On Mon, Oct 25, 1999 at 12:59:54PM -0600, Brett Johnson wrote:
> I share your hesitancy. I don't have any illusions about this vote
> being the final word though. I'm just trying to do _something_ to
> move us forward on a decision. We've been stalled for a long time,
> and it doesn't appear to me that we're ever going to all come to a
> consensus. So we have to have some mechanism for guiding a decision,
> since there doesn't appear to be a formal one. If anyone can suggest
> a better mechanism than a vote, please do so.
>
> Jon, maybe you could outline for us how you envision decisions being
> made as to what is contained in the spec?
Straw polls are useful; we should have done this one a long time
ago. Thanks for taking the initiative.
We all need to back off from trying to achieve the 100% perfect
all-singing all-dancing answers in favor of those that (a) the driver
writers agree to implement (b) the app writers agree to use and (c)
happen soon enough to be useful.
As far as actual suggestions, how about: the next time we have
dissention, everyone think about whether they are adding information to
the debate by posting something, or just repeating what they or someone
else has already said instead. If you boil down the basic points in the
context-dependence argument, there aren't many - but there have been
several hundred messages about them. As a forward-looking example: the
actual name of the GL library DOES NOT MAKE ANY FUNDAMENTAL DIFFERENCE.
What matters is that there *is* a single name for it that apps can rely
on.
Not to sound unduly harsh, but if we get into extended pissing
matches over whether "libGL<a>" is a better or worse filename than
"libGL<b>", it's time to give up on doing this via an open mailing list
and try some other model involving only the people actually writing
drivers and libraries.
Jon Leech
SGI
