Stephen J Baker wrote:
> 
> On Thu, 16 Sep 1999, Michael I. Gold wrote:
> 
> > We will probably never agree on this, it seems, so I'm going to stop
> > debating it now.  There has been very little input from other implementors,
> > and I would be interested in hearing more of their opinions and less of
> > mine.  :)

Ahem..  Michael, there has been quite a bit of input from other
implementors (myself included).  Input that doesn't agree with your
opinion still counts as input... ;o)

> OK, I've listened hard to the arguments on both sides and I
> think I'm convinced that the realtime penalty for a context-
> independent glGetFuncAddress is either negligable or zero.

Yay!  I'm glad more folks are understanding this concept.

> However, as an application writer, I have to say that I have
> no problem at all with the address returned by glGetFuncAddress
> being context-specific.

I don't see a problem with it in the applications I've written either,
but that doesn't mean that it isn't a problem for some applications
(or libraries as Allen succinctly pointed out).

> It would be marginally nicer if it didn't have to be - but
> in the interests of making life easier for the driver writers,
> I think we should simply say in the spec that addresses returned
> by glGetFuncAddress are *not* guaranteed to work in other contexts.
> 
> I would rather make it easy for all OpenGL's to implement
> glGetFuncAddress with minimal code change than make it harder
> to implement for such a small application-side gain. That
> gives us a better chance of having this Linux specification
> spread to other OS's - which is VERY important IMHO.

I've only heard one implementor opine that a driver would be more
difficult to implement due to GetProcAddress returning context 
independent pointers.  As another implementor, I don't agree.

It's probably true that if we keep the GLX API as similar as possible
to wGL (i.e. glXGetProcAddress has the same semantics as
wGLGetProcAddress) it will be slightly easier to port an existing
Windows ICD.  But nothing about the concept makes a driver inherently
more complex or more difficult to write.

Being able to easily port existing Windows ICD's is the most compelling
argument I've heard so far for making GetProcAddress return context
specific pointers. As an implementor who has an existing Windows ICD
just waiting to be ported to Linux though, I'd gladly trade the small
amount of effort required to implement context independent pointers for
a cleaner GetProcAddress specification.

Cheers!
begin:vcard 
n:Johnson;Brett
x-mozilla-html:FALSE
org:Hewlett-Packard;Worstation Systems Lab
adr:;;;;;;
version:2.1
email;internet:[EMAIL PROTECTED]
x-mozilla-cpt:;19584
fn:Brett Johnson
end:vcard

Reply via email to