Re: Re: cvs: php4 /ext/standard incomplete_class.c php_incomplete_class.h type.c

Zeev Suraski wrote:

At 14:20 24/07/2002, Yasuo Ohgaki wrote:

I don't think breaking compatibility is not needed now. However, I'm not against to raise fatal error in PHP5. PHP5 breaks scripts in many ways. Why don't we save it for PHP5?

Without paying attention to the specific issue, please snap out of this state of mind. PHP 5 will not be a wholesale compatibility breaking version. To the best of my knowledge, there are no significant compatibility breaking changes in PHP 5. Once the object model is finalized, we'll add a mode in which PHP 5 employs the same object model as v4, which means people should be able to run their PHP 4 scripts without any modifications, out of the box. We should strive to keep it like this as much as we can.

Most PHP4 scripts may be executed w/o modification to run under PHP5, IMO also. Anyway, I'm not against keeping comaptibility between PHP4 and PHP5, too. I'm againt too high warning level that terminates script execution w/o fatal reason now. Without my patch, PHP 4.3.0 will terminate scripts with E_ERROR when user access incomplete object property. It was working w/o any warnings up untill 4.2.x. My patch works the same for both PHP4 and PHP5. It raises E_NOTICE for property access, E_ERROR for method. Older code was raising E_ERROR... If many people think this BC is required, I'm not againt having this BC, too. All I have to do is cast object to array then use the array to access incomplete object's property. If I need to save it back, cast the array back to object. I'm not sure if this works or not, though. (If it doesn't, I'll stringly disagree with the BC) Any comments? -- Yasuo Ohgaki