Wobbly AP PE and Nodes

[Daveroski]

Easy to see when plotting a course to Mimnus from a Kerbin orbit of around 100k.

First the AP and PE can be seen to be wobbling on the orbit around Kerbin. Particularly if the orbit is very close to circular. and the value of AP is Close to the value of PE.

It's as if there are two values for each and the blue orbit line quickly but noticeably changes from one to the other.

At first I thought it might be mods causing issues but this is a vanilla installation.

Next Set mimnus as target.

At AN or DN equalise the angle of the orbit. You have to do this by eye as no accurate information about the angles of the orbits is available to the user.

Once equalised, plot a course for Mimnus.

Multiple times I have noticed this effect.

A course can be plotted which almost intersects Mimnus - a PE of less than 10K

Switch to Standard camera view and line up with your new node. Use Time-warp as necessary but carefully and patiently. If you go too fast you may overshoot and you can't send time-warp in reverse. (Wouldn't THAT be cool!?)

When it is time to fire up the engine using the Half the duration as a calculation of when to fire the engine eg. a duration of 1m 20s half of which is 40s means fire the engine at full power at T-40s. The instant you notice your little blue target node fly out from under the cross-hair cut the engine. Reaquire the little blue node and gently fire towards it using shift and control for power until the little meter at the side of the targeting globe reads less than 0.2.

Switch to Map view.

At this point I can see that I have plotted my course and all seems well.

I delete the node I created.

It is at this point things get creaky

It shows me the course I plotted and the PE at Mimnus but it flashes between the course I plotted and a clear miss where I will get close to mimnus, but no cigar. The course is just a long elliptical orbit around Kerbin.

Both of these courses are on my map at the same time. Quickly flashing from one to the other. There is no way to tell which is real. I deleted my node so it's not that.

If I time warp at this stage, even briefly, My plotted course disappears and the elliptical orbit is all that remains.

Now I have to make course corrections just to get into Mimnus' SOI again.

It's almost as if there are two universes, one slightly out of sync with the other and the craft is switching between the two.

It may be a secondary cause but it would explain the wobble I get now in larger craft. It's as if it's trying to orient for two slightly different orbits. But as this could be a different issue I will report it as such.

[Bilfr3d]

This is a known bug. It is an issue with the floating-point variables (or something like that) which is used in the custom calculations of the orbits. Thanks for reporting it, however I believe it is known and has been reported several times in the bug tracker. Not too sure about the bug tracker bit. It won't ACTUALLY affect your orbits though. So, no need to worry about losing your ship to a wobbly PE or AP - you'll only see it change when it is loaded into the calcs as well. As for your orbit being different than calculated, that's usually a... erm... pilot problem. Most times, you just need to start burning around 10-5s before the actual burn node. Also ALWAYS follow the blue thing. xD

Hope this helps,

ToTheMun

[Daveroski]

This is a known bug. It is an issue with the floating-point variables (or something like that) which is used in the custom calculations of the orbits. Thanks for reporting it, however I believe it is known and has been reported several times in the bug tracker. Not too sure about the bug tracker bit. It won't ACTUALLY affect your orbits though. So, no need to worry about losing your ship to a wobbly PE or AP - you'll only see it change when it is loaded into the calcs as well. As for your orbit being different than calculated, that's usually a... erm... pilot problem. Most times, you just need to start burning around 10-5s before the actual burn node. Also ALWAYS follow the blue thing. xD

Hope this helps,

ToTheMun

Wow that was rude.

Pilot problem? Always follow the blue thing?

If you bothered to actually read my post. I said that the calculation was correct. My plotted course was clearly shown.

As for your 10-5 seconds remark. Its simple mathematics. We have to provide X amount of power at 'T' for the node to be precise. We can't just instantly deliver that much power. so we spread the power evenly on either side of 'T' If the node wasn't constantly being estimated my way would still work. You would be somewhere near Duna.

Why do people always assume noob and that they know better than the OP?

Why did you thank me for reporting it? Are you a developer? The games producer perhaps? Perhaps you run the forum and were thanking me for using it the way it was intended? Or perhaps you were just being patronising.

I am correct.

There is a bug.

I reported it.

