crazy idea: artificial planet ring

[Norby]

A test under 1.77.1 gives exactly the same results as under 1.79 for me (under Windows). So as I recall you're on Linux (?) we may have a difference there?

In my Linux+HD4000 the ring is perfect under 1.77.1 also.

[Thargoid]

In my Linux+HD4000 the ring is perfect under 1.77.1 also.

Thanks - so alternatively it may be a driver issue perhaps?

I hate these oddities...

[Zieman]

Did some testing with this OXP on my work laptop.

Win 7 Pro 64-bit & NVIDIA Quadro 1000M &

- Oolite 1.77.1 32-bit : ok
- Oolite 1.77.1 64-bit deployment : ok
- Oolite 1.77.1 64-bit test : ok
- Oolite trunk-1.79.0.c272d89-dev.x64 : ok, except transparent waystations

[Thargoid]

Thanks - is the 1.79 the same set up you had the probs on before? Also did you use the OXP as dewnloaded, or with the modified shipdata code I posted the other day? (for the station transparency issue).

But I have no idea where Dizzy's odd results are from?

A question for the dev's - could there be a graphics card/driver and/or OS dependence here?

[Zieman]

Thanks - is the 1.79 the same set up you had the probs on before? Also did you use the OXP as dewnloaded, or with the modified shipdata code I posted the other day? (for the station transparency issue).

But I have no idea where Dizzy's odd results are from?

A question for the dev's - could there be a graphics card/driver and/or OS dependence here?

I've so far used the oxp as downloaded, no edits.

And the 1.79 setup should be the same, although the nightly I just downloaded looks to be one step newer that the one I build on my home desktop.
To reiterate, only difference between those 1.79 setups I've used should be the fact that one I downloaded as a pre-built nightly (work laptop - no probs other than transparency on waystations) and the other I built myself, as per the instructions https://bb.oolite.space/viewtopic.php?f=8&t=5943#p73980 (messed up).

Work Laptop:
i7-2630QM, NVidia Quadro 1000M, 16 GB RAM, SSD, Windows 7 Pro 64bit

Home Desktop:
i7-4770K, Asus GTX780-DC2OC-3GD5, 16 GB Kingston Hyper-X Beast 2400 MHz, Kingston Hyper-X 3K 240 GB SSD, WD Caviar Black 2 TB, Windows 7 Ultimate 64bit

[Thargoid]

Thanks again. So similar OS's but different graphics set ups.

I'm not sure I like the way this glitch is headed for this concept becoming something useful in-game.

[Zieman]

I'm now downlaoading stuff to test the self-build version on my work laptop, to see if that makes any difference. If not, then I''l test with my home desktop, and make sure no other Oolite version is installed when I test a particular version.

[cim]

Looks impressive in-game, though the entity count is quite significant.

Transparent stations: both the 1.77 Coriolis and 1.79 Coriolis have subentities, but different ones. So in 1.79 you override the standard subentities (back to the 1.77 ones), making much of the station's hull disappear. Your fix works fine for that. (having several levels of like_ship is fine, by the way)

Alignment and position: looks fine here on 1.79 except near the stations where the visual effect parts seem a little off in position.

If it's only a problem on (some) 1.77.1 builds then it's probably not a problem for the future, but obviously tracking down what fixed it for potential backporting to 1.78 would be extremely difficult.

[Diziet Sma]

@Dizzy - So as I recall you're on Linux (?) we may have a difference there?

Correct.. but thanks to a new 750GB hard drive in this laptop, I now have XP, Win7, Linux Mint16, Ubuntu Studio and ArchLinux all booting on the same machine. So far all the testing I've done has been under my main workhorse, Mint16, so I'll check it out under Windows, and fire up the ol' MacBook (which also dual-boots Mint16) for a couple of tests as well. I'll also try some different graphics settings in Oolite, see if that makes any difference.

re the rest of the discussion since your post, there may well be a graphics chip/driver problem behind this.. my laptop uses a reasonably obscure AMD/ATI chip (Radeon X1300 - a laptop version of the X2300) that was quickly made obsolete. Older AMD/ATI drivers in Linux (and Windows, for that matter) are a bit of a sore point in general. On the other hand, the same drivers are being used by both 1.77.1 and 1.79, so who knows?

