Re: cvs: php4 /ext/standard output.c php_output.h /main main.c

> > This change breaks several tests which rely on an implicit > > flush of output buffers before the end of the request. > > There are also various real-world applications which rely on > > this feature.. > > I imagined it could break some things (as I said in the CVS commit > message), but I think this is still a good step towards consistency, as the > Apache module cannot emit output during shutdown. Do you think it's worth > it to create (or leave) an inconsistency between the modules? The implicit flush at the end of the request works beautifully with Apache, so I'm not sure that we are talking about the same thing. Specifically, you seem to be referring to output which is generated by if (PG(modules_activated)) zend_try { php_call_shutdown_functions(); } zend_end_try(); while I'm concerned about output from zend_try { php_end_ob_buffers((zend_bool)(SG(request_info).headers_only?0:1)); } zend_end_try(); - Sascha Experience IRCG http://schumann.cx/ http://schumann.cx/ircg