[sal_vager]

Sorry Daveroski but it really is a floating point issue with the display, even if you're spot on with the burn KSP can display a different orbit to what was predicted.

There's already been a few attempts by the devs to address this but it's not been completely fixed yet, and may never be, not without deliberately fudging the orbits to match what was predicted anyway.

Also, there's no need to address ToTheMun in that way, he was trying to help.

[Daveroski]

Sorry Daveroski but it really is a floating point issue with the display, even if you're spot on with the burn KSP can display a different orbit to what was predicted.

I think that information may be in error. I think it is an error in the base coding of the physics of the universe and has nothing to do with how it is displayed. If it was a display error, the craft would not try to follow two different alternating values and wobble in the way that it does. It would affect the display. Not the craft.

I think the display merely presents that universe according to available data. If a fundamental part of that data is constantly changing and not being corrected properly it would explain two possible courses flashing alternately on the display. I feel that the root source of the variation should be found and corrected and not patched after the fact as I believe is what is currently happening now.

While it is possible that I may be in error. I think that all data is relevant and that my observations may help to rectify the situation. The displayed result and the resulting course seem to be getting data from two separate sources. I feel it is relevant and so point it out.

There's already been a few attempts by the devs to address this but it's not been completely fixed yet, and may never be, not without deliberately fudging the orbits to match what was predicted anyway.

Something needs to be done about it because it makes the game feel broken. It feels broken to me. Therefore it must feel broken to some others. The reason I pointed it out.

Also, there's no need to address ToTheMun in that way, he was trying to help.

Being rude helps? Insulting my intelligence helps? Patronising me helps?

You may feel that there was no need but if ToTheMun isn't informed that they are being insulting they may feel that it is acceptable to talk down to others and continue to do so and so, there was a need to tell him that his remarks were offensive.

I'm not about to show off and list my abilities and my skills but have worked hard to get them and so found his remarks offensive.

I responded fairly and appropriately to him and no longer feel the need to pursue it.

[Starwaster]

Being rude helps? Insulting my intelligence helps? Patronising me helps?

You may feel that there was no need but if ToTheMun isn't informed that they are being insulting they may feel that it is acceptable to talk down to others and continue to do so and so, there was a need to tell him that his remarks were offensive.

I'm not about to show off and list my abilities and my skills but have worked hard to get them and so found his remarks offensive.

I responded fairly and appropriately to him and no longer feel the need to pursue it.

Ok, I read over his message and nope, not seeing anything particularly rude there. He has no idea what your skill level and calling into question the possibility of user error is not uncalled for.

You're overreacting.

[vexx32]

Guys, this is a support forum, not an IRC chatroom. If you're not contributing to the thread, there's no need to post -- and personal attacks will not be tolerated. If more information is requested, it is only courteous to provide it. More data is always useful in solving a problem.

Daveroski, we have to cover all our bases, as it would hardly be fun for anyone (yourself included) if we overlook a potentially useful piece of information that could get your issue resolved almost straight away; spending a week on a one-hour problem is annoying for everyone involved.

As the others have said, this is somewhat due to the floating-point errors that appear in a great many applications. In macroscopic distances like those involved in KSP, these errors compound immensely. It is potentially possible to mitigate this by forcefully making the Ap/Pe markers stay in place if their altitude is unchanging, but that has not yet been done, nor do I know for certain how complex such a function would be; it depends how the code is currently set up. The wobble exists (in its most common incarnation) because in cases where the orbit is near circular, large regions of the orbit tend to end up being at about the same altitude. Combined with the errors inherent in floating-point variables (especially with larger numbers), it becomes tricky to get a definite fix on exactly where a Periapsis or Apoapsis is.

[Daveroski]

Guys, this is a support forum, not an IRC chatroom. If you're not contributing to the thread, there's no need to post -- and personal attacks will not be tolerated. If more information is requested, it is only courteous to provide it. More data is always useful in solving a problem.

Daveroski, we have to cover all our bases, as it would hardly be fun for anyone (yourself included) if we overlook a potentially useful piece of information that could get your issue resolved almost straight away; spending a week on a one-hour problem is annoying for everyone involved.

