Celestia Forums

Chris, would you be kind enough to do a point of the actual state of Celestia?
I ask that because since some weeks (perhaps months) I have the feeling the dev is quite wimpy... (without offending, perhaps I'm all wrong after all)
What are the actual challenges, what are the coming ones?
What about Qt Celestia?
Etc...

ElChristou wrote:Chris, would you be kind enough to do a point of the actual state of Celestia?
I ask that because since some weeks (perhaps months) I have the feeling the dev is quite wimpy... (without offending, perhaps I'm all wrong after all)
What are the actual challenges, what are the coming ones?
What about Qt Celestia?

I've been busy with development of ESA's STA tool, so I've had limited time to work on Celestia lately. But, I'm not the only Celestia developer. Andrew has spent a lot of time working on updating stars.dat. This new star database is an important contribution for 1.6.0, introducing a great deal more rigor in the processing of the raw HIPPARCOS data. Recently, I've been modifying the solar system parser and data structure to allow multiple names for solar system bodies (as is possible now with stars and deep sky objects.) The first patch for this should be ready today.

You can see 1.6.0 feature development status here:
http://en.wikibooks.org/wiki/Celestia/160Features
Note that not all of these features will necessarily appear in 1.6.0; there needs to be a discussion among the developers about

I'd say that most of the major coding for 1.6.0 features is done. The work that remains is mainly bug fixes, data file updates, and documentation changes. There are some UI adjustments that need to be made as well: keys assigned for the new grid types, reassignment of the D key to do something other than play the demo, making sure that the new features are properly exposed via the GUI, etc.

I intend to make a Qt4 preview release available for all platforms this summer. Although I've been using the Qt4 version of Celestia every day (much more often than the Windows and Mac-specific GUIs), there is still a lot of work to be done on the UI.

As for future development of Celestia (after 1.6.0), I feel that possibly the most important task is to create an add-on manager. It is simply too difficult to install and manage add-ons for Celestia. With the large number of add-ons available, it becomes important to be able to enable and disable individual add-ons without restarting Celestia; otherwise, even the most capable computer becomes overwhelmed when everything thing is loaded at once. There has been a lot of discussion about creating an add-on manager; Christophe T even created a forum dedicated to the discussion of the topic. We should develop these ideas further and finally code a proper UI for installing, enabling, and interacting with add-ons.

--Chris

Here is my view of the same situation...

Presumably all this is due to the summer "black hole"...

Chris wrote:But, I'm not the only Celestia developer. Andrew has spent a lot of time...

Indeed, apart from Chris, it seems Andrew is currently the ONLY remaining active (and at the same time "youngest") developer .

Remaining Developers
==================

• Vincent had his last post > 1 month ago
  
• Dirkpitt (aka DW) had his last post in the forum March 29 (!) and in the dev list May 05 , i.e. a long long time ago! . His nice HDR code is still "experimental", i.e. not brought to a "happy end" ...
  
• Christophe Teyssier had his last post March 25/27(!), both in the forum and the dev list, i.e. > 3.5 months ago!
  I am actually a bit worried...
  
• Toti has completely vanished since many months. His email is also bounced back... (here I am worried, too)
  
• Pat Suwalski usually has only brief "reappearances" in the dev list, when it comes to uploading a new Celestia version . The last one was May 05.
• Hank: well, after trying in vain to revitalize Celestia development with people who mostly don't know how to code, and after being designated by Chris L as his "backup project admin",

  Chris wrote:Agreed. It's an essential part of making Celestia development less bottlenecked by me (though I have no plans to leave the project). Would you be up for it, Hank?
  

  vanished completely from the scene a long long time ago...
  
• Grant Hutchison: despite his official "dismissal", he appears to be among the most "steady" devs . His own recent description reads like so:
  

  granthutchison wrote:I've long since given up cluttering the developers' list with my updates to the data files. I restrict myself to forum announcements concerning updates that people might find interesting:...
  

