nyrath
-
Posts
604 -
Joined
-
Last visited
Content Type
Profiles
Forums
Developer Articles
KSP2 Release Notes
Posts posted by nyrath
-
-
I've finished editing my post now, hopefully it's more clear / useful. Didn't mean to overstep.
No, no, you are not overstepping. I welcome your input. You solved the original problem, remember?
Progress report:
Ohboy, I'm surprised any of these Orions ever got off the launch pad.
First off due to some sloppy factoring on my part, there were not one but two places that I recalculate the mass, and of course I forgot to change the second (for people who do not write computer code for a living the translation is BAD programmer! BAD, BAD! No dog biscuit for you!) In other words, the magazines soon had their full mass. I've refactored it like I should have in the first place. This mistake is done by programmers who are amateurs, sloppy, or lazy. Sloppy in this case.
Secondly, if you have a launch pad accident and you do a restart, KSP will re-use the updated engine mass and start re-adding the virtual magazine mass for a second time. I will check but I'm afraid this will happen if you, say, switch from one vessel to another and back again. I hope not.
And I will try to make a new bomb type in between 3.5MN and 80MN. Would 40MN be suitable?
-
In the above system as you stated it, the magazine's mass is being recalculated after every bomb. And the new mass is being added to the engine. (and the old mass deducted).
Since you're adding the mass of the magazine to the engine, please subtract it from the magazine
Better yet, calculate what the magazine's mass would be and add that to the engine, without ever changing the magazine's actual mass. (perhaps this is what you're actually doing)
Possibly better yet (though it may not allow for ejecting magazines or replacing them in flight or other strange behavior) but you could just Initialize by calculating the appropriate mass of all magazines and adding that to the engine, then when a bomb is expended simply deduct the mass of that bomb from the engine. Rather than recalculating magazine mass at all. (The magazines just always have their low constant mass).
Yes, I was already subtracting the mass of the magazine from the magazine, less 1 ton. I didn't want to get into tedious details. If anybody wants those, the source code is in the mod.
I might go with the system you propose (the third one, not recalculating, just deducting the mass from the engine), except it will get a tad more complicated when NovaSilisko starts jettisoning his Orion magazines. Or adding new full magazines when one refuels.
I did not use your proposed system initially since a career of being burnt by unexpected side effects make me habitually try to program the absolute minimum change possible. The more code your change, the more chances to spawn new bugs. Of course apparently that strategy did not work in this case.
-
The Engine with the column works very good for me, even with heavy payloads under strong acceleration.
Holy Oberth Effect! That's a big spacecraft stack! But I'm glad the central column isn't doing anything weird.
Am I the first person to try turning this into a horizontal vehicle since its release?]Great Tsiolkovsky's Ghost! Yes, kahlzun, you are the first. That's almost as terrifying as the seven Orion array I saw.
-
Well, it's pretty clear from all of your reports that I've created a bug in the mass-transference code. Not surprising, it is a little involved. I have to initialize it by pre-loading each magazine's mass (less one ton) into the engine's mass. Then when the magazine burns a bomb, it recalculates its mass, I subtract its old mass from the engine and add in the new recalculated mass. Obviously I made a mistake somewhere.
I'll put in lots of debug log statements and try to track this down. And try to not fret that by fixing this bug I might bring back the dancing magazine problem...
My only other worry is that I might have made another problem. In prior versions, there did not seem to be any problems with fancy payloads placed on the top of the engine column. Please be on the look out for such problems.
I also have the odd feeling that the engine missing the center column is more stable than the one with the column. You might want to experiment with that engine, foregoing the magazine rack and making your own center column out of struts and stuff.
-
This mission and picture was taken by user Anglave , but it is my failure. That's because it is my mod that is failing in such a spectacular manner.
-
And never mind that, missed the other 40 pages of thread... I'd be willing to help if it's needed.
We are mostly in the testing stage. So you are welcome to download the mod, install it, and bang on it until it breaks.
Then tell us about the breakage (a recipe to recreate the bug would be super).
-
Glad I could be of some (albeit minor) value to the project. I think it's an awesome mod.
Thanks! But do not minimize your contribution. I was frankly pulling my hair out trying to fix this bug. Your solution was both elegant and was a shining example of thinking outside of the box. Literally.
