Re: session module

brad lafountain wrote:

I looked at your changes to php_session.h. I saw that you added PHPAPI to the .h but don't you need to add them to session.c too?

Thanks, I forgot to commit session.c. I'll commit it later :)

Additionally i think php_get_session_var and php_set_session_var should be exported also.

You are going to add external serializer, right? Then these function does not have to be called from other module, since it is called by session module. IIRC.

i think we need PHPAPI int ps_globals_id too

Hmm. I'm not sure. I thought it is not needed, but I don't build PHP under windows. I could be wrong. Anyone?

As for startup, we probally need a non-static version of php_session_start? What do you think?

This one need to discuss, since this could be a problem. For example, 1) a module start session internally. 2) User set session module paramter, since user is not starting session automatically. 3) User try to start session with session_start(), but session is started already... IMHO, you are better to document that your module requires to session module and session is activated. If session is inactive, you can return error. There may be better to have a function returns current session status for other modules. -- Yasuo Ohgaki

- Brad --- Yasuo Ohgaki <yasuo_ohgaki@yahoo.com> wrote:

derick@php.net wrote:

Hello, Sessions are complex enough already. Doing hacks on the hash by yourself is a very bad idea IMO. Derick

I agree, it could be a bad idea. I only remeber how session module initilizes $HTTP_SESSION_VARS/$_SESSION. Besides, adding element to PS(http_session_vars) will lead inconsistency with global variables unless PS(vars) are taken care of. (there are enough inconsistency between global session vars and $HTTP_SESSION_VARS, though) Call for new function? -- Yasuo Ohgaki

_________________________________________________________ Do You Yahoo!? Get your free @yahoo.com address at http://mail.yahoo.com