Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
On Dienstag 04 Oktober 2005 22:17, Melchior FRANZ wrote: > * Curtis L. Olson -- Tuesday 04 October 2005 22:02: > > You've been granted CVS commit access so use your best judgement. > > Yes. I don't usually touch such things, because I don't understand much > of this. I did it anyway, because: > > - this change was already in cvs since a great while, and only had been > reverted recently > > - the commit log of the reverting patch didn't explain why this was > reverted; it was part of a completely different change and looked like an > accident Well, I reverted. Just because, as it was introduced the first time it was a workaround for something, at this time, hard to fix. At that time, the renderer had a different understanding of ground level than the gear code. I changed that at some time and removed the workaround. I thought that it was clear that it was a workaround, and I silently restored the old, more correct, behavour. Greetings Mathias -- Mathias Fröhlich, email: [EMAIL PROTECTED] ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
Curt, Is on my todo list for tomorrow (friday) since I saw Melchior's patch. Greetings Mathias On Dienstag 04 Oktober 2005 20:52, Curtis L. Olson wrote: > For what it's worth, I don't like this patch. It shouldn't make much > difference on 24/32 bit cards, which is probably most everyone now > anyway, but I think there is a different problem brewing somewhere. > > I haven't had time to look into it, but the AGL reading on the HUD no > longer reads correctly. Somewhere along the lines we have introduced > some sort of height above ground bugs. I don't know if that is in the > ground cache code or elsewhere, but the HUD above ground display isn't > working right anymore. > > If we get that problem fixed so the system knows the correct AGL, then > we wouldn't need to make this particular huge hack 5 times worse. > > Somehow the gear still knows where the ground is, but I recall specific > patches to the individual FDM's. I've lost track of what is going on > with this section of code, but it's important and it really should get > fixed before we get too much further! > > I'm going out of town on thursday and rushing to get a bunch of other > stuff done in the mean time, so I really can't look at this in the near > term, but someone really needs to volunteer to step up and track down > what is going on here. > > Regards, > > Curt. > > Melchior Franz wrote: > >Update of /var/cvs/FlightGear-0.9/FlightGear/src/Main > >In directory baron:/tmp/cvs-serv754 > > > >Modified Files: > > renderer.cxx > >Log Message: > >prevent view through big hole in carrier deck > > > > > >Index: renderer.cxx > >=== > >RCS file: /var/cvs/FlightGear-0.9/FlightGear/src/Main/renderer.cxx,v > >retrieving revision 1.27 > >retrieving revision 1.28 > >diff -C2 -r1.27 -r1.28 > >*** renderer.cxx 1 Oct 2005 09:56:53 - 1.27 > >--- renderer.cxx 4 Oct 2005 18:01:45 - 1.28 > >*** > >*** 499,503 > > - cur_fdm_state->get_Runway_altitude_m(); > > > >! if ( agl > 10.0 ) { > > scene_nearplane = 10.0f; > > scene_farplane = 12.0f; > >--- 499,503 > > - cur_fdm_state->get_Runway_altitude_m(); > > > >! if ( agl > 50.0 ) { > > scene_nearplane = 10.0f; > > scene_farplane = 12.0f; > > > > > >___ > >Flightgear-cvslogs mailing list > >[EMAIL PROTECTED] > >http://mail.flightgear.org/mailman/listinfo/flightgear-cvslogs > >2f585eeea02e2c79d7b1d8c4963bae2d -- Mathias Fröhlich, email: [EMAIL PROTECTED] ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
Frederic Bouvier wrote: So the HUD is displaying the height for the last known QFE ? I think so. I suppose it just a barometric instrument with a digital display. It is synchronized by ATC reports. Erik ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
Quoting Erik Hofman: > Dave Culp wrote: > > > This sounds more like HAA (height above airport) or HAT (height above > > touchdown). Height AGL should be the current height above the ground > > directly below the aircraft. > > > > Height AGL should change as the terrain below the aircraft changes. > > What would expect the HUD to display? I'm quite sure that the F-16 > doesn't have a terrain database or an AGL radar. So the HUD is displaying the height for the last known QFE ? -Fred ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
Dave Culp wrote: This sounds more like HAA (height above airport) or HAT (height above touchdown). Height AGL should be the current height above the ground directly below the aircraft. Height AGL should change as the terrain below the aircraft changes. What would expect the HUD to display? I'm quite sure that the F-16 doesn't have a terrain database or an AGL radar. Erik ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
> So what basically happens now is that at the (startup) airport the AGL > would be reported correctly, but once the terrain elevation increases > the reported AGL won't change (like in real life). This sounds more like HAA (height above airport) or HAT (height above touchdown). Height AGL should be the current height above the ground directly below the aircraft. Height AGL should change as the terrain below the aircraft changes. Dave ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