If I had remembered more of my rocket science, I would have remembered how sensitive rockets are to being balanced. From my website I was focusing more on having the rocket balanced radially around the thrust axis. But here there was a problem with the rocket balanced between nose and tail. That's why the launch struts were formerly required to keep the rocket from jitterbugging on the launch pad.
But the main problem was that the KSP physics engine does not play well with objects of high mass and low volume, which could not be predicted.
-
Downloading now. Will update this post with report.
Update: I'd like to report a success of the "no beans" mass-routing system!
No jitter on the pad and I expect orbit to be reasonably achievable.
Thank you so much! You solved the problem!
And I discovered that my ships no longer need launch struts. They will sit on the pad with no wobble at all.
I did have to fix it so that if a magazine was undocked or something, the mass was removed from the engine.
I do wonder if the mass is "realistic" or not though. Maybe it's being multiplied somehow?This craft:
uses the 400MN charges to ascend, and it's massive enough that the 3.5MN charges won't get it off the pad.
The RCS I installed, which was easily turning earlier versions, won't turn this craft perceptibly.
Second Update: The 400MN bomb magazines are REALLY heavy. You essentially need to use 400MN charges to lift them.
A more reasonable craft with two 80MN magazines, four 0.88MN magazines, and six standard 3.5MN magazines made a nice circular orbit. Though I don't see myself using the 0.88's. The 3.5's were quite small enough.
I will confess that I interpolated the mass of the individual 80MN charges as I had no data. But the 400NM are as per the resource material.
However, I do not have a figure for the number of charges in a 400NM magazine. It could very well be less than 60.
I do agree that the 400NM magazines are ultraheavy. About 200 tons each.
I will add a debugging readout to make sure that the mass being added to the engine is not being multiplied or something like that. Did you ever get a ship with 400MN into space before, so you could test the RCS on it?
On my ships, I have three rings of 4 RCS, one ring at the edge between engine top and bottom of magazines, one at the bottom of the engine just before the shock absorbers, and one at the top of the support column.
The 0.88 were my invention, they were intended for Jebidiah, who wanted to land his ship under atom bomb power.
The 2NM were from the USAF Orion study, the 3.5NM are from the later NASA study. The NASA charges increase thrust at the expense of specific impulse, compared to the USAF charges.
Third Update: Minor bug report, if I return to the space center and then resume my flight, all magazines are feeding.Yes, that is a known bug, it is in my Bugzilla list. The jumping magazines bug had priority.
-
Markarian421 and Anglave, when you get a chance please test the new version in the first post.
It is now in KSP 0.20 format, so please follow the directions to uninstall the old version first.
This implements Anglave's idea of removing the mass from the magazines and storing it in the engine. Warning: I have no idea if this fixes anything at all. All that I do know is that I can successfully fly into orbit and circularize the orbit while carrying a large load of magazines. But I usually can, when the same version fails on your systems.
I'm crossing my fingers.
-
My system is quite similar to yours, except I have 6 cores. and an NVIDIA GeForce GTX 570 (because I use Blender 3D, and it can utilize the GPU).
So I don't know what the problem is.
I've been working on implementing Anglave's suggestion of sucking the mass out of the magazines and injecting it into the engine. And making the magazines release upon command so Novasiliko can use it in his Orion. I'm working as fast as I can, but a bit more documentation on the API would help.
-
I tried v 0.11 -- so far no self destructs on the pad. But if I load it up with 3.5MN magazines it can't get off the pad -- it initially slowly rises, then sinks almost as if its mass is increasing.
That's pretty weird! I'm wondering if this is a processor problem. Due to the fact that I'm paranoid about people adding or removing magazines during a mission, when the engine requests a bomb from the MagazineManager, the manager does a re-inventory of the magazines. This might stress a CPU with lower specs than the computer I'm testing on. I'll look into optimizing the MagazineManager.
Never mind -- next craft I tried to make, using only one level of the 80NM magazines, spontaneously exploded on the pad. *(As soon as physics kicks in, before I hit launch.)But the MagazineManager has nothing to do with dancing magazine problem. Still working on that.
