NetNews Usenet Archive 1992 #26/NN_1992_26.iso/spool/comp/graphics/opengl/171

Internet Message Format | 1992-11-07 | 5.1 KB

Path: sparky!uunet!nwnexus!ole!rwing!fnx!sgihbtn!jb
From: [e-mail]
Newsgroups: comp.graphics.opengl
Subject: Re: (FAQ?) Where to get opengl spec/conformance test
Message-ID: <[e-mail]>
Date: 5 Nov 92 22:50:42 GMT
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
Sender: [e-mail]
Organization: Sierra Geophysics, Inc.
Lines: 97
Nntp-Posting-Host: mcknly

In article <[e-mail]>, [e-mail] (Nitin Borwankar) writes:
|> In article <[e-mail]> [e-mail] (Allen Akin) writes:
|> >In article <[e-mail]> [e-mail] (Nitin Borwankar) writes:
|> >|
|>
|> >We want to do both, of course.  Without consistent implementations, the
|> >interface won't be used in "volume" applications.  If you ask, I think
|> >you'll get similar opinions from other vendors and potential users of
|> >OpenGL.
|>
|> X is used in volume applications that are cross platform, although everyone's
|> X is a little different from the others - enough to make it inconsistent
|> by your definition.  However the vendors find it worth there while to take
|> responsibility for porting across those acceptable variations.
|> The result *is* high volume in spite of the inconsistencies.
|> Hell, you even have to rebuild some stuff when the OS goes through a *minor*
|> revision. We have to live with it.

  Ha! I've got more "works with X" applications that blow up than I can
shake a stick at. Especially in the area of HIGH DEMAND, STRESSES THE EDGES
scientific work, its a BIG BIG PAIN to port to "brand X" X11. For example,
we just don't support X11R3 in some of our apps, and never will. DO YOU?
Do you use shm extensions? shape extensions? bezier ...? Blecch.

|>
|> There will always be those things, and part of the reality of being a
|> cross-platform software developer is living with these and creating a
|> software development framework that can manage these inconsistencies.

 Right, so as a developer, I fully endorse SGI to maintain a "1 open GL,
no subsets allowed" ideal . And PEX and all its subsetting can go rot
if I had my way.
|>
|> At some point it becomes important to ask the question -
|> "How close is good enough ?"
|> If you're shooting for 100%, I'm afraid that may be more than you need to
|> generate volume.
|>
|> >
|> >| That's one of the reason's you can buy NT today for development at $69.
|> >| The development costs will be paid for by *later* sales jump started by
|> >| this very wise marketing decision to make things easy for developers.
|> >
|> >Note that Microsoft is not giving away source code for NT. :-)
|>
|> I was only hoping for an SGI certified *run-time* environment on selected
|> high-volume platforms.
|> I don't think everyone wants the source to OpenGL.
|> I think people want a low cost *deployment* platform that will kick start
|> high volume.
|>
|> I know you want purity as well as high volume.
|> However at some point these may be impossible to achieve simultaneously.
|> At that point which one is more important ?
|> Considering that SGI is a business and not a standards body I suspect
|> volume is the bottom line and purity is seen as a means to it.
|> In case purity is becoming counter-productive to the original intent
|> I propose that  the earlier premise ( without consistency high volume
|> will not be generated) be re-examined and gently relaxed till some
|> equilibrium between desired purity and vendor demand is observed.

   Short-term concessions are what destroy standards efforts.
|>
|> >
|> >I agree with your reasoning, though.  What we're really talking about
|> >here is accessibility.  People are annoyed that OpenGL isn't available
|> >at low cost on a wide variety of platforms right away.  That's
|> >understandable.  Fortunately, it's a temporary situation.

   Sure is. From what I've heard from hardware vendors, there are OPENGL
solutions firming up in all the major camps -- realistically, with orders
or magnitude better overall usability than half-baked PEX & such solutions.
For example, the PEX "Hello world" handout at the SIGGRAPH was pages &
pages & pages long -- and had dozens of places where it was incomplete &
non-functional, depending on your particular vendor's "brand" of PEX.
How many lines of OPEN GL does it take to draw one colored 3-d line?
How many if-this-server-then-do-this-instead code do you need to write?

|>
|> I missed the earlier part of this discussion.  Was a relatively short
|> timeframe mentioned ? Dates ?
|>
|> In any case, all the comments here seem to enthusiastically support OpenGL
|> and the annoyance is actually a good sign for SGI, since people do want it.
|>
|> >
|> >Allen
|>
|> Nitin Borwankar,
|> [e-mail]        ***** visualize whirled peas *****
|>

--
 JB West,  Manager of Graphics Dev't          |          Halliburton
 Sierra Geophysics, Inc.                      |     EEEEEE    SSSSSS  GGGGGG
 A Halliburton Energy Services Group Company  |    EEE        SSSS   G   GG
 [e-mail] "My opinion, not ESG's"        |   EEEEEEE   SSSSSSS GGGGGGG