As the others have said, this is somewhat due to the floating-point errors that appear in a great many applications. In macroscopic distances like those involved in KSP, these errors compound immensely. It is potentially possible to mitigate this by forcefully making the Ap/Pe markers stay in place if their altitude is unchanging, but that has not yet been done, nor do I know for certain how complex such a function would be; it depends how the code is currently set up. The wobble exists (in its most common incarnation) because in cases where the orbit is near circular, large regions of the orbit tend to end up being at about the same altitude. Combined with the errors inherent in floating-point variables (especially with larger numbers), it becomes tricky to get a definite fix on exactly where a Periapsis or Apoapsis is.

I don't have to tell you Newton's first law.The problem seems to be at the origin. Call it (0,0,0) A you get further from the origin the error seems to increase. I have had several Duna space stations, shake themselves to pieces.

Now I build stations in as few parts as possible. No science modules. No habitats. No pretty bits. They have to be purely functional.

I delete debris regularly so the poor program doesn't have to track so many parts and get confused. It does help most of the time.

If I am not using a station or a ship in space, I will retire it. (rescuing any kerbals first of course)

Keeping the number of space objects and landed objects on other worlds to a minimum seems to reduce the wobble al little.

But it upsets me because I want to build bigger and better, I want to take huge things vast distances. But I don't any more, because I have lost so many. When it is fixed, I may spend another 3000 hours playing it.

[vexx32]

KSP uses a floating origin system. Basically, the entire universe uses the currently focused spacecraft as the origin point. You'll almost never see any issues caused by being 'too far' from the origin anymore; this was implemented some time ago, around the time of the original Space Kraken.

But yes, large craft can tend to have problems -- heck, even building some ludicrously large craft has been known to break the editor scene completely (a forum user by the name of Whackjob is quite familiar with this). The glitches are hard to pin down precisely, but Squad are doing as much as they can.

[Bilfr3d]

Why did you thank me for reporting it? Are you a developer? The games producer perhaps? Perhaps you run the forum and were thanking me for using it the way it was intended?

I think that information may be in error. I think it is an error in the base coding of the physics of the universe and has nothing to do with how it is displayed. If it was a display error, the craft would not try to follow two different alternating values and wobble in the way that it does. It would affect the display. Not the craft.

I think the display merely presents that universe according to available data. If a fundamental part of that data is constantly changing and not being corrected properly it would explain two possible courses flashing alternately on the display. I feel that the root source of the variation should be found and corrected and not patched after the fact as I believe is what is currently happening now.

While it is possible that I may be in error. I think that all data is relevant and that my observations may help to rectify the situation. The displayed result and the resulting course seem to be getting data from two separate sources. I feel it is relevant and so point it out.

Well, apparently you are a game producer of KSP. KSP is not 'real life'... don't expect it to act like real life and take in all the relevant laws of physics. It's also an ALPHA. It's still in development, everything is not yet fully complete. The KSP devs may yet re-haul the entire orbital physics system. Who knows?

I'm not about to show off and list my abilities and my skills but have worked hard to get them and so found his remarks offensive.

I responded fairly and appropriately to him and no longer feel the need to pursue it.

If you had skills, you would have realised that there are three stickied threads to read before posting a request for support. That includes the thread on providing information like logs, screenshots etc.

So far it seems that with every support/bug request you have made, all attempts to fix/diagnose the bugs by both myself AND staff and moderators has resulted in you replying by saying you know better than people who are STAFF AND MODS and also other users, whom you do not know. If you don't like the responses you will receive from support requests if you have not completed the request properly, then request for support properly or don't request for support at all.

And the further you are from the origin has no affect on your ships shaking themselves apart. Same goes with the whole number of ships and debris. It's just the kraken - it is random. The main reason for the kraken is part clipping, and SAS. Its worse when SAS is on. Sometimes, the kraken will not seem to exist - other times it will be a horrid explosion on months of flying and launching. This is a completely separate bug to the one you are reporting here, and I believe the KSP devs are looking for a solution to fix it.

[Vanamonde]

Alright, Daveroski. You feel you know the reason for the glitch, others disagree, but you wanted to report it, and now you have. Let's all move on, please, and put an end to this unpleasantness. Thread closed.