Melchior FRANZ wrote: * Curtis L. Olson -- Tuesday 04 October 2005 22:22: Somewhere since the last release, that got broke and it must get fixed. If that was fixed you wouldn't be seeing a hole in the carrier deck. The bug was AFAIK there ever since we have helicopters. The same holes were on rooftops. Looking at the code (and only at the code) it looks more like a misunderstanding than a bug. What happens with the HUD is that it behaves like a normal instrument now (and not a perfect one) by that it specifies the AGL relative to the last known good elevation (the airport elevation). I assume it worked more like a radar that could precisely determine the AGL at the aircraft location. So what basically happens now is that at the (startup) airport the AGL would be reported correctly, but once the terrain elevation increases the reported AGL won't change (like in real life). Maybe we need a different naming for exact AGL (which is computed correctly BTW, but under a different name). Erik ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
[Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
* Curtis L. Olson -- Tuesday 04 October 2005 22:22: > Somewhere since the last release, that got broke and it must get fixed. > If that was fixed you wouldn't be seeing a hole in the carrier deck. The bug was AFAIK there ever since we have helicopters. The same holes were on rooftops. m. ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
Melchior FRANZ wrote: * Curtis L. Olson -- Tuesday 04 October 2005 22:02: You've been granted CVS commit access so use your best judgement. Yes. I don't usually touch such things, because I don't understand much of this. I did it anyway, because: - this change was already in cvs since a great while, and only had been reverted recently - the commit log of the reverting patch didn't explain why this was reverted; it was part of a completely different change and looked like an accident - I mentioned it in this message and got no reactions: http://mail.flightgear.org/pipermail/flightgear-devel/2005-October/039285.html not that this is necessarily an agreement, but together with the other two reasons I though it would be OK, and better than the whole, which I consider a show-stopper. I'd just hate to have this slip through the cracks, and when someone tries to land on an object that is 50.01 meters tall or more, they are going to get a hole again. We could just remove that check and leave the near clip plane in close all the time, but then our terrain rendering will really stink for anyone with a 16bit depth buffer ... Andy (via IRC) has also looked at the code and suggested that the whole 'if' case is probably not needed any more. I just tested it, and indeed, with only scene_nearplane = groundlevel_nearplane->getDoubleValue(); scene_farplane = 12.0f; the hole doesn't occur any more. I'll be doing some more tests. But I won't touch that code again without explicit OK from an expert. :-) Just know that with the near plane set close in, there isn't enough depth buffer resolution on 16 bit cards to properly draw the terrain. If you look at mountains in the distance, you get lots of odd z-buffer fighting. This is on 16 bit cards. If we don't care about 16 bit cards any more (that used to be our only option in the old voodoo-1/2/3 days) then we could remove that whole if statement. For what it's worth, my laptop can only run FlightGear acceptably in 16 bit mode so I'm slightly worried about the ramifications of this change. Ultimately we *really* need to fix the above ground level calculations. Somewhere since the last release, that got broke and it must get fixed. If that was fixed you wouldn't be seeing a hole in the carrier deck. (And the AGL computations in the rest of the sim would start working correctly again.) Regards, Curt. -- Curtis Olsonhttp://www.flightgear.org/~curt HumanFIRST Program http://www.humanfirst.umn.edu/ FlightGear Project http://www.flightgear.org Unique text:2f585eeea02e2c79d7b1d8c4963bae2d ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
[Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
* Curtis L. Olson -- Tuesday 04 October 2005 22:02: > You've been granted CVS commit access so use your best judgement. Yes. I don't usually touch such things, because I don't understand much of this. I did it anyway, because: - this change was already in cvs since a great while, and only had been reverted recently - the commit log of the reverting patch didn't explain why this was reverted; it was part of a completely different change and looked like an accident - I mentioned it in this message and got no reactions: http://mail.flightgear.org/pipermail/flightgear-devel/2005-October/039285.html not that this is necessarily an agreement, but together with the other two reasons I though it would be OK, and better than the whole, which I consider a show-stopper. > I'd just hate to have this slip through the cracks, and when > someone tries to land on an object that is 50.01 meters tall or more, > they are going to get a hole again. We could just remove that check and > leave the near clip plane in close all the time, but then our terrain > rendering will really stink for anyone with a 16bit depth buffer ... Andy (via IRC) has also looked at the code and suggested that the whole 'if' case is probably not needed any more. I just tested it, and indeed, with only scene_nearplane = groundlevel_nearplane->getDoubleValue(); scene_farplane = 12.0f; the hole doesn't occur any more. I'll be doing some more tests. But I won't touch that code again without explicit OK from an expert. :-) m. ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
