Page 1 of 2

Has core Celestia development stopped?

Posted: 23.03.2005, 01:56
by Evil Dr Ganymede
Has there been any recent development of Celestia's core at all (i.e. rendering engine, etc)?

I'm not really interested in new/updated textures and I'm certainly not interested in fictional addons, but since the addition of multiple light source support last year it seems like it's been all quiet on the core development. And there hasn't been a new prerelease for quite a while now.

I'm also not particularly interested in people saying I'm impatient, or that I should do some of my own development, before anyone starts on that. I'd just like to know whether development of the parts of Celestia that I'm interested in (mainly the rendering engine, and more complete sscs for asteroids, comets, moons, extrasolar planets etc) is moving forward at all. Or has such development stalled completely now?

Posted: 23.03.2005, 07:10
by Brendan
I'll be busy this spring, but this summer, I should have time to do some work. What I'd likely do is do little feature additions in BCelestia. My moon ssc files have all of the data that I've found on the known moons. With extrasolar planets, minor bodies and locations, I'd look into where data about them is available and find if there's any way to fetch the data and automatically convert it into ssc files. I'll be learning the Ruby language that I'd use on text files like that.

Posted: 23.03.2005, 10:27
by Evil Dr Ganymede
That's all well and good, but will this info be included in the default version of Celestia? I'm not really interested in alternative versions of the program (though I'm not knocking your version at all).

I know Chris has other things to do, but I've not heard a peep about what he's currently working on (even if it is slowly)

Posted: 23.03.2005, 17:47
by Paolo
I think that everyone of us is somewhat worried about this long period of pre-releases and news shortage concerning the Celestia development.
IMHO it is necessary that Chris reassures the community explaining which is the current situation.
Moreover I would like to know if he is available to evaluate and recruit new coders in order to extend the development crew.

Posted: 23.03.2005, 22:04
by t00fri
Paolo wrote:...
Moreover I would like to know if he is available to evaluate and recruit new coders in order to extend the development crew.


The latter is not an easy task. There are surely many people whose C++ coding abilities are excellent. But Celestia has become meanwhile a most complex machinery packed with lots of high-precision astro-mechanics, astrophysics etc. While porting tasks from one OS to another mainly require good knowledge of C++ and the respective graphics toolkit, /creative/ further core development requires much more know how about physics, etc. And as to OpenGL graphics implementations, Chris is hard to replace ;-)

I see the problem along a quite different line, but I won't go into it here in public.

Bye Fridger

Posted: 24.03.2005, 08:54
by Paolo
Fridger

t00fri wrote:...Chris is hard to replace ;-)


this is quite true. My idea is that if someone gives Chris a hand and if he should coordinate the efforts of a pair of coders the development should remain continuously in progress.

The point about the internal complexity is well known. But also in the past some new core features was discussed, sketched, implemented and then refined through the test-debug cycle by the experts like you, Selden, Granthutchison and so... Almost nothing of new and complex is perfect at first tentative. Specially in software implementation :wink:.
The most important thing is to have new prototypes to test and work on.

After that there are other things important: e.g. to prepare a final release for distribution collecting all the files and updating the gui of the program.
Debug is a tedious and expensive task but is very important too. The time for these three tasks almost for the windows version is in charge to Chris alone and can be surely delegated to someone else.

I know that extending the crew should generate problems. To avoid them it is important for the members to follow two simple rules: share and discuss ideas before implementation and document every change in CVS tree.

Bye - Paolo

Posted: 24.03.2005, 11:47
by ElChristou
t00fri wrote:I see the problem along a quite different line, but I won't go into it here in public.


What a pity!! as newcomer all those rumors are worrying... wanted to know more... :(

Posted: 24.03.2005, 11:56
by ElChristou
Paolo wrote:After that there are other things important: e.g. to prepare a final release for distribution collecting all the files and updating the gui of the program.
Debug is a tedious and expensive task but is very important too. The time for these three tasks almost for the windows version is in charge to Chris alone and can be surely delegated to someone else.