• myself: notably since Chris L resigned his NVIDIA job and started to work fulltime on Celestia about one year ago, it became increasingly obvious that Chris' priority list and mine differ a lot! So oscillating between being demotivated and boored, I run back my public dev activities a lot and coded a number of features for my own pleasure. After my recent preview of my globular cluster code, Chris L had so many proposals to make (without having seen my code), that I decided to put my code into the "fridge" for a while ...

Testers:
==========

• Selden: somewhat less frequently visible, since involved in complex Celestia add-ons . But basically "reliable" as ever...
  
• Cham: much less active since his return from India.
  
• ElChristou: busy at the French riviera, yet asking "piercing questions"
  
• ...

Did I get the picture wrong somewhere??

Fridger

Chris,

please, don't forget about the planetographic grid. Currently, it's uncomplete, as we already discussed by email, some weeks ago.

Tx Chris!
Tx also Fridger for this portrayal of the dev team. Indeed the situation seems to be... not really cool... Hope this is only the summer gap...

ElChristou wrote:Tx Chris!
Tx also Fridger for this portrayal of the dev team. Indeed the situation seems to be... not really cool... Hope this is only the summer gap...

My name may also be included in order to do whatever it is possible for me to do. As Christophe (ElChristou) knows, I have tried to help in a number of areas.

By the way, Christophe, I have received still absolutely nothing from anyone else at NASA. Apparently, they don't give a damn if the people who pay their salaries would like answers to questions.
I did see, a television program of the Apollo 7 mission, however, which displays very accurately, the details we are trying to find for your SLA panels. Should be a way to convert that program into something from which the details can be extracted, so I'm currently looking into your project from this angle now.

If you people find mundane tasks which can be done by some Brain-Dead Bozo, please feel free to give me a shot at these kinds of tasks. Organizing, testing, documentation, packaging, or whatever else you would prefer not to have to do yourselves. I don't mind this at all, and would gladly be willing to do what ever I can do in order to further development from those of you who are obviously brighter than I am.

The Motherlode now has to have every add-on installed, tested, and approved by myself before it gets placed into the catalog. This is not a very glamorous task, and I have managed to piss some folks off, by requiring things like Read Me files which explain how to install the new add-on, and instructions pertaining to how the add-on should be used and/or located. This is something I am really fairly good at, and you may have noticed that the overall quality of that site has now been somewhat improved over the last six months. It is still not perfect, but we are trying very hard to improve the quality of the site.

If a potential developer would like me to apply these skills to new parts of the Celestia package, I would be happy to volunteer for these tasks also. And, just in case you haven't noticed, I sit here in front of my PC on a daily basis for anywhere from 10 to 14 hours a day now since my health lets me do little else. Just FYI...

Thanks, Bob

Thanks, Bob

Tx Bob, despite not said enough, I'm sure people here DO appreciate your efforts!

My name may also be included in order to do whatever it is possible for me to do.

Of course, Bob, your help is very much appreciated by everyone! So you count among the last few active testers. Note that my above post had a certain message to convey and referred mainly to people with formal Celestia developer status ...
http://sourceforge.net/project/memberli ... p_id=21302

Fridger

Hmmm... so the active developers are heavily biased towards data file handling rather than coding...

It is unfortunate that the globular clusters have been put on hold: I would have thought this would be a high priority.

What would be helpful would be to have some kind of specified date after which new features do not get added to the upcoming version: right now the timescale seems fairly arbitrary and it is not clear whether a feature will be accepted into the upcoming version or postponed until later, nor is it clear what the features that are regarded as critical to the new version are.

ajtribick wrote:It is unfortunate that the globular clusters have been put on hold: I would have thought this would be a high priority.

In my personal version of Celestia, my globular clusters code is fully operative since quite a few months (including the corresponding data base) . I indicated the reason why I have decided to freeze the project from my part...

Fridger

t00fri wrote:

ajtribick wrote:It is unfortunate that the globular clusters have been put on hold: I would have thought this would be a high priority.