If you can figure out some logging code to print out the actual positions of the segments vs the calculated positions, might it be easier to see what's going wrong? But then, since apparently I'm the only person getting this effect, it may not be worth your while.. your call.

Edit: I've just had another thought.. I wonder if this is some weird interaction with another OXP that I have installed under 1.77.1, but not under 1.79.. more testing, methinks!

Edit 2: MYSTERY SOLVED.

You are not gonna believe this.. it's the Griff_Shipset_[color=#FF0000]Replace[/color]_v1.34.oxp. When I remove the OXP, the ring looks identical to when it's under 1.79.

The Griff_Shipset_[color=#FF0000]Addition[/color]_v1.22.oxp, on the other hand, is not a problem.

This explains why it was not turning up under 1.79, since the replacement set is basically redundant for 1.79, unless you really want the shadered ships, and therefore I didn't install it.

You can check it for yourself by installing just the Rings demo, the replacement set, and, of course, you'll need the Griff_Shipset_Resources_v1.2.25.oxp for the textures, etc..

[Zieman]

Edit 2: MYSTERY SOLVED.

You are not gonna believe this.. it's the Griff_Shipset_[color=#FF0000]Replace[/color]_v1.34.oxp. When I remove the OXP, the ring looks identical to when it's under 1.79.

The Griff_Shipset_[color=#FF0000]Addition[/color]_v1.22.oxp, on the other hand, is not a problem.

This explains why it was not turning up under 1.79, since the replacement set is basically redundant for 1.79, unless you really want the shadered ships, and therefore I didn't install it.

You can check it for yourself by installing just the Rings demo, the replacement set, and, of course, you'll need the Griff_Shipset_Resources_v1.2.25.oxp for the textures, etc..

Ninja'd by Dizzy!

[Norby]

it's the Griff_Shipset_[color=#FF0000]Replace[/color]_v1.34.oxp.

Confirmed: the effect is the same under 1.77.1 and 1.79, if I undock immediately in a new game then I can see many explosions which push away the remaining parts from the originally correct positions.

[cim]

Of course...

The ring segments are like_ship of an asteroid. Asteroids, in Griff's shipset, get a ship script which on spawning gives them a small random velocity. So with the Replace shipset, that gets applied to the basic asteroid, and so to the segments too...

[Norby]

get a ship script which on spawning gives them a small random velocity.

So a script=""; in ring_segment part of shipdata solved it well at me.

[Thargoid]

I'll try and grab some time later and rewrite the demo to remove the like_ships where possible and see where we go from there. I'd still like to know why things are misaligning near the gateway station too.

[Thargoid]

OK, I've just uploaded a new version of the demo, with a few changes:

• Removed the like_ship for the segment (but the gateway station still uses one at the moment - TBD list includes to make a proper ring station rather than re-using a Coriolis) to overcome the issue with Griff's replacement set.
• Moved the ring up to mid-way between the planet surface and the main station orbit height, to stop NPC issues with low altitude warnings (allow them to launch and dock).
• Changed the model for the ring segments and joints, to look a little nicer.
• Tweaked the script a little to try and overcome the gaps at the end of the ring spawning (removed a few of the Math.floors).

The gaps are still there near the end of the ring - it looks to me like we're just seeing rounding errors or similar magnified on a planetary scale to form the gaps (which are <100m in terms of position). I'm not sure how that can be improved (even in 1.79), here any ideas are welcome.

Download link is - here -

BTW a thanks to everyone who has chipped into this demo - I just love how these fun little ideas grow into such community projects [/color]