Quite agree... indeed I begun to worder if the forum was only to discuss the add-ons!! There is a lot of little implementation/request wich are not part of the physics laws... but for know all seems really so... pre-dead... :?
Who are the others developpers?

Posted: 24.03.2005, 14:58
by t00fri
Paolo wrote:Fridger

t00fri wrote:...Chris is hard to replace ;-)

...

The point about the internal complexity is well known. But also in the past some new core features was discussed, sketched, implemented and then refined through the test-debug cycle by the experts like you, Selden, Granthutchison and so...
...


Paolo,

I was NOT referring to the debugging-testing part. Almost everyone knowing how Celestia is operated and being a 'diligent character' can do this task eventually.

I was talking about the creative part of joint /new/ core development.
Here plenty of know-how is unavoidably required besides mere coding ability. For quite a while Chris, Christophe and myself formed a nicely tuned, effective team, notably since the really challenging coding tasks were well discussed beforehand, planned, implemented and debugged jointly. Grant is really hard to replace, since he continuously implemented and updated astronomical data input with /great knowledge/ and utmost /care/. He was never involved in coding, however. /All/ 3d Graphics tasks rest exclusively in Chris' hands, since none of the others (unfortunately) knows much about OpenGL (except dirkpitt, who as a recent member, specializes on the OSX porting).

But our individual "boundary conditions" all changed with time. My professional duties left less and less spare time for the very time consuming part of coding. Well,... and the others were also getting interested in new activities or just went away, as we know...

This left-over configuration still can lead to code checkins here and there, but major new implementations requiring a 'brainstorming' of a team with complementary expertise, seems out of the question, meanwhile. That's at least how I see it...

Anyway, as a nice optimistic example, Christophe submitted two days ago a complete i18n (multi-language) implementation of Celestia!

Let me end by saying that I find all this far from ideal, but it seems there is not much what I can do about it...

Happy Easter...

Bye Fridger

Posted: 24.03.2005, 17:50
by Paolo
t00fri wrote:For quite a while Chris, Christophe and myself formed a nicely tuned, effective team, ...


It is exactly the situation that was in my mind: a development coordinator + 2 developers (coders/testers).

As you and Christophe has already announced it, we know that you have only a spare time to dedicate now. Chris ?€¦

We know that some tasks are complex and requires extended brainstorming.

The point is that if no one else is invited, involved, and trained to the Celestia Core development it is almost improbable that the situation will change.

Happy Easter to everybody.

Posted: 24.03.2005, 22:08
by Christophe
Well, I actually wonder why we don't have many more contributors. The community seems active in the forums, and astronomy is geeky enough to pick the curiosity of the casual or confirmed coder.

Why does KStars for example, while having a much smaller user base has more contributors than Celestia?

Even considering the fact that Celestia requires an additional OpenGL expertise, this doesn't seem sufficient to explain the gap.

My theory is that the Celestia community while huge compared to that of KStars is mainly on Windows. The visibility of Celestia on Linux is quite low: it's not part of KDE (like KStars is) nor Gnome and it's not included in the major commercial distributions. And the thing is that the proportion of Windows users becoming active contributors is far lower than that of Linux/Unix users, partly because the proportion of coders is simply higher on Linux and also because Windows just doesn't come with a development environment thus requiring an additional effort from the would-be contributors.

What does this give? Well I think the best way to get more contributors is to get a larger community of Linux users. Increasing the popularity of Celestia on Linux is the actual problem and not an easy one.

Posted: 25.03.2005, 02:40
by codrinb
I agree with Christophe here.

There are so many people on the boards and Celestia is such a great application.

Why not allow a larger group of people contribute?

I know that many times quantity is the opposite of quality, but I don't think that is true for open source.

-Codrin

Posted: 25.03.2005, 07:30
by Evil Dr Ganymede
codrinb wrote:I agree with Christophe here.

There are so many people on the boards and Celestia is such a great application.

Why not allow a larger group of people contribute?

I know that many times quantity is the opposite of quality, but I don't think that is true for open source.