One curious thing I've noticed, I'm not sure if it's 0.20.2 or this mod -- in the VAB if I disconnect a part I sometimes can't reattach it. Or let go of it. I have to start a new craft or load one or sometimes kill the process to escape. I have only encountered it when I have a craft with an Orion engine, and it happens most often with the Orion engine part.That happens occasionally to me. Only seems to happen with the Orion engine with the column on top, not with the flat-top engine. One accidentally disconnects the engine, engine turns red, the six magazine attachment points show as green spheres but the attachment point on the top does not appear, cannot let go or re-attach engine. It's on my list of bugs.
-
OK, false alarm.
Problem was due to programmer stupidity.
Everybody, you might want to check the USAFOrion3Engine part.cfg file.
Make sure it contains
linearStrength = 2200
angularStrength = 2200NOT
linearStrength = 2200
}
angularStrength = 22000.20 version working fine now that I removed that dastardly "}" character. I have no idea how it got in there.
-
Are the folders formatted properly for .20?
I don't really want to remove these parts from my partsfolder every time there is an update
Arrgh. I formatted it for 0.20, and now it doesn't work. The items appear in the VAB inventory, you can add them to your vessel, but the Orion engine no longer adds an icon to the staging display, and it will not take off the launch pad. I'm working on it.
I'm curious that you had to put on so many magazines to get the same jumping-bean failure I've been getting. I wouldn't have thought that would be machine dependent, but I can't think of anything else between our two installs that would be different. And for me (at least in v9) even two layers of any types of magazines (except the least massive) caused a failure.Yes, as a programmer I know that is a very bad sign. If it is machine dependent, it becomes several orders of magnitude harder to root out the bug.
I had a thought. I see that you're dynamically calculating the mass of the magazines as bomb count is decremented. Any chance you could just assign the mass to the engine rather than the magazines? This might throw off the center of mass a bit, but it would allow you to keep the overall mass of the craft appropriate, and allow the mass to scale properly when adding/removing magazines.That is a very insightful and interesting suggestion. I'm certainly going to try that. I certainly sounds like it would avoid the small dense object problem.
Might be better just to depart from reality a little and have fewer, but larger, magazines. Small, heavy objects hate existing.I will certainly try to optimize in that direction. KSP has made it quite clear it does not like such objects at all.
The thing is that the one with the individual magazines worked perfectly before the last update.Please, until I get all of this sorted out, you'd be better off reverting to the version that works for you. If you no longer have the zip file PM me and I'll get it to you.
Oh lordy, that's sweet! A Kerbalized Orion!
Is it possible yet to jettison empty bomb containers? It doesn't seem to work on my design, though it could be a problem with the docking nodes themselves.Alas, jettisoning empty cannisters has not yet been implemented. It had a lower priority because it does not actually interfere with performing a mission.
Is it not possible to add a decopuler module to the magazine part.cfg file?
Currently the Orion engine is magic. As long as a magazine is part of a vessel, the engine can magically find it and suck nuclear bombs at will.
-
Are the folders formatted properly for .20?
I don't really want to remove these parts from my partsfolder every time there is an update
Sorry, I'll get on that right away.
Edit: Weird, the new version doesn't seem to fix the problem for me?Phooey. Back to the drawing board. But that's why I rely upon the good people of this thread to do testing.
Canopus, you might try manually editing the engine part.cfg and increasing the mass
It's a little bit sad that this thing can't destroy Roche anymore. Otherwise this would be the perfect defense against killer asteroids:DWell, if you want to destroy Roche, edit all the magazine part.cfg files and set destroyMass = 1000
-
New Version Up for Testing
Go to first entry in thread
NovaSilisko suggestion to increase the engine mass to 30 seems to have fixed things (I hope). The new version restores the mass of the magazines, and increased the mass of the engines to 30. The largest magazine has a mass of 220.98 tons, this puts it within x10 of the engine mass. The smallest magazine has a mass of about 5, this puts the engine within x10 of the magazine mass.
Anglave, Markarian421 please try this version (or edit the two Orion engine part.cfg files).
s20dan, you may have been ninja'd, but you are to be congratulated for independently coming up with a solution.