In my personal version of Celestia, my globular clusters code is fully operative since quite a few months (including the corresponding data base) . I indicated the reason why I have decided to freeze the project from my part...

Fridger

Aaaaaaarrrrrgggghhhh...

So we can regard globular clusters as a "not in the foreseeable future" issue. Because after all, there's no reason for anyone else to do any work on them when there's already a working implementation out there (after all, you can come along and say that you've already done it). What exactly do you want from Chris Laurel before you'd be willing to share the code for the globular clusters implementation?

I mean, what's wrong with Chris Laurel coming along and trying to discuss a bunch of rendering issues? After all, without the code there he has no way of knowing how you've managed to do things, and with the HDR that was being developed, I don't see why it would be unexpected?

What's the point of Celestia development if the first thing we do when we get asked questions is to decide to freeze the project?

t00fri wrote:[*]Vincent had his last post > 1 month ago

Indeed, I've been very busy professionally these last months.
But I still have a few changes about 'ridiculous details' waiting to be checked in for the Qt4 version , plus some patches proposals here and there in the SF tracker: https://sourceforge.net/tracker/index.p ... tid=371302

Then, as you know, my scope is mainly centered on adding scripting methods related to important new features.
So I'm just waiting for your globular clusters code to get back working on Celestia...

ajtribick wrote:So we can regard globular clusters as a "not in the foreseeable future" issue. Because after all, there's no reason for anyone else to do any work on them when there's already a working implementation out there (after all, you can come along and say that you've already done it). What exactly do you want from Chris Laurel before you'd be willing to share the code for the globular clusters implementation?

I presented a preview of my globular cluster project quite some time ago, when it was largely completed from my point of view after spending MANY thoughts and plenty of time with it. In principle my plan was to release the corresponding patch /immediately/ after the preview for general testing.

I mean, what's wrong with Chris Laurel coming along and trying to discuss a bunch of rendering issues? After all, without the code there he has no way of knowing how you've managed to do things, and with the HDR that was being developed, I don't see why it would be unexpected?

Being a very experienced scientist and Celestia developer, it should have been understood that much of the long list of OBVIOUS proposals for improvement generated by Chris BEFORE even seing my code, were long examined (and mostly discarded) by myself already. I don't appreciate such a /superficial/ "peer reviewing" attitude, before the code has even been released for testing and before Chris was even aware of the complex underlying theoretical modelling constraints (Ivan King...) .

It would just have been fine to write instead at the time of the preview:
looks good & looking forward to your patch...

What's the point of Celestia development if the first thing we do when we get asked questions is to decide to freeze the project?

Questions would have been great, but instead I got a long lecture by Chris about how to do these things "right", without him being aware of the tricky underlying 3d - 2d modelling constraints that certainly exclude the /naive/ blending approach he was lecturing about to me...

Throughout my (long) scientific career I have ALWAYS subject my professional work to competent peer reviewing. So I have NO problems whatsoever with this. Actually, I am a "glowing" defender of this procedure as long as it has a competent basis!

But if Chris thinks he knows everything better from the beginning, then such things tend to happen... After that kind of reaction by Chris I just felt demotivated to proceed, got desinterested and turned to other activities instead.

Fridger

t00fri wrote:...But if Chris thinks he knows everything better from the beginning, then such things tend to happen... After that kind of reaction by Chris I just felt demotivated to proceed, got desinterested and turned to other activities instead.

Fridger, don't be so strict, you know very well that (perhaps not you but) in general one can post/talk too quickly resulting in a strange "flavor". This should not be the cause of such demotivation. In such case a bit more of communication should dismiss such bad "flavor" and this way keep the working flow open. We should try to by pass such small problems and go ahead because it's not worthy of the level of this project! Courage!

Vincent wrote:

t00fri wrote:[*]Vincent had his last post > 1 month ago

Indeed, I've been very busy professionally these last months.
But I still have a few changes about 'ridiculous details' waiting to be checked in for the Qt4 version , plus some patches proposals here and there in the SF tracker: https://sourceforge.net/tracker/index.p ... tid=371302

Then, as you know, my scope is mainly centered on adding scripting methods related to important new features.
So I'm just waiting for your globular clusters code to get back working on Celestia...

