I should chime in here; Michael is right. It's far too late (by at least 12
months) to make changes in OpenGL in Win2K. If there was overwhelming
support, we could look at making changes post W2K, but getting uniform
support across all the legacy platforms would be extremely difficult to pull
off. For Win95 and Win98, we could put a patch on our website, but we'd
also have to make a redist that ISV's (and maybe IHV's??) could ship with
applications. For NT4 and W2K we are required to use the service patch
mechanism. So we end up having to deal with 6 different release pipelines
and QA efforts.
Steve
-----Original Message-----
From: Michael I. Gold [mailto:[EMAIL PROTECTED]]
Sent: Thursday, October 14, 1999 10:08 AM
To: [EMAIL PROTECTED]
Subject: Re: [oglbase-discuss] Updated GL_EXT_get_proc_address spec
At 08:00 AM 10/14/99 -0500, Stephen J Baker wrote:
>On Wed, 13 Oct 1999, Brian Paul wrote:
>
>> I believe the only outstanding issue is whether glGetProcAddressEXT
>> is indeed the right function name. The alternative is to use
>> glXGetProcAddressEXT to return pointers to core GL functions in
>> addition to GLX functions.
>>
>> I'm in favor of the former (having glGetProcAddressEXT).
>
>I vote with you on this one. It reduces the amount of #ifdef
>hacking when you move to a non-X-based OpenGL. (We could fondly
>hope that Microsoft would support glGetProcAddressEXT, so the
>only time you'd need wglGetProcAddress would be when querying
>wgl functions)
Okay, there are a couple of problems here:
1) The extension as now written CANNOT be implemented on Windows (w/o
assistance from Microsoft). No way, no how. The previous spec could have
been. This IMO is not a good property for an ARB extension candidate. If
its a GL extension, it should be implementable by the ICD, without any
assistance from Microsoft. And again, semantically it makes no sense from
opengl32.dll to export a context-independent GL entry point.
2) From a practical standpoint, I believe its unlikely MS would support
this extension; they have already solved the problem with
wglGetProcAddress, and many many Windows apps use this without the level of
pain that some here would assert. Even if MS did decide to support this in
Win2k (ain't gonna happen), there is no way to get it working on Win95,
Win98, and WinNT4 without a patch. So much for cross platform
compatibility.
If we make this a GLX extension, all considerations of the impact on
Windows go out the window.