I did notice that if I made a stack using the multi-magazine on top of columnless engine, the result was actually stable enough for the vessel to sit on the launch pad with no launch supports. This is not true with the individual magazines.
I am considering making the center column into a separate part. Or something. The engine with the center column seems to have balance problems compared to the column-less engine.
-
It's not so much that as having said dense loads dangling off the side of something that's very lightweight. Try increasing the engine mass to something like 30, see if that does anything? Last I checked it only weighed 2 units.
Anyway, the collision mesh issue - you can use as many colliders as you want. Just model a bunch of different chunks of the whole and put them all together.
Ah! I'll try increasing the engine mass. Didn't think of that.
But there may be problems if I stack a 4 ton magazine on top of a 200 ton magazine. Time to experiment.
Multiple colliders, obvious in hindsight, boy to I feel stupid.
Also, you replied to my post twice
oops, sorry about that.
-
nyrath, while you were working on this latest version this morning, I did as you suggested and tested some altered magazines with reduced mass.
The magazines work fine when they're paper-light.
Unfortunately, as you noted, this does somewhat affect the mass/weight ratio of the craft. It's Ludicrous
In my latest version, the fix is basically making the part.cfg of the 80MN magazine have bombMass = 0.2 instead of 1.152, and part.cfg of the 400MN magazine have a bombMass = 0.3 instead of 3.68
I could lower the bombImpulse to address the mass/weight ratio, but we want this thing to be able to lift off.
I get the feeling that the KSP physics engine was not designed to handle such dense loads.
-
So! Inspiration struck me and I cranked out this thing over a period of about two hours or so. Going to be powered by this plugin. Three parts, the cabin (built-in docking doors and payload bay, crew of 4 or 6 kerbs), bomb canister (built in docking port, 2.5m diameter), and orion engine (6 built-in docking ports to attach bomb canisters). The whole thing is 9m diameter. The payload bay has more than enough room for a big lander, or huge amounts of cargo.
That model is gorgeous!
I have a question, but you do not have to answer if you care not to.
I have problems with the mesh of my engine in the central column section. The areas inside the column for the stacks of magazines makes the collision mesh have concavities, which are not allowed in a convex collision mesh. Your engine with the build-in docking ports for the bomb canisters seem to have the same concavity problem? How does one address this?
I'm trying to think of how you could possibly land it to put in more cargo, since flying into an atmosphere retrograde with an orion engine means having your nukes blasted by air resistance back into the rear of your spacecraft, where they then detonate and kill everyone (though this isn't simulated by the mod)The USAF and NASA 10 meter Orions used a cannon powered by compressed ammonia to fire the nuke through a hole in the pusher plate.
For a tail lander, offhand I'd say you'd have to make the nukes mounted on tiny missiles in order to get them to the desired detonation point despite the air resistance.
At least that's my story and I'm sticking to it.
The mod does not try to simulate the nuke placement since there was no point. Happens too fast to see, and it contributed nothing to gameplay.
I did try using girders and things to make long skinny "legs" with retractable landing feet on the ends. The Orion still came down hard enough to blow up when I tried playing atomic "Lunar Lander" to the left of the space center launch pad. I'll have to try it with the 0.88MN tiny nukes.
-
New Version Up for Testing
Go to first entry in thread
I drastically reduced the mass of the larger magazines. I did finally manage to reproduce the problem by making tall stacks of the 400MN and 80MN magazines, including the hovering in mid-air physics breakage.
I did note that with massive stacks of magazines, the advanced SAS was not up to the task of keeping the vessel's nose pointed in the desired direction. You need lots of RCS as well. In VAB, use center of gravity indicator.
Anglave, Markarian421 try this version. If you are still having problems I'll reduce the mass some more.
I did try reducing the mass to under 1 ton. The control capsule popped off. When I tried tying down the capsule with lots of struts, two 400MN detonations was enough to send the vessel into Kerbin escape trajectory.
-
So, I'm at home now, on a much more powerful machine. I'm running Win7 x64 and have a fairly serious gaming rig.
I've got 0.20.2 installed, and the only mod I use or have installed is Kerbal Engineer Redux.
I downloaded a fresh copy of Orion v9 and unzipped it. Launched KSP and built the same stack.