Melchior FRANZ wrote: * Curtis L. Olson -- Tuesday 04 October 2005 20:52: For what it's worth, I don't like this patch. I find the hole more annoying. Unfortunately, I can't fix what you think is the real problem. Shall I revert for now? I'm not saying the hole isn't annoying, I'm just saying that there is a bug because for some reason, the sim thinks you are > 10 meters AGL when you are sitting on the carrier deck. There is some ground intersection problem going on there. If the ground interesection was computed correctly, the system would think you are < 10 meters AGL and everything would work the way it is intended. I'd really like for this to get fixed the right way. When we slap on bandaids without fixing the underlying problems, we end up with a system that has a lot of bandaids on top of a rotting infrastructure. Similarly whenever we see a stray crash or segfault we should pursue it with our utmost agression and stamp those out right away. Anytime we leave these sorts of crashes and problems for later, we end up with a system full of unexpected, unexplained, impossible to debug crashes. That kind of software is an incredible pain to operate. In the past I had more time to defend against these things, right now I don't. You've been granted CVS commit access so use your best judgement. I'd just hate to have this slip through the cracks, and when someone tries to land on an object that is 50.01 meters tall or more, they are going to get a hole again. We could just remove that check and leave the near clip plane in close all the time, but then our terrain rendering will really stink for anyone with a 16bit depth buffer ... It's not an easy problem, but slapping a bandaid ontop will probably mask it long enough so that the person who introduced the orignal problem will be long gone before we get bit again and no one will know how to fix it ... Regards, Curt. -- Curtis Olsonhttp://www.flightgear.org/~curt HumanFIRST Program http://www.humanfirst.umn.edu/ FlightGear Project http://www.flightgear.org Unique text:2f585eeea02e2c79d7b1d8c4963bae2d ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
[Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
* Curtis L. Olson -- Tuesday 04 October 2005 20:52: > For what it's worth, I don't like this patch. I find the hole more annoying. Unfortunately, I can't fix what you think is the real problem. Shall I revert for now? m. ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
[Flightgear-devel] Re: [Flightgear-cvslogs] CVS: FlightGear/src/Main renderer.cxx, 1.27, 1.28
For what it's worth, I don't like this patch. It shouldn't make much difference on 24/32 bit cards, which is probably most everyone now anyway, but I think there is a different problem brewing somewhere. I haven't had time to look into it, but the AGL reading on the HUD no longer reads correctly. Somewhere along the lines we have introduced some sort of height above ground bugs. I don't know if that is in the ground cache code or elsewhere, but the HUD above ground display isn't working right anymore. If we get that problem fixed so the system knows the correct AGL, then we wouldn't need to make this particular huge hack 5 times worse. Somehow the gear still knows where the ground is, but I recall specific patches to the individual FDM's. I've lost track of what is going on with this section of code, but it's important and it really should get fixed before we get too much further! I'm going out of town on thursday and rushing to get a bunch of other stuff done in the mean time, so I really can't look at this in the near term, but someone really needs to volunteer to step up and track down what is going on here. Regards, Curt. Melchior Franz wrote: Update of /var/cvs/FlightGear-0.9/FlightGear/src/Main In directory baron:/tmp/cvs-serv754 Modified Files: renderer.cxx Log Message: prevent view through big hole in carrier deck Index: renderer.cxx === RCS file: /var/cvs/FlightGear-0.9/FlightGear/src/Main/renderer.cxx,v retrieving revision 1.27 retrieving revision 1.28 diff -C2 -r1.27 -r1.28 *** renderer.cxx1 Oct 2005 09:56:53 - 1.27 --- renderer.cxx4 Oct 2005 18:01:45 - 1.28 *** *** 499,503 - cur_fdm_state->get_Runway_altitude_m(); ! if ( agl > 10.0 ) { scene_nearplane = 10.0f; scene_farplane = 12.0f; --- 499,503 - cur_fdm_state->get_Runway_altitude_m(); ! if ( agl > 50.0 ) { scene_nearplane = 10.0f; scene_farplane = 12.0f; ___ Flightgear-cvslogs mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-cvslogs 2f585eeea02e2c79d7b1d8c4963bae2d -- Curtis Olsonhttp://www.flightgear.org/~curt HumanFIRST Program http://www.humanfirst.umn.edu/ FlightGear Project http://www.flightgear.org Unique text:2f585eeea02e2c79d7b1d8c4963bae2d ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d