.0.0b4RC1.tar.gz

My copy of that doesn't have a configure script in it.

Jan

nevermind, sorry.

Jan

Hi again :)

I rolled a preliminary beta 4 package just to make
sure nothing is seriously broken. You can still
commit fixes in the next few hours and if I don't
hear of any serious show stoppers, I'll re-bundle
and release beta 4 later today.

tests/classes/array_access_005.phpt is failing
for me for quite a long time now. This is on
Linux, FreeBSD and OS X.

The .out file gives:
string(3) "Joe"
string(6) "JoeFoo"
string(9) "JoeFooBar"
---ArrayOverloading---
string(3) "Joe"

Fatal error: Objects used as arrays in post/pre increment/decrement
must return
values by reference in
/home/jan/src/php-5.0.0b4RC1/tests/classes/array_access_0
05.php on line 40

I am not sure if PHP or the testcase is broken here.
Some more tests fail as well, see the complete result at
http://jan.prima.de/~jan/rc1_result.txt

regards,
Jan

== Quote from Andi Gutmans (andi@zend.com)'s article

I rolled a preliminary beta 4 package just to make sure nothing is
seriously broken. You can still commit fixes in the next few hours and if I
don't hear of any serious show stoppers, I'll re-bundle and release beta 4
later today.

Andi,

Not a big issue, but a different behaviour compared with beta3 and some following snapshots.

<?
class Test {
function __set($n, $v) {
throw new Exception('error');
}
}

function f() {
$o = new Test();
$o->foo = 5;
}

f();
?>

Reports:
Fatal error: Uncaught exception 'exception' with message 'error' in /var/www/html/set.php:5
Stack trace:
#0 /var/www/html/set.php(11): Test->__set()
#1 /var/www/html/set.php(11): f()
#2 /var/www/html/set.php(14): f()
#3 {main}
thrown in /var/www/html/set.php on line 5

There is an additional #1 in the stack trace.
When executed outside the context of a function, the reported function is "unknown()".

This also happens with __get, __toString, __call and __clone. But not with __construct and __destruct.

Regards,
Stephane

At 02:33 PM 2/12/2004 +0000, Stephane Drouard wrote:

== Quote from Andi Gutmans (andi@zend.com)'s article

I rolled a preliminary beta 4 package just to make sure nothing is
seriously broken. You can still commit fixes in the next few hours and if I
don't hear of any serious show stoppers, I'll re-bundle and release beta 4
later today.

Andi,

Not a big issue, but a different behaviour compared with beta3 and some
following snapshots.

<?
class Test {
function __set($n, $v) {
throw new Exception('error');
}
}

function f() {
$o = new Test();
$o->foo = 5;
}

f();
?>

Reports:
Fatal error: Uncaught exception 'exception' with message 'error' in
/var/www/html/set.php:5
Stack trace:
#0 /var/www/html/set.php(11): Test->__set()
#1 /var/www/html/set.php(11): f()
#2 /var/www/html/set.php(14): f()
#3 {main}
thrown in /var/www/html/set.php on line 5

There is an additional #1 in the stack trace.
When executed outside the context of a function, the reported function is
"unknown()".

This also happens with __get, __toString, __call and __clone. But not with
__construct and __destruct.

This is a result of the exceptions having been broken in B3 and being
triggered too late (thus, not finding f() twice on the stack).
We will need to investigate it before RC1.

Andi