[Flightgear-devel] Issue collection

[Renk T. <tho...@jy...>]

> I've created a merge request
> https://www.gitorious.org/fg/fgdata/merge_requests/183
> containing lots of my recent shader work(...)

Um... what should I do with this? I've gotten one comment pointing to a problem with the urban shader, but I can reproduce the same problem on a pristine master branch, so it's not related to the merge request. 

I've tested the things I usually do a bit more - it seems to be working more or less fine. I suppose in principle I could push this myself now, but since it's effects file it can potentially damage any of our rendering schemes, so I'd really appreaciate a second pair of eyes. I'm also still mired in hardware problems, my current FG setup is spred out across 3 setups on 2 computers between Windows and Linux, so I can't really do so much stuff right now.

> * Restarting Advanced Weather seems to have a timing issue for really  
> high  framerates. 

Is there anyone else observing this issue? The symptom is that after selection of a weather tile and 'Okay', no clouds get drawn but the system throws an error to the comsole complaining about NaN passed to setprop() or an array out of bounds. This also happens when restarting weather tiles (pressing Okay again). If it is a timing issue, it needs to be tested on different systems, because I've never ever seen it before on my old computer.


> * strange bug - all generated clouds (on Linux and Wndows) used only one  
> texture of a multi-texture sheet, which gives some odd repetitions.

Is anyone else seeing this one? I'm seeing it on all my three setups, in both atmospheric light scattering and default rendering scheme (so it isn't related to any shader I touched), it goes away when I roll the binary back three weeks.

It's a bit subtle for many clouds (which tells how nice the algorithms are) but for whispy clouds like Altocumulus or small Cumuli it's really apparent that they all look the same. This doesn't appear something I can fix - was there any change to the C++ side in the last weeks which could have caused this? It's a pretty serious bug in my opinion. It might also affect trees, as they rely on the same multi texture coordinate strategy, I haven't looked if all trees now look the same.

> I noticed that after starting up fgfs with metar activated, the "normal
> weather comes up".
> Now if the "normal" weather shows rain and i then open the weather menu  
> and
> switch the advanced weather on with "OK" the weather is changing, but if
> advanced weather shows now rain the the wet runways stay wet.

I *think* the runway shader of the default scheme probes the rain-norm property to determine if the runway is to be drawn wet. Basic Weather sets that by METAR. Advanced Weather sets it only if you're inside a patch of rain and hence Atmospheric Light Scattering uses /environment/surface/wetness to determine if terrain appears wet or not, because terrain can be wet if it rained an hour ago, but currently the sun is shining.

The upshot is that Advanced Weather can decide that even if the METAR says rain, you may not currently be inside a patch of rain and thus sets the rain norm to zero. In this case the runway would suddenly apper dry when Advanced Weather starts, although the METAR is the same.

> If i press the apply button instead the runways change from wet to dry
> instantly. If i press OK after that all clouds get redrawn again.

Advanced Weather doesn't uniquely map the METAR into a cloud/weather configuration - it creates a wide range of configurations which are all compatible with the METAR (4/8 at 3000 ft doesn't tell you which cloud type this is, how large the patches are, where they are,...). Every time you press okay, a different configuration is created as an interpretation of the currently selected METAR string. Some of them might give you a dry runway (see above), some might not.

> How is this menu supposed to be operated :)

We thought it'd be more intuitive than the last one... How would you expect it to operate?

> This is now done.  You can now set
> /environment/params/use-external-precipitation-level=true to disable the
> previous behaviour of only showing rain under the heaviest cloud layer,  
> and
> set /environment/params/external-precipitation-level-m to set the maximum
> altitude at which precipitation will occur.  0 means it will occur at all
> levels.

Very cool - thanks.

* Thorsten