| Author |
 Topic  |
|
MPusz
Aquila imperiale
Poland
683 Posts |
 Posted - 31/01/2010 : 18:49:23
|
| Can someone move that thread to bugs section? Thanks |
 |
|
|
Coolwind
Moderator
Italy
7511 Posts |
Posted - 31/01/2010 : 19:07:27
|
| Only after checking xcsoar: i need to know if it is "normal" for xcsoar or not. I need to know if I am going to fix a bug of someone else or mine!! |
|
|
|
MPusz
Aquila imperiale
Poland
683 Posts |
Posted - 31/01/2010 : 19:20:07
|
| If only I could I would be glad to check it. However it seems that Blue Arrow on XCSoar have way more bugs than on LK8000 (you mentioned that you have fixed something already). I set up a waypoint 140km from the glider and even 60km/h side wind didn't make the blue arrow point in other direction than exactly to the destination (it is different in current LK8000 versions). |
|
|
|
MPusz
Aquila imperiale
Poland
683 Posts |
Posted - 03/02/2010 : 09:03:18
|
| As stated above that problem cannot be reproduced at XCSoar (because it is hidden under other bugs you fixed already in LK8000), but I am pretty sure it is "old" XCSoar bug. Could you please move that thread to the bugs section so it doesn't get lost? |
|
|
|
eric.carden
Aquilotto
USA
206 Posts |
Posted - 09/03/2012 : 22:23:54
|
Is this issue still open/unresolved? If so, I may dig around in the code a little and see what I find. I agree that the behavior you report, MPusz, is undesirable. It sounds like the math used doesn't take into account (other than to hide the arrow when above final glide) how much more time you'll spend climbing/drifting. It sounds likely that the code assumes you want to arrive at the next waypoint at your current altitude, even if doing so would mean rounding the next waypoint way above final glide. That's my first-thought theory, anyway.
Eric |
|
|
|
eric.carden
Aquilotto
USA
206 Posts |
Posted - 19/03/2012 : 20:28:24
|
| I confirmed (LK v3.0a + Condor) that this issue still exists. I dug around in the code, and it's as I suspected. The code assumes you want to arrive at goal at your current altitude. I've scribbled out a solution, but before I go any further... Paolo, may I fix this? |
|
|
|
Coolwind
Moderator
Italy
7511 Posts |
Posted - 19/03/2012 : 21:06:56
|
Sure you can fix it, but I confess I dont see the problem, at least using a single goto.
Can someone else among developers please check this out? I dont want to fix one problem, and open another at the same time of course! |
|
|
|
MPusz
Aquila imperiale
Poland
683 Posts |
Posted - 19/03/2012 : 21:12:38
|
| Blue Arrow never worked like it should. It seems that Eric found the right purpose of that problem. |
|
|
|
eric.carden
Aquilotto
USA
206 Posts |
Posted - 19/03/2012 : 21:24:35
|
Here's the basic problem. Assume you lack only 500' to begin final glide. During the glide (if begun here), you'd lose 5,000'. LK SHOULD only guide you far enough upwind of the course line to account for the drift you'd experience while climbing 500'. Instead, it guides you far enough upwind to allow 10X that amount of climbing/drifting. This can cause a considerable unnecessary "detour", especially in high winds in a slow, low-performance glider with slow climbs. The problem exists as much for a single goto as for a complex course.
I totally agree about not wanting to break one thing while fixing another, of course! :-)
So how precise do we need "best cruise track" (the direction pointed by the blue arrow) to be? An iterative approximation is needed to fix this, and I don't want to waste any processing power getting more precision than is needed. I'll start the bidding at +/-1 degree. How about it? |
|
|
|
Coolwind
Moderator
Italy
7511 Posts |
Posted - 19/03/2012 : 21:32:45
|
If we are talking about displaying the blue arrow, 5 degrees are far enough accurate.
It is supposed to give only an hint , visually. I am not keen on overloading calculations for an approximated direction used only as a visual reference on a small screen.
Are you sure we are not already calculating this drift, for final glides?
I ask because in these long months, I have practically changed everything in the code except the calculations (most of them, and task included). Reason: not doing competitions, I made LK best for cross country. So I am not sure at all we dont already calculate this.
By the way I have just started another big code cleanup, and this time I should take care of calculations because some people want to start working on them for general aviation.
|
|
|
|
eric.carden
Aquilotto
USA
206 Posts |
Posted - 19/03/2012 : 21:48:19
|
| There's no need to circle on final glide, so no circling drift (the whole point of BestCruiseTrack and the blue arrow) is calculated for final glides. In fact, the blue arrow is currently designed to disappear once you're on final glide. And the code skips the calculation of BestCruiseTrack if on final glide. BestCruiseTrack is calculated whenever navigating a task and not on final glide, and this is where the error lies - and is where the fix needs to be applied. |
|
|
|
Coolwind
Moderator
Italy
7511 Posts |
Posted - 19/03/2012 : 21:51:34
|
| If I do a simple Goto I get the blue arrow! |
|
|
|
eric.carden
Aquilotto
USA
206 Posts |
Posted - 19/03/2012 : 22:00:25
|
| The problem isn't that the arrow doesn't appear when it should. It's that it points in the WRONG DIRECTION. Try a scenario like the one I described, and you'll see. |
|
|
|
eric.carden
Aquilotto
USA
206 Posts |
Posted - 21/03/2012 : 23:03:55
|
Paolo said... quote:
not doing competitions, I made LK best for cross country
Interesting... (I assume you mean "best for NON-RACING cross-country".) A racing pilot would definitely prefer the arrow to behave as MPusz and I think. When NOT racing, though, the way it is now is probably plenty acceptable - and maybe even slightly preferable. I guess the non-racing pilot will just need to remember that he needs to head a little more into the wind than the arrow points, unless he's okay with arriving at goal only at his safety altitude.
If any non-racers ever wish for it to be back like it is now, I suppose we could add a user-configurable option. I doubt, though, that anyone besides racing pilots will notice the change, and they should be happy with the change.
Eric |
|
|
|
Coolwind
Moderator
Italy
7511 Posts |
Posted - 22/03/2012 : 00:44:44
|
I think it is ok to fix it as you say, Eric, no need to configure it.
If it's wrong, it's wrong. It cannot be half wrong. |
|
|
|
Topic |
|
(LOW) XCSOAR Blue arrow problem