New Celestia-1.4.0pre-FT1.2 Version for Download

[t00fri]

Fridger is right - regardless of whether or not one likes the forum interface, one can't deny that the forum gets a lot more traffic than the mailing list does. Yes, some of that is noise on the development board, but a lot of it isn't. I think the extra comments also give an added impetus to developers to keep working because they're getting more feedback and they know that people are encouraging and helping them to reach their goals.

Right

[t00fri]

Pat,

incidentally, if you want me to test your new GTK hacks for Windows, just send me the project files or nmake Makefile and I'll have a go at it. Are you now professionally related to GTK development somehow??

Certainly I can see the "charm" that a truly "unified" Celestia approach would have, based on a professionally maintained cross-platform toolkit like GTK...

As I very honestly expressed to Paulo several times, I cannot get excited very much about "home-brewn" cross-platform toolkit solutions, however.

Cheers,
Fridger

[t00fri]

Just consider that meanwhile (via this Forum interface) we managed to have an almost synchrone release of FT-OSX, including lots of valuable feedback from the MAC community!

This is great news. Are they using gcc or something a la Codewarrior?

They are using gcc 4.x (!). It took us a few tweaks to get the code compiled with 4.x. But now it compiles "off the shelf"...

I expect that building the GTK version on OSX would be easy with the built-in X server running. By next year I expect that it will run completely natively. Check this out:

http://micke.hallendal.net/archives/200 ... acosx.html

Neat!

Cheers,
Fridger

[Malenfant]

Don't suppose there's any chance that the 'offset Sol' problem could be solved in FT2.0 could it? i.e. the fact that Sol isn't at 0,0,0 of the coordinate system?

I asked Selden why it wasn't at 0,0,0 in this thread but he didn't know... is there any good reason why this is still offset?

[Cham]

Malenfant,

the only reason the sun still has this stupid offset is to maintain compatibility with old URLs and scripts. It's a very old blunder. I don't know what exactly it will break, if we place the sun at its real position (0, 0, 0).

[t00fri]

Malenfant,

the only reason the sun still has this stupid offset is to maintain compatibility with old URLs and scripts. It's a very old blunder. I don't know what exactly it will break, if we place the sun at its real position (0, 0, 0).

It's even worse: we still have this (quite significant) typo in the speed of light! The main reason why it is still there, is that its correction would invalidate many existing cel://urls etc ...

Lots of work ahead, friends

Bye Fridger

[Malenfant]

Malenfant,

the only reason the sun still has this stupid offset is to maintain compatibility with old URLs and scripts. It's a very old blunder. I don't know what exactly it will break, if we place the sun at its real position (0, 0, 0).

It's even worse: we still have this (quite significant) typo in the speed of light! The main reason why it is still there, is that its correction would invalidate many existing cel://urls etc ...

Lots of work ahead, friends

Bye Fridger

What typo in the speed of light?

I'm still not seeing what would break if it was changed to 0,0,0 though. Do many URLs and Scripts rely on the exact xyz coordinate of Sol rather than a coordinate relative to where Sol is? Personally I'd be inclined to say "screw the cel URLs and scripts" and change it to 0,0,0 anyway... it's the surely scriptwriters' responsibility to update them to keep up with Celestia, not the developers' obligation to keep things the same (and incorrect) for the scriptwriters.

[Cham]

Personally I'd be inclined to say "screw the cel URLs and scripts" and change it to 0,0,0 anyway... it's the surely scriptwriters' responsibility to update them to keep up with Celestia, not the developers' obligation to keep things the same (and incorrect) for the scriptwriters.

I agree with this. To Hell the scripts. Make Celestia better !

[t00fri]

Here is a relevant post from Chris about that typo:

Yes, I believe that we should be using the Julian year. Either way,
there's a big discrepancy with the current value of light year.

--Chris

Isn't the reference for a light year the julian year rather than the tropical or sideral?

In any case that's what Wikipedia and my units.dat file say.

Code: Select all

1 ly = 365.25 * 86,400 * 299,792,458 = 9,460,730,472,580,800 m


Celestia's present value is incorrectly

Code: Select all

astro.h:#define KM_PER_LY     9466411842000.000


The typo has been spotted by the user 'tlittle' in March 2005 here:

http://www.shatters.net/forum/viewtopic.php?t=6975&highlight=

Bye Fridger

[Malenfant]

Huh.

OK... so is there any reason why this can't be corrected in FT 2.0? And does this screw anything up if it is corrected?

I mean, if some values are incorrect then what other opportunity do you have to correct them if not in a new version of Celestia?

[t00fri]

Huh.

OK... so is there any reason why this can't be corrected in FT 2.0? And does this screw anything up if it is corrected?

I mean, if some values are incorrect then what other opportunity do you have to correct them if not in a new version of Celestia?

Malenfant,

clearly, changing this value would take only a few seconds.

But...the real work would be to check /again/ many phenomena predicted by Celestia that are sensitive to that value. Like checking the famous mutual events ( Jean Meeus) among the Galilean Moons. Will they still be fine as they were with the wrong value of KM_PER_LY? etc...

As a reminder of what we did previously (May 2002),

http://www.shatters.net/forum/viewtopic.php?t=1506&postdays=0&postorder=asc&start=0

Or this

http://www.shatters.net/forum/viewtopic.php?t=1492&start=4

and many other precision tests...

So this seems like a big chunk of precision work, really, and any experienced volunteers would be highly appreciated.


Bye Fridger

[t00fri]

If there are voluteers willing to pusue these important and most interesting/rewarding tests once more, I could easily provide a compiled Windows or Linux version with the corrected value of KM_PER_LY!

