Re: [ZEND-ENGINE-CVS] cvs: ZendEngine2(PHP_5_1) / zend.h zend_object_handlers.c zend_objects.c zend_reflection_api.c php-src NEWS

Hello Dmitry, Tuesday, November 15, 2005, 2:35:25 PM, you wrote: > > http://cvs.php.net/diff.php/php-src/NEWS?r1=1.2027.2.189&r2=1.2027.2.190&ty=u > Index: php-src/NEWS > diff -u php-src/NEWS:1.2027.2.189 php-src/NEWS:1.2027.2.190 > --- php-src/NEWS:1.2027.2.189 Tue Nov 15 08:29:15 2005 > +++ php-src/NEWS Tue Nov 15 08:35:20 2005 > @@ -1,6 +1,7 @@ > PHP NEWS > > ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||| > ?? Nov 2005, PHP 5.1 > +- Allow recursive calls to __get/__set for different properties. (Dmitry) > - Upgraded PEAR to version 1.4.4. (Greg) > - Fixed bug in mysqli extension with unsigned int(11) being represented as > signed integer in PHP instead of string in 32bit systems. (Andrey) > > http://cvs.php.net/diff.php/ZendEngine2/zend.h?r1=1.293.2.3&r2=1.293.2.4&ty=u > Index: ZendEngine2/zend.h > diff -u ZendEngine2/zend.h:1.293.2.3 ZendEngine2/zend.h:1.293.2.4 > --- ZendEngine2/zend.h:1.293.2.3 Thu Sep 1 06:05:30 2005 > +++ ZendEngine2/zend.h Tue Nov 15 08:35:22 2005 > @@ -17,7 +17,7 @@ > +----------------------------------------------------------------------+ > */ > > -/* $Id: zend.h,v 1.293.2.3 2005/09/01 10:05:30 dmitry Exp $ */ > +/* $Id: zend.h,v 1.293.2.4 2005/11/15 13:35:22 dmitry Exp $ */ > > #ifndef ZEND_H > #define ZEND_H > @@ -265,13 +265,18 @@ > typedef struct _zval_struct zval; > typedef struct _zend_class_entry zend_class_entry; > > +typedef struct _zend_guard { > + zend_bool in_get; > + zend_bool in_set; > + zend_bool in_unset; > + zend_bool in_isset; > + zend_bool dummy; /* sizeof(zend_guard) must not be equal to sizeof(void*) */ > +} zend_guard; > + > typedef struct _zend_object { > zend_class_entry *ce; > HashTable *properties; > - unsigned int in_get:1; > - unsigned int in_set:1; > - unsigned int in_unset:1; > - unsigned int in_isset:1; > + HashTable *guards; /* protects from __get/__set ... recursion */ > } zend_object; > [...] Wouldn't it be better to have the HashTables as members rather than pointers? It would prevent a malloc/free that is imo not necessaary at all. regards marcus