the 1.4prev9 bug report thread

[cartrite]

I think I may have found the root cause to this crtl+v crash. It seems to have something to do with star style=points in basic rendering.

cartrite

[Vincent]

Ok, here's the script line that causes the "Control V crash" :

Code: Select all

set { name "StarStyle" value "points" }

I also found out that when I manually set the star style option to "Points" in the Render Menu, the crash comes back. It doesn't happen when the star style option is set to "Fuzzy Points" or "Scaled Discs".

Added later :
Cartrite, I didn't see your post. I think we both found it

[cartrite]

10/4 Vincent I hope this is realy it.

[t00fri]

10/4 Vincent I hope this is realy it.

OK at last, we got it: I rebooted Windows and indeed can consistently now also reproduce the crash: I first put the star style to "points" with CTRL+S and then I cycle through the render paths with CTRL+V. Between OpenGL 2.0 and Basic the crash is reproducably provoked!

If one takes another star style (as I do by default) then the crash does never occur.

+++++++++++++++++++++++++++++++
So Chris must have introduced another bug there, since he was experimenting last week quite a lot with new rendering techniques for point stars!
+++++++++++++++++++++++++++++++

Bye Fridger

[cartrite]

I don't know. I don't have the source code. I hope this really gives some help tho.

cartrite

[cartrite]

Oh Yeah. The crash also occurs if your in basic rendering and change star styles to points.

cartrite

[BlindedByTheLight]

Just curious... is this build a straight build of CVS? The version number for that lists pre8 not pre9?

[cartrite]

Can't be. CVS does not display level 11 or 12 textures. 1.4-pre6,7,9 does.

cartrite

[dirkpitt]

Can someone else (Chris?) confirm the VT issue (alas, VT's render as white on my Rage 128 machine at home)? Presumably it's a known issue (or possibly a non-issue?) if the prerelease builds are somehow mysteriously working around it...

[jdou]

I confirm,
- Stars style Points = Ctrl+V crash
- Levels 11-12 are ok on Windows pre9

[dirkpitt]

I confirm,
- Levels 11-12 are ok on Windows pre9

No, what I meant was, are Levels 11-12 not working in CVS builds? (and why?)

[t00fri]

Please note that I uploaded here

http://www.celestiaproject.net/forum/viewtopic ... 8&start=36

a modified 1.4.0pre9-F WINDOWS executable containing some fixes for further testing. It also does not show the CTRL-V crash for point style=points.

Bye Fridger

[fsgregs]

Selden:

I have never had a shader log actually write anything into it, noto nly with pre9 but with all the previous versions. What is supposed to happen when you open Celestia? When/how/what is supposed to be in the shaders.log file inside the Celestia main folder?

Frank

[brokfn]

Hi,

With 1.4.0pre9-F :
- no more crash by cycling through ctrl-V
- galaxy labels are a lot better. To my mind they should display a little earlier yet.
- the bug display with Radeon 9600 + latest drivers is still there (i.e. white oceans, problem with specular textures.
- strange behaviour when clicking on set time boxes. Not really a bug, but disturbing...

Bye
Brokfn

[BlindedByTheLight]

- galaxy labels are a lot better. To my mind they should display a little earlier yet.

I had mentioned it before... but this seems apropos again... perhaps a either a slider or a user-controlable property in the cfg file (if that's easier)?

Just a thought - one I'm sure the relevant parties have already had. Still, I like to share.

[t00fri]

Hi,

With 1.4.0pre9-F :
- no more crash by cycling through ctrl-V
- galaxy labels are a lot better. To my mind they should display a little earlier yet.
- the bug display with Radeon 9600 + latest drivers is still there (i.e. white oceans, problem with specular textures.
- strange behaviour when clicking on set time boxes. Not really a bug, but disturbing...

Bye
Brokfn

Thanks for the feedback! Did you note that the threshold for the label appearance may be set by you via the "[" and "]" keys. You may use a cel function in the start.cel script to restore the preferred settings upon startup.

In the time dialog, I did not touch anything.

Bye Fridger

[chris]

I uploaded 1.4.0pre10 for Windows:

http://www.celestiaproject.net/~claurel/celest ... 1.4.0pre10

It contains a fix for the crash when switching render paths and a workaround for the OpenGL 2.0 problems on new NVIDIA drivers. The latter fix may also help with ATI cards. I'd encourage anyone with and ATI card who wants to use OpenGL 2.0 to try out the new Celestia prerelease.

--Chris

[BlindedByTheLight]

Thanks for the feedback! Did you note that the threshold for the label appearance may be set by you via the "[" and "]" keys. You may use a cel function in the start.cel script to restore the preferred settings upon startup.

I did not know that! Thanks. Two things though. One, the [ and ] keys not only seem to toggle the labels... but also brightness. Is that supposed to happen?

And, secondly, even on the highest "show the most possible labels" I would say it still doesn't show nearly enough. In other words, on the the lowest setting I get usually one 1 galaxy (the one in the center - and only if it's close enough... most of the time I can't see any galaxy labels). On the highest setting, it's not much better. Perhaps the range should be tweaked? Or is this specific to my configuration somehow?

[t00fri]

Thanks for the feedback! Did you note that the threshold for the label appearance may be set by you via the "[" and "]" keys. You may use a cel function in the start.cel script to restore the preferred settings upon startup.

I did not know that! Thanks. Two things though. One, the [ and ] keys not only seem to toggle the labels... but also brightness. Is that supposed to happen?

Yes, indeed. The "[" and "]" set the "faintestMag" parameter. Increasing it, lets you see more galaxies, makes the visible ones more bright and displays more labels. By means of the "(" and ")" keys, instead, you may adjust galaxy brightness without affecting stars and label display.

And, secondly, even on the highest "show the most possible labels" I would say it still doesn't show nearly enough. In other words, on the the lowest setting I get usually one 1 galaxy (the one in the center - and only if it's close enough... most of the time I can't see any galaxy labels). On the highest setting, it's not much better. Perhaps the range should be tweaked? Or is this specific to my configuration somehow?

But not showing all labels at a time is the whole point of the code! Only those labels corresponding to relatively close-by apparent galaxy brightness are shown beyond a certain threshold. The deeper you zoom in, the more labels pop up. But if too many would pop up, you'd be back at label overlap again. The best tool is actually to switch on the marker (CTRL+K) that shows you the actually selected galaxy location before you see it and it's label.

Upon startup, only the labels are shown that correspond to DSO's visible with the current faintestMag setting, typically 7.0mag, i.e ~ naked eye visibility (in empty space).

You should just see the labels of M 31, M33, Milky Way, SMC, LMC, ....

Bye Fridger

[BlindedByTheLight]

But not showing all labels at a time is the whole point of the code!

Well, lol... that would explain why I'm the only one bitching about it!

I can see from a scientific perspective (which I understand is the most important consideration for the coders) why the feature would behave as such.

I will just miss my Celestia demonstrations that have become quite popular in my circle. Missing being able to zoom out from the Milky Way - after impressing upon my fawning masses just how big our galaxy is with a quick tour.... then zooming out even further and turning on labels and having them all go "Wow... look at all those stars."

To which I impishly add: "Those aren't stars...."

Blows 'em away EVERY time....