-Codrin


Possibly because "too many cooks spoil the broth", as the saying goes? I think something like this needs to be fairly focussed, and having more than about a dozen active contributors to the core will just make it a nightmare to organise. You can see that to an extent with all the addons and textures - different people have different ideas of what they consider good, or realistic. If you're developing something, I would have thought you'd need more active guidance from someone who ultimately has final say in what goes in and out.

I don't know how Celestia's development works, but if Chris is as inactive on that side of things as he has been on these forums lately, then other contributors might simply think that the brains behind it all isn't listening to them.

Posted: 25.03.2005, 08:11
by Christophe
I'd just love for Celestia to have a dozen contributors. At the moment there are less than half a dozen, with only one touching the OpenGL part and doing maybe 90% of the work. We have place for growth.

I don't know how Celestia's development works, but if Chris is as inactive on that side of things as he has been on these forums lately, then other contributors might simply think that the brains behind it all isn't listening to them.


It's not an ego thing, it's just a matter of finding people to do the actual work.

Posted: 25.03.2005, 09:30
by t00fri
Christophe wrote:I'd just love for Celestia to have a dozen contributors. At the moment there are less than half a dozen, with only one touching the OpenGL part and doing maybe 90% of the work. We have place for growth.

I don't know how Celestia's development works, but if Chris is as inactive on that side of things as he has been on these forums lately, then other contributors might simply think that the brains behind it all isn't listening to them.

It's not an ego thing, it's just a matter of finding people to do the actual work.


Christophe,

I think your arguments are valid for GUI work, porting tasks, internationalization, add-on managers,... etc.

These tasks are to some extent quite similar in /many/ applications.
So no very deep insight into Celestia-specific aspects is really required.

But come on, I think I can judge from my professional background that only very few people in this forum would have the necessary background in physics, astro-mechanics, astrophysics, cosmology, planetary science that is required to bring Celestia further /at the front/!

Most of the original developers of Celestia actually do have some kind of physics/mathematics degree, /including Chris/.


And without getting too explicit here, the real problem relies in the following "squeeze":

On the one hand Chris (as creator and lead-developer) wants to keep complete control of which code is committed. In view of his great competence and professionalism this is excellent and the high quality of Celestia serves as perfect justification.

On the other hand, writing many emails is NOT one of his "favorite" activities ;-) . This way long, unpleasant breaks in the /necessary/ communication among the various developers and himself arise very often. This gives rise to a lot of frustration. At some point people just prefer to walk away...Increasing the number of developers just aggravates this essential problem!

As emphasized above, Celestia development worked excellently and efficiently some years ago, when we were a modest number of well-tuned developers with complementary expertise. Communication among us was regular and efficient.

Bye Fridger

Posted: 25.03.2005, 11:36
by dirkpitt
t00fri wrote:On the one hand Chris (as creator and lead-developer) wants to keep complete control of which code is committed. In view of his great competence and professionalism this is excellent and the high quality of Celestia serves as perfect justification.

On the other hand, writing many emails is NOT one of his "favorite" activities ;-) .


I agree, writing emails and forum posts take a lot of time. Perhaps IRC might be
a more efficient communication channel? I know Nasa's World Wind project
has had a lot of success with using IRC for communication between developers.

Posted: 25.03.2005, 13:10
by Christophe
I don't think IRC would work that well for us considering the geographical location of current contributors.

Posted: 25.03.2005, 13:24
by Christophe
On the one hand Chris (as creator and lead-developer) wants to keep complete control of which code is committed. In view of his great competence and professionalism this is excellent and the high quality of Celestia serves as perfect justification.


Fridger, I don't think Chris is such a control freak, I have yet to see him reverting a commit, or taking back cvs write access.

Okay, he often doesn't answer to emails in a 'timely' fashion, but he's hardly the only one to blame here. When a patch is submitted to the list or the forum there is no reason why it should be left to him and only him to make an answer, if only to aknowledge the patch.

The thing is that we're not getting that many patches to begin with, and that has little to do with the way Chris handles his email.

