Re: [oglbase-discuss] OK, back to Re: 2nd draft:GL_EXT_get_proc_address spec

[Thomas Roell]

In your message of 17 September 1999 you write:

> 
> Thomas Roell wrote:
> | 
> |         ... Suggesting a predefined mechanism for distpatching that
> | all libGL implementations should use is simply not acceptable. 
> 
> I believe it's required, if a single libGL must dispatch commands to
> drivers written by independent developers.  Especially in a multihead
> system.

The goal here to define a list of prerequisites so that multiple
different libGL implementations that are binary compatible can coexist
on IA32 LINUX. The goal is not to define a driver API for any specific
libGL. It is quite clear that for any give libGL implementation there
has to be a driver API (which is what you are talking about).

However going into all the implementation details distracts to much
from the original question of whether glXGetProcAddressEXT() should
return a pointer that is associated with the context that was current
at call time. 

It has been shown by multiple people with different approaches that
there is no technical requirement to enforce this. On the other hand,
as extensions are only queryable via glGetString() which in itself
relative to the current context, it makes semantically sence to have
the same behavoir for glXGetProcAddressEXT(). If an implementation
still returns only one unique pointer for all contexts, nothing is
lost.

- Thomas
-- 
             Thomas Roell   /\         An imperfect plan executed violently
             Xi Graphics   /  \/\ _     is far superior to a perfect plan. 
         [EMAIL PROTECTED]   /   /  \ \     
                         / Oelch! \ \             George Patton