On Tue, 27 May 2003, Michael Mauch wrote:
> Michael Bretterklieber wrote:
>
> > we discovered a crash using 4.3.2RC4 with oci8, here's the bt:
>
> Is your Apache linked to pthread, like it's written on
> <http://www.php.net/oci8>?
sure, and all other neat things like library-preload, etc.
we usualy start this script with the cli-php, but sometimes also from
apache.
>
> And your extension_dir .../no-debug-non-zts-20020429
> is probably the wrong one, as you are using --enable-debug.
>
we don't load any extensions dynamicaly (all exts are staticaly
compiled in) and usualy we use a non debug-build :-)
however we could workaround this, we had a function like this:
function foo()
{
$db = new DB;
$db->query(..)
..
}
this function was invoked several times, and our database-objects
connected also several times to oracle, then I declared $db static:
function foo()
{
static $db;
$db = new DB;
$db->query(..)
..
}
et voila, the script didn't crashed anymore :-)
It looks like that the oci extension has some bugs in freeing resources,
because according to the bt, the oci module freees an already freeed
resource, if I'm right, or maybe a problem with the reference counting,
...
tomorrow I can send the output with ociinternaldebug(true)
another hint, I tried all php versions down to 4.2.1, and all versions
caused the script to crash.
bye,
--
------------------------------- ----------------------------------
Michael Bretterklieber - http://www.bretterklieber.com
JAWA Management Software GmbH - http://www.jawa.at
Tel: ++43-(0)316-403274-12 - GSM: ++43-(0)676-84 03 15 712
------------------------------- ----------------------------------
"...the number of UNIX installations has grown to 10, with more
expected..." - Dennis Ritchie and Ken Thompson, June 1972
