New Celestia 1.4 FT1.2ish OS X Build 11-16-2005

BlindedByTheLight · Post #1by **BlindedByTheLight** » 16.11.2005, 11:36

Here you go again... hot off the presses from Fridger and Toti. This is essentially a straight compile from their source code.

Available at:

http://www.stevenbinder.net

As usual, I added Doctorjoe's great GOTO spacecraft start-time patch. Just select, say Mir (which isn't viewable in the present day) and then hit shift-G and you will be brought there in BOTH time and space.

Also as usual, I added keyboard short-cuts to toggle specific orbit lines.

I did NOT clean up the menu bars... that's a lot of work for still essentially a beta release. But I'll get to it next time.

If you want to read more about what's been fixed:

http://celestiaproject.net/forum/viewtopic.php?t=8334

That's probably a good place to post any bugs. If you're SURE something is Mac OS X specific, then post it here. Otherwise... in F&T's thread. Remember, as I have stated numerous times, I really have NO idea what I'm doing. Don't be fooled by the fancy words I'm using (like "compile")... the experts here know the truth. I am clueless - though I think they would give me an 'A' for effort...

Post #2by **ElChristou** » 16.11.2005, 18:45

Tx Steven!

julesstoop · Post #3by **julesstoop** » 16.11.2005, 20:30

Almost forgot to say "thank you!" ;)
It's working fine, it seems from a small preliminary test run.
Does anyone have an idea if the Open GL2 renderpaths would be fixable in the near future?
They are available on my machine (PowerBook Al with 64MB 9600, 10.4.3), but produce a very 'silly' result.

Post #4by **selden** » 16.11.2005, 20:37

Jules,

My understanding is that the OpenGL v2 render path works fine on Windows and Linux systems (except for the bug in the current Nvidia drivers for Windows). This suggests that the problem is in the MacOs graphics drivers provided by Apple.

Could you provide an example of the "silly" result?

julesstoop · Post #5by **julesstoop** » 16.11.2005, 23:36

I think you are right that it's probably Apple's Open GL implementation.
Anyhow, it only seems to happen with planets and I've posted screenies from Mars and Saturn.

Note that mars is sunlit from about 3/4 of the front. On the Sarturn shot, the nightside is where the bright white part is on the right side. The squarish parts on the bottom-front of Saturn are continuously shifting (about 4 times a second)

Thanks for taking a look Selden.

BlindedByTheLight · Post #6by **BlindedByTheLight** » 16.11.2005, 23:46

If I'm not mistaken, Apple only has drivers that support OpenGL 1.5 - at least on my ATI card that's the case.

julesstoop · Post #7by **julesstoop** » 17.11.2005, 07:30

Obviously but unfortunately you might be right.
Still realtech's OpenGL extensions Viewer tells me 10.4.3 supports most of the Open GL 2.0 spec.
Besides, I recall clearly that the Open GL 2 renderpaths in Celestia were not available with previous versions of the Mac OS. These two facts led me to believe that open GL 2.0 support had finally landed in OS X..

BlindedByTheLight · Post #8by **BlindedByTheLight** » 17.11.2005, 22:44

Then perhaps, though the OS can now handle 2.0, actuall support is graphics card driver specific? So I have ATI to blame, then?

julesstoop · Post #9by **julesstoop** » 18.11.2005, 07:20

The errors I experience (see the screenshots posted above) are reminiscent of other but similar Open GL-related errors on OS X which seem to occur when the GPU is short on VRAM. I don't recall if they were ATI-specific, but I remember speculation that they might have something to do with Apple using the GPU and VRAM for the GUI, alongside any Open GL app that might happen to be running.
Anyway, Apple is probably to 'blame' in both cases.
I have to say though that Celestia is the only application I have experienced Open GL related errors with on OS X.

Post #10by **dirkpitt** » 21.11.2005, 01:53

julesstoop wrote:The errors I experience (see the screenshots posted above) are reminiscent of other but similar Open GL-related errors on OS X which seem to occur when the GPU is short on VRAM. I don't recall if they were ATI-specific, but I remember speculation that they might have something to do with Apple using the GPU and VRAM for the GUI, alongside any Open GL app that might happen to be running.
Anyway, Apple is probably to 'blame' in both cases.
I have to say though that Celestia is the only application I have experienced Open GL related errors with on OS X.

That's true. Anyway, good news: I've verified via test code that GLSL shaders can now run in the default hardware-accelerated pixel format meaning the OS X builds should be able to access the GL 2.0 render path with small changes to the source code.

Unsupported hardware (e.g., Radeon 9200) may be able to render GLSL features using slow, software emulation (partially accelerated with altivec) while cards that can do fragment shading (e.g., my Radeon 9700) may be able to render some or all GLSL stuff fast.

I'll try to do a test with Celestia soon.

julesstoop · Post #11by **julesstoop** » 21.11.2005, 07:02

That's good news indeed Dirk, thanks for sharing it!

Post #12by **dirkpitt** » 21.11.2005, 17:49

Hmmm.. GLSL seems to be broken for me as well. I compiled fresh from CVS and I too get the "butterfly" effect on my planets (you'll see the same characteristic pattern if you paste the shader code into Apple's Shader Builder program too). No bright-red error shaders either. I'll file a bug report with Apple. Why every single driver update and bug escalation has to go through Apple is beyond my comprehension, especially since it makes no difference in code quality...