Re: [PHP4BETA] CVS update: php4/ext/session

From: Date: Mon, 07 Jun 1999 05:38:40 +0000
Subject: Re: [PHP4BETA] CVS update: php4/ext/session
References: 1 2 3 4 5  Groups: php.version4 
Request: Send a blank email to php-version4+get-1349@lists.php.net to get a copy of this message
At 21:55 06/06/99 +0200, Sascha Schumann wrote:
On Sun, Jun 06, 1999 at 10:51:31PM +0200, Zeev Suraski wrote: At 21:44 06/06/99 , Sascha Schumann wrote:
On Sun, Jun 06, 1999 at 06:39:48PM -0000, andi wrote:
Date: Sunday June 6, 1999 @ 14:39
Author:       andi
Update of /repository/php4/ext/session In directory php:/tmp/cvs-serv11552/ext/session Modified Files:
      session.c
Log Message: - Use PS() everywhere.
I wrote it the way it was on purpose. Don't try to outsmart original authors of source code you work on. ;)
Mind sharing that purpose with the rest of us mortals? :) I don't like to reinvent the wheel or double code. That is why the session module uses php_rinit_globals not only for ZTS where this function is necessary, but also for non-ZTS where the function is called directly from the startup code. Using PS(x) in php_rinit_globals will work on Windows, because PS(x) is there some_ptr->x. On Unix PS(x) is some_ptr.x which is obviously different and will fail there.
I don't quite understand where I went wrong. I added that patch so that you'll have thread safety (and I was compiling on win32). I am probably missing something but I didn't quite understand what you were saying. Don't you always need thread safety in rinit? You can have two rinit's in a threaded environment running at the same time. Why will PS(x) on UNIX fail for you? some_ptr.x should work unless you're not defining the variables in a standard way or you're doing something special. Please explain what I'm screwing up. Andi --- Andi Gutmans <andi@zend.com> ICQ: 1568297

« previous php.version4 (#1349) next »