Posted: 25.03.2005, 13:57
by t00fri
Christophe wrote:
On the one hand Chris (as creator and lead-developer) wants to keep complete control of which code is committed. In view of his great competence and professionalism this is excellent and the high quality of Celestia serves as perfect justification.

Fridger, I don't think Chris is such a control freak, I have yet to see him reverting a commit, or taking back cvs write access.

Okay, he often doesn't answer to emails in a 'timely' fashion, but he's hardly the only one to blame here. When a patch is submitted to the list or the forum there is no reason why it should be left to him and only him to make an answer, if only to aknowledge the patch.

The thing is that we're not getting that many patches to begin with, and that has little to do with the way Chris handles his email.


Christophe,

at the heart of this discussion are not little patches! Of course here you are right, the two of us have often enough handled such things between us.

I am exclusively concerned with MAJOR coding steps in the CORE of Celestia. Structural issues in the code etc that we really need to move ahead significantly.

In all these cases Chris has very well taken influence /beforehand/. And also the checkin (not the patches!) ONLY happened /after/ he gave green light.

I can easily trace every single one, since I have retained /ALL/ mails exchanged among us since 3 years...

Personally, I am not very much interested in a development style, that proceeds with a little (uncoordinated) patch here and there. That's how things go right now. 1 little thing/month or so.

We have so many overdue and very significant tasks to add in since almost 2 years!

Just a few examples:

1) Precession of the equinoxes.
2) Stars with the same coordinates in stars.dat and in an stc do not plot in the same position.
3) The Earth is 200AU from the coordinate origin, which introduces huge parallax effects for nearby stars, but this has not been rectified because it would "break cel URLs".
4) A number of custom orbits are still quite badly in error.
5) We have no custom orbits for a number of major moons for which suitable formulae are available.
6) Node and pericentre precession is not implemented.
7) We can't select a view by sky coordinates and corresponding grids in any of the standard astronomical frames.
8.) After all this work on multiple star 3d graphics by Chris and my extensive orbit data preparations, we still don't have a decent multiple-star browser GUI, such that people can actually localize all the thousands of binaries!
9) many further much discussed issues like "filters", "cosmo-Celestia", modified "goto" for binary systems etc.
10) improved manoevering on the surface of objects

+ .....

Bye Fridger

Posted: 25.03.2005, 14:37
by Harry
t00fri wrote:I am exclusively concerned with MAJOR coding steps in the CORE of Celestia. Structural issues in the code etc that we really need to move ahead significantly.
I'd like to emphasize that for a lof of people it's not these major steps which matter, but the polishing - i.e. easily available documentation and a nice out-of-the-box experience. A lot of users don't really care about inaccuracies in the data. For those users it's more important to have a lot of stuff to investigate without going through the additional step of finding, downloading and installing Addons. I am not saying that scientific accuracy should be traded for more fancy stuff, but that when you can have both there is no reason not to include the latter.

Examples:
- IMHO some or all of the Messier objects should be included in the default Celestia installation. This doesn't necessarily have to include textures, it's just important to show that the universe isn't filled with just a couple of stars.
- how about more surface locations on Earth? The data is available and is AFAIK accurate (is this true?)... We just would have to trim it if we want to save space.
- why not add a few scripts and make it easier to access them? They hardly take any space and provide real additional value for any newcomer. We just have to decide which ones...

This is not that much work, but it simply isn't happening. Maybe there is something wrong with these ideas, but I believe it's simply because there isn't anybody around who takes care of stuff like that, and lack of feedback is turning down everybody who's going to try...

Of course this has nothing to do with the improvements you wrote about, I am just pointing out often the small bits are neglected.

3) The Earth is 200AU from the coordinate origin, which introduces huge parallax effects for nearby stars, but this has not been rectified because it would "break cel URLs".

Apparently the cel URLs are going to be broken anyway, so this may be the right time to do it... And IMHO they should be broken in a way that Celestia doesn't even accept the old URLs anymore (it should display an error message explaining that old-style URLs are no longer supported), unless a compatibility mode can be provided.

Another thing which IMHO needs to be fixed:
- the 16kly limit for stars.

Harald