Partial bug in Web Info on 1.5 pre 5

[rthorvald]

Try getting a fresh copy from CVS. I'm having no problems with today's CVS.

Will do that, and report back...

- rthorvald

[rthorvald]

Weird, but problem solved: my new build of today removed the bug. I must have got a corrupted download last time or something...

Thanks, everyone

- rthorvald

[chris]

This is a different problem than the one that Runar is talking about. The add-on just uses a regular old cel URL, but unfortunately URLs produced with older versions of Celestia are not compatible with 1.5.0 due to a number of changes.

--Chris

So what you're saying is that we can no longer specify - as an
example - that an add-on "works with Celestia 1.4.1 or later."

Thanks a lot...

Yes, add-ons can work with Celestia 1.4.1 or later, but URLs they use will be broken for one build or the other.

Seems to me that this development team could use a better
organized approach to these matters.

And that is what exactly? There's really not a good solution for URL compatibility due to the nature of the KM_PER_LY bug that is in largely responsible for the problem. The best that Celestia could do is make a guess at what position in space an old URL is really pointing to. The fact that the Sun changed from being a static object ~200 AU from the origin to an object that orbits the SSB position right at the origin is also a problem, albeit a lesser one. Celestia could guess that any URL referring to a point within say a light year from the origin should be adjusted for the new position of the Sun. And since the Sun is moving in 1.5.0, that adjustment is time dependent! See how messy this gets? It's actually even worse, as the interaction between the KM_PER_LY bug and the change in the Sun's position compounds the difficulty of disentangling an pre-1.5.0 cel URL.

The fundamental problem with cel URLs now is that they specify position in Celestia's universal coordinate system. Whenever the position of an object changes, any cel URLs that would have placed the observer near that object get broken. Grant recently checked in some revisions for the orbits of Pluto, Charon, Nix, and Hydra. Any cel URL that placed the observer close to one of these objects is unlikely to work now, and it's not a result of programmer laziness that these URLs are broken.

The solution is to make URLs relative to the currently followed object (or more generally, that the URL stores observer coordinates in the current reference frame), so that they're less susceptible to breaking whenever an orbit is revised. And don't worry: when we do get around to changing cel URLs to use relative coordinates, old absolute cel URLs can still be made to work.

--Chris

[BobHegwood]

Sorry Chris...

Didn't mean to imply that you were doing something wrong.

It's just that - if these CEL URL problems ARE in the state you say that
they are - then why do you allow any references to them within
Celestia in the first place?

Seems to me that this problem will only get worse as the technology
gets more and more exciting (and confusing.)

I know I'm not the smartest guy in the world, but I very much doubt
that the creators of the add-ons I referenced above gave ANY thought
to what you just explained here. They - like me - would simply assume
that it will work from now on.

Thanks, Brain-Dead

[chris]

Sorry Chris...

Didn't mean to imply that you were doing something wrong.

It's just that - if these CEL URL problems ARE in the state you say that
they are - then why do you allow any references to them within
Celestia in the first place?

Because cel URLs created with 1.5.0 are still very useful.

Seems to me that this problem will only get worse as the technology
gets more and more exciting (and confusing.)

No, it should be less of a problem now that 1) bugs have been fixed, and 2) Celestia now has a better defined coordinate system (with an origin at the solar system barycenter.) Switching to relative coordinates would further improve the long term consistency of cel URLs.

I know I'm not the smartest guy in the world, but I very much doubt
that the creators of the add-ons I referenced above gave ANY thought
to what you just explained here. They - like me - would simply assume
that it will work from now on.

Certainly the bugs were not unforeseen and I don't expect Celestia users to have to concern themselves with exactly how those broke cel URLs. However, people creating cel URLs should understand that they record an absolute position in space, and that another user clicking on the URL might not see the same view if they have a different set of SSC files installed.

--Chris

[t00fri]

Quite unemotionally stated: if add-on creators will not care to adapt their add-ons to the progressing core code of Celestia they should not be further supported.

We cannot stagnate with Celestia development, just because some add-ons are NOT updated by their creators.

Bye Fridger

[BobHegwood]

Certainly the bugs were not unforeseen and I don't expect Celestia users to have to concern themselves with exactly how those broke cel URLs. However, people creating cel URLs should understand that they record an absolute position in space, and that another user clicking on the URL might not see the same view if they have a different set of SSC files installed.

--Chris

Okay, well at least I may be able to explain some of these problems
given the explanations here.

Maybe I can add some more comments to the aforementioned
add-ons to avoid confusion.

Thanks, Brain-Dead