On Mon, 8 May 2000, Michael Vance wrote:
> On Mon, May 08, 2000 at 10:17:57AM -0500, Stephen J Baker wrote:
>
> > * Carry on debating because I must have either A or C.
>
> Dear God, are you insane man? My small bit of input (I don't consider
> myself smart enough to actually vote):
No,no,no - you misunderstand. I (personally) am not saying that
I could live with A or I could live with C. I'm saying that
this attempted compromise vote should be "Are you willing to
make this compromise (B) - or not?" - with no implied statement
of what the alternative should be if you are not prepared to
compromise.
If you can't stand B and absolutely must have A or absolutely
must have C then vote "NO" - and we'll figure out some other
way to resolve this.
> - It is trivial for me to fix things for recompilation--in that
> respect, the Most Correct solution is best, not the Most Preserving.
I understand that position - but a significant number of people
(a majority in fact - as of the last vote) don't agree with you.
Allowing non-ABI programs to run as they did before is an important
goal for some of us here.
But let's not get into the whole debate again - we are here to
decide. If you need more debate - just re-read the archives.
Everything that could be said has been said - it's all down to
philosophy and opinion at this stage.
> - However, the above must work within the constraints of "The ABI
> shall never break". As a binary application vendor with users that
> recompile their system libraries, this is very important.
Indeed - but the question is "Should the ABI be allowed to break
existing source-code distributed programs in the name of some
future convenience?"...that's entirely a matter of opinion,
philosophy and politics...which is why we have twice voted...and
clearly intend to do so again.
> And, as an aside:
>
> - The weird discussion the other day about some glext structure with
> pointers sounded very hokey. For my last two projects, C++ classes
> encapsulated the GL pointers, with additional logic for tracking state
> changes, etc. Ie, I'm doing the work anyway. If nothing else, Bernd
> has said he would contribute the qgl-style interface the Quake games
> have used, but we both agree this should be contrib-style work,
> nothing codified by the ABI. Keep things clear and focused.
Yep - certainly "Not In Version 1.0.0" - I think we are all agreed
on that. This would be some new library with a 'glut-like' status
with respect to the ABI (ie you can expect it to exist everywhere
but it's not a formal part of the spec).
Steve Baker (817)619-2657 (Vox/Vox-Mail)
L3Com/Link Simulation & Training (817)619-2466 (Fax)
Work: [EMAIL PROTECTED] http://www.hti.com
Home: [EMAIL PROTECTED] http://web2.airmail.net/sjbaker1