Reports in the forum about the tests would be most interesting.

Bye Fridger

[Malenfant]

Exactly what tests would you propose running? Would it be just to compare Celestia's predicted eclipse values with the 'official' ones in e.g the Astronomical Almanac?

I wouldn't mind testing it out anyway, if you want a volunteer.

[t00fri]

Exactly what tests would you propose running? Would it be just to compare Celestia's predicted eclipse values with the 'official' ones in e.g the Astronomical Almanac?

I wouldn't mind testing it out anyway, if you want a volunteer.

I think the best /start/ might be to compare Jean Meeus tabulated mutual events (there are many!) with Celestia's predictions. Both as to the appearance of the event and to the timing if the LT is activated (light travel delay).

After I did all these comparisons for the first time (thereby discovering a most cucial bug in the code ) I was /forever/ deeply impressed about Celestia's amazing precision (based of course on VSOP87 the father of which was Jean Meeus).

I suppose much information can be found in those old threads from 2002 where Grant and I interacted in a most fruitful and enjoyable way (that I still miss today).

The mutual events are listed regularly in S&T in a column by Jean Meeus. The one we used was in Dec. 2002 S&T, p. 100 ff.

In principle it's easy. Look up the event time in S&T set it in Celestia go to the Jovian system and activate LT. See whether there is an event at this time (within a few SECONDS!) and whether it looks as predicted. It's really amazing to predict an /annular/ eclipse shadow of Europa on Io. Both moons are really "small" i.e. the shadow cone must be pointing very close to perfection .



Do you have S&T available? I am sure Selden could also find these event tabulations somewhere on the net


Bye Fridger

[t00fri]

Or look at these most successful tests of Pluto-Charon eclipse events.



from this thread of mine in June 2003

http://www.shatters.net/forum/viewtopic.php?t=2477&start=0

In all cases the resulting event timings are sensitive to the value of KM_PER_LY, notably for large distances like Pluto.

Bye Fridger

[Malenfant]

In principle it's easy. Look up the event time in S&T set it in Celestia go to the Jovian system and activate LT. See whether there is an event at this time (within a few SECONDS!) and whether it looks as predicted. It's really amazing to predict an /annular/ eclipse shadow of Europa on Io. Both moons are really "small" i.e. the shadow cone must be pointing very close to perfection

I've got the Observers Handbook 2005 and 2006 for the Royal Astronomical Society of Canada. They've got mutual event timings in there calculated by the "Institut de Mechanique Celeste et de Calcul des Ephemerides" in Paris - will that be the same as the S&T values?

The times are given in UT of 2005 (or 2006) geocentric phenomena.


Howveer, I presume I don't actually GO to the Jovian system though do I - shouldn't I stay at Earth, select Jupiter and centre it in the display (and zoom in), and see if the time of the event agrees with the prediction?

[t00fri]

Here are some online S&T mutual events:
(Jean Meeus)

http://skyandtelescope.com/printable/observing/objects/planets/article_774.asp

Bye Fridger

[t00fri]

Howveer, I presume I don't actually GO to the Jovian system though do I - shouldn't I stay at Earth, select Jupiter and centre it in the display (and zoom in), and see if the time of the event agrees with the prediction?

Precisely. Sorry for being sloppy.

You land on Earth, Push "*" (turning towards the sky), get rid of the clouds ('I') and center Jupiter on the screen.

Set the mutual event time (UTC). Then you zoom with movement of 'SHIFT+mouse Left' until you see the moons of interest in high 'magnification'. Toggle LT on|off and watch the effect.

Try Jean Maeeus' event timings I gave above from S&T.

Bye Fridger

PS: It seems I am going to compile a Windows FT1.2 executable with the correct KM_PER_LY and deposit it in the Texture foundry . Got to go to bed now (past 1 am). Will do it tomorrow.

[Malenfant]

Here are some online S&T mutual events:
(Jean Meeus)

http://skyandtelescope.com/printable/observing/objects/planets/article_774.asp

Bye Fridger

Hm, I found a calculator on the S&T website at
http://skyandtelescope.com/observing/ob ... 830_2.asp#

but the times don't seem to exactly agree with the ones given in my Observers Handbook. Sometimes they're a few minutes ahead, sometimes they're a few minutes behind, sometimes they're bang on.

eg the S&T calculator says:

Friday, December 2, 2005

10:40 UT, Io's shadow begins to cross Jupiter.
11:20 UT, Io begins transit of Jupiter.
12:50 UT, Io's shadow leaves Jupiter's disk.
13:32 UT, Io ends transit of Jupiter.

Whereas the times I have for those events in my book are:

10:40 UT ShI
11:19 UT TrI
12:50 UT ShE
13:30 UT TrE

(the events are the same, it's just abbreviated in the book).

I'm assuming that the times in the Handbook are the more accurate ones, but is that correct?

[t00fri]

Well, just quickly before really going to bed: I own /many/ "Astronomical Ephemeris Handbooks" older than 1995, which have never disappointed me. However, at that time VSOP87 was not yet used for mutual events! Do you have the "Supplement ..." as well? It should be explained there what theoretical approach was used.

I would be very careful with WEB-based calculators (of mutual events) . To achieve the required precision for our purpose, one will really need a fairly extensive calculational engine. I seriously doubt that this would be implemented in such a javascript WEB-calculator. The S&T event calculator you quoted really looks more like a tool for hobby astronomers...


It's much safer to compare the /tabulated/ events in Jean Meeus' article above. This guy has been a real perfectionist throughout his life! So I trust him very much.

Bye Fridger