PHP 4.0 Bug #9746 Updated: pg_close (or autoclose) closes wrong connections

ID: 9746 Updated by: romolo Reported By: rich@horde.org Status: Open Bug Type: PostgreSQL related Assigned To: Comments: Try current CVS release, at lest in my case the bug seem to be solved. The problem appeared to be in the close function as it was closing first the specified connection and than the default one (that in the case of same connection ID are exactly the same) so that the 2nd unwilled close was really closing the connection and not just removing a ref to it. Previous Comments: --------------------------------------------------------------------------- [2001-03-14 16:38:02] sniper@php.net Try latest CVS snapshot from http://snaps.php.net/ as this should be fixed. --Jani --------------------------------------------------------------------------- [2001-03-14 15:37:28] rich@horde.org I tried with php4-200103141145; still as described above. --------------------------------------------------------------------------- [2001-03-14 16:21:41] rich@horde.org I recall this coming up before, but I can't find it in the list of open or closed postgres bugs. It appears as though PHP is using the arguments to pg_connect to uniquely identify connections. For example: $dba = pg_connect("user=foo pass=bar dbname=baz"); $dbb = pg_connect("user=foo pass=bar dbname=baz"); pg_close($dba); # $dbb is closed now too Were one to use $dbb at this point, one would see PHP Warning: 1 is not a valid PostgreSQL link resource in [file] at [line]. But, $dba = pg_connect("user=foo pass=bar dbname=baz"); $dbb = pg_connect("pass=bar user=foo dbname=baz"); pg_close($dba); # $dbb is still open and $dba = pg_connect("user=foo pass=bar dbname=baz"); $dbb = pg_connect("user=foo pass=bar dbname=baz "); pg_close($dba); # $dbb is still open The bug doesn't require that the handle be closed explicitly; I discovered it here where one of our programmers was having connections close when one fell out of scope. I tried moving back to PHP 4.0.2 but the bug's there too (so it doesn't look to be a regression problem); on the other hand, a DEC Alpha running Digital Unix 4.0e and PHP 4.0.2 does *not* exhibit this behaviour. If you're unable to duplicate it, please get in touch with me, and I'll see if I can make arrangements for a testing environment. I'll poke through the code myself for something obvious, but I'm unfamiliar with PHP internals in the first place (up to now, at least). --------------------------------------------------------------------------- ATTENTION! Do NOT reply to this email! To reply, use the web interface found at http://bugs.php.net/?id=9746&edit=2