I've found a problems with stars if I use the built 4903 (not done by me, I must admit, but by Cartrite).
The problem is shown in another thread, you'll find below the links:
http://www.shatters.net/forum/viewtopic.php?f=4&t=12221&start=135#p119782
and
http://www.shatters.net/forum/viewtopic.php?f=4&t=12221&start=135#p119784 (thanks @abramson)
I put again the picture.
I have nothing to add. Except that's funny.
Teto.
Half star! (built 4903 for Windows)
-
- Site Admin
- Posts: 4211
- Joined: 28.01.2002
- With us: 22 years 9 months
- Location: Seattle, Washington, USA
Re: Half star! (built 4903 for Windows)
Here's a patch that fixes this bug:
http://www.celestiaproject.net/~claurel/celest ... here.patch
The problem was a tricky floating point precision issue. The reason that stars were affected and not planets is that stars generally are much bigger than planets. The larger radius of stars was enough to cause a terrible rounding errors in a matrix inversion. For the fix, I changed the problem code to use a coordinate system scaled so that one unit = one planet (or star) radius.
--Chris
http://www.celestiaproject.net/~claurel/celest ... here.patch
The problem was a tricky floating point precision issue. The reason that stars were affected and not planets is that stars generally are much bigger than planets. The larger radius of stars was enough to cause a terrible rounding errors in a matrix inversion. For the fix, I changed the problem code to use a coordinate system scaled so that one unit = one planet (or star) radius.
--Chris