Aha! Welcome back, Vincent . One remaining question might still be whether practically ALL previously active, coding developers (apart from Chris) are "very busy recently", or whether there is another reason for their coincident "retirement" ...

Let's hope it's the former reason

Fridger

t00fri wrote:

ajtribick wrote:So we can regard globular clusters as a "not in the foreseeable future" issue. Because after all, there's no reason for anyone else to do any work on them when there's already a working implementation out there (after all, you can come along and say that you've already done it). What exactly do you want from Chris Laurel before you'd be willing to share the code for the globular clusters implementation?

I presented a preview of my globular cluster project quite some time ago, when it was largely completed from my point of view after spending MANY thoughts and plenty of time with it. In principle my plan was to release the corresponding patch /immediately/ after the preview for general testing.

I mean, what's wrong with Chris Laurel coming along and trying to discuss a bunch of rendering issues? After all, without the code there he has no way of knowing how you've managed to do things, and with the HDR that was being developed, I don't see why it would be unexpected?

Being a very experienced scientist and Celestia developer, it should have been understood that much of the long list of OBVIOUS proposals for improvement generated by Chris BEFORE even seing my code, were long examined (and mostly discarded) by myself already. I don't appreciate such a /superficial/ "peer reviewing" attitude, before the code has even been released for testing and before Chris was even aware of the complex underlying theoretical modelling constraints (Ivan King...) .

It would just have been fine to write instead at the time of the preview:
looks good & looking forward to your patch...

What's the point of Celestia development if the first thing we do when we get asked questions is to decide to freeze the project?

Questions would have been great, but instead I got a long lecture by Chris about how to do these things "right", without him being aware of the tricky underlying 3d - 2d modelling constraints that certainly exclude the /naive/ blending approach he was lecturing about to me...

Throughout my (long) scientific career I have ALWAYS subject my professional work to competent peer reviewing. So I have NO problems whatsoever with this. Actually, I am a "glowing" defender of this procedure as long as it has a competent basis!

But if Chris thinks he knows everything better from the beginning, then such things tend to happen... After that kind of reaction by Chris I just felt demotivated to proceed, got desinterested and turned to other activities instead.

This is ridiculous. I raised some legitimate concerns about using sprites to render globular clusters at small apparent sizes. I absolutely do not believe that I've got everything figured out from the beginning. But, I've been dealing with similar additive blending related problems throughout the seven years I've been working on Celestia, so I brought up an issue that is largely independent of the exact distribution of stars in a cluster. If that does constitute competent peer reviewing, I don't know what does. The 'lecture' that you refer to was in reponse to your own request for me to clarify my proposed technique. You had problems with it and that's fine--I'm ok being wrong, and I'm not expecting you to adopt my approach. But I do expect any Celestia developer to respond constructively to feedback from other people on the team. Being an experienced physicist doesn't exempt you from that. And pouting in the corner with your code doesn't help anyone. It would be much more useful if you just shared your globular cluster patch so that it could be reviewed and incorporated into Celestia.

--Chris

chris wrote:But, I've been dealing with similar additive blending related problems throughout the seven years I've been working on Celestia, so I brought up an issue that is largely independent of the exact distribution of stars in a cluster. If that does constitute competent peer reviewing, I don't know what does.

You did start right away with proposing how to do things 'better', although you obviously were unaware of King's definitions of cluster parameters that are to be read out from Harris' catalog to specify the cluster morphology.

The main point which that 2d textbook alpha blending argument of yours missed out is that we have to switch between a proper 3d definition and the corresponding 2d skyplane projection. That makes things "non-local" via the non-trivial integration involved, rendering the proper matching between a resolved star-sprite display and an unresolved central "soft blob" thoroughly nontrivial.

