Re: cvs: php4 /ext/xml xml.c

From: Date: Sun, 15 Jun 2003 16:55:06 +0000
Subject: Re: cvs: php4 /ext/xml xml.c
References: 1 2 3  Groups: php.cvs 
Request: Send a blank email to php-cvs+get-21745@lists.php.net to get a copy of this message
I think all memory handling should be done with PHP, as you can see there are many places in simplexml where I avoid a copy by relying on the fact that the libxml2 memory will be allocated by php's mm. -Sterling On Sun, 2003-06-15 at 13:04, Rob Richards wrote: > ----- Original Message ----- > From: Moriyoshi Koizumi > > > > Looks pretty much pretty much how it is done in libxml > (xmlGetGlobalState). > > > Will this cause any problems mandating the memory behavor in /etc/xml? > > > > As far as I looked over the libxml code, xmlMemSetup doesn't work > > in sync with the other part of the library as it seems to take it for > > granted that xmlMemSetup() is called prior to any libxml initialisation > > routines. That's why I'm under the impression that it's our task to to do > > it in the PHP side. > > If you want to override the memory handling, i.e. currently it uses > free/malloc/realloc/strdup, then yes it will need to be setup. And it has to > be setup prior to xmlInitParser as the globals will be allocated via that > particular memory handler. > > Another thing, if the memory handler is changed, I am not sure what the > affect on trying to debug the libxml memory will be. This is very important > as, at least working in dom, there are so many scenarios that can happen and > it would take forever trying to fix leaks coming from libxml. > > any reason you want to change the memory handling? I would just think that > it would be one more thing that we would have to deal with. > > What I meant about the xmlGetGlobalState was how it handle the one time > initialzation under threading. > > Rob -- "C makes it easy to shoot yourself in the foot; C++ makes it harder, but when you do, it blows away your whole leg." - Bjarne Stroustrup

« previous php.cvs (#21745) next »