Clicked launch and almost perfectly recreated the same vibration->airborne spinning->explosion. Never activated the engine nor fired a nuke.
The mod seems perfect and the engine is quite usable so long as I only try to use one layer of magazines, and only 2MN or smaller magazines.
I just installed this tonight, and it's great fun, thanks!Whenever I tried the engine with the built in rack and manually added magazines, the results were similar to what's being reported above -- the ship would tear itself apart/explode as soon as I hit launch. The magazines shake and violently eject. I started over from scratch with the same end result. Using the other engine and any combination of magazines I've tried so far works fine. OS X, KSP 0.20.2
Anglave, Markarian421 if you feel up to it, I'd like you to try an experiment. Only if you feel up to it, mind you.
In the Parts folder there are three folders with the problem magazines: USAFOrionMag1kt3_5mn, USAFOrionMag5kt80mn, USAFOrionMag15kt400mn. I'm reasonably sure the problem is because the magazines have too much mass. Let's try changing that.
In each of those folders is a part.cfg file. You should have the originals in the zip file you downloaded, so you can edit the ones in your KSP folder.
For each part.cfg file, open the file with notepad or some other simple text processor. Do not use MS word or some fancy word processor because it might alter the file in such a way that KSP cannot read it (specifically by altering the "end of line" character).
In each file, find the line that says
mass = 13.3386
(or whatever mass)
and change it to
mass = 1
In each file, fine the line that says
bombMass = 0.141
(or whatever bombMass)
and change it to
bombMass = 0.01
Save the files, then try making big stacks again and see if they blow up on the launch pad.
As near as I can figure, the problem is due to the fact that if a part is linked to another part where the mass difference is larger than x10 or so, the parts start to jump around. Since some of these magazines have a mass difference of x100 or so, that could be the cause.
Again I'm sorry the mod is being so difficult.
-
So! Inspiration struck me and I cranked out this thing over a period of about two hours or so. Going to be powered by this plugin. Three parts, the cabin (built-in docking doors and payload bay, crew of 4 or 6 kerbs), bomb canister (built in docking port, 2.5m diameter), and orion engine (6 built-in docking ports to attach bomb canisters). The whole thing is 9m diameter. The payload bay has more than enough room for a big lander, or huge amounts of cargo. I'm trying to think of how you could possibly land it to put in more cargo, since flying into an atmosphere retrograde with an orion engine means having your nukes blasted by air resistance back into the rear of your spacecraft, where they then detonate and kill everyone (though this isn't simulated by the mod) Could just be a space-only craft, though.
Anyway, here's a render of the whole mess:
That is glorious NovaSilisko! Looking forwards to seeing the finished product.
Let me know if you have any questions about the code, though I'm sure you will have little problem figuring it out (especially since it is based on the code you gave me). Any difficulties understanding my code will probably be because I'm a novice at C#. The readme file has a few notes about making the mod work with other mesh models.
-
I clicked the "launch" button, which loaded it sitting on the pad.
Then the photo, above.
Ah, what was the kind and number of each magazine? I'm trying to get the bug to reproduce on my system here. Until I can reproduce it, it is almost impossible to fix it.
Sorry the mod is being so uncooperative.
-
{ facepalm }
Ohboy, that's bad. I'm sorry you are having so much trouble. I'll have to think about this.
You might try using some struts on the magazines, bottom and top, to see if that changes anything.
You did launch your ship by setting the throttle, turning off all the magazines except the most powerful one, and hitting the space bar, right?
-
And while you are at it, be sure to shut down the entire game and restart it again. Just to be sure.
I've had KSP start acting weird sometimes, but a restart fixes it. Things like in the VAB, I set the symmetry button to "4" then it re-sets itself to "6"
Update:
I'm on Steam, so I am forced to use the latest version. According to the title screen I'm using v0.20.2.186
I'm on a Windows 7 (64-bit) with an AMD 3.50 GHz six-core processor with 8 GB RAM, not that should make a difference.
Orion aka "Ol' Boom-boom"
in KSP1 Mod Development
Posted
Looks good. Lets see, 10MN is about x3, 15MN is about x4, and 20 is about x6 compared to a 3.5MN.
How about 10 MN?