The argument which you then laid out at length

Chris wrote:I think the problem can be simplified quite a bit if we just use a single circular (or ellipsoidal, if we get eventually get orientation information) 'cloud' sprite with a luminosity profile matching the projection of the theoretical distribution. There are several ways to do this--a dynamically generated texture, a pixel shader, or simply tessellated circle geometry.
...

completely ignores precisely that non-trivial issue!

So when you absolutely have to make proposals for improvement of my approach before the code is even released, then I think one should expect that you are at least aware of that basic underlying 3d-2d problematics related to King's parameter definitions...

Chris wrote:The 'lecture' that you refer to was in reponse to your own request for me to clarify my proposed technique. You had problems with it and that's fine--

Of course, I had no problems (REALLY! ) with what you were saying, I just wanted to be sure that your argument was indeed ignoring the most crucial point that made me dismiss that approach, quite some time earlier...

Despite all your writing, don't you agree that it's kind of odd from your side to embark into pages and pages of arguments, before having seen what was done and notably before knowing the literature that formed the basis of my approach?? I am certainly ready for criticism, but after having spent a good deal of time with the stuff and knowing what I am talking about, I do expect that the criticising person is sufficiently prepared...

Fridger

t00fri wrote:

chris wrote:But, I've been dealing with similar additive blending related problems throughout the seven years I've been working on Celestia, so I brought up an issue that is largely independent of the exact distribution of stars in a cluster. If that does constitute competent peer reviewing, I don't know what does.

You did start right away with proposing how to do things 'better', although you obviously were unaware of King's definitions of cluster parameters that are to be read out from Harris' catalog to specify the cluster morphology.

The main point which that 2d textbook alpha blending argument of yours missed out is that we have to switch between a proper 3d definition and the corresponding 2d skyplane projection. That makes things "non-local" via the non-trivial integration involved, rendering the proper matching between a resolved star-sprite display and an unresolved central "soft blob" thoroughly nontrivial.

The argument which you then laid out at length

Chris wrote:I think the problem can be simplified quite a bit if we just use a single circular (or ellipsoidal, if we get eventually get orientation information) 'cloud' sprite with a luminosity profile matching the projection of the theoretical distribution. There are several ways to do this--a dynamically generated texture, a pixel shader, or simply tessellated circle geometry.
...

completely ignores precisely that non-trivial issue!

So when you absolutely have to make proposals for improvement of my approach before the code is even released, then I think one should expect that you are at least aware of that basic underlying 3d-2d problematics related to King's parameter definitions...

Chris wrote:The 'lecture' that you refer to was in reponse to your own request for me to clarify my proposed technique. You had problems with it and that's fine--

Of course, I had no problems (REALLY! ) with what you were saying, I just wanted to be sure that your argument was indeed ignoring the most crucial point that made me dismiss that approach, quite some time earlier...

Despite all your writing, don't you agree that it's kind of odd from your side to embark into pages and pages of arguments, before having seen what was done and notably before knowing the literature that formed the basis of my approach??

I don't think it is odd at all. As I said, I've seen similar issues crop up before--it's a potential problem any time you are potentially rendering a bunch of sprites in one place. The specifics of the King parameters don't even matter: somehow you have to deal with the case when lots sprites pile up on the same pixel.

I am certainly ready for criticism, but after having spent a good deal of time with the stuff and knowing what I am talking about, I do expect that the criticising person is sufficiently prepared...

This is insulting. Of course I was prepared; my method may not be perfect, but given the constraints we have to deal with (8-bit per channel frame buffers, etc.), it's likely good enough. But if you've got a globular cluster rendering working and the clusters look good at all sizes and distances, then just share the code patch already. Actually, even if the rendering isn't perfect, you should still share the code. Or if you're unwilling to participate in the development of globular clusters, then say so, and we'll get the task assigned to others.

--Chris

chris wrote:I don't think it is odd at all. As I said, I've seen similar issues crop up before--it's a potential problem any time you are potentially rendering a bunch of sprites in one place. The specifics of the King parameters don't even matter: somehow you have to deal with the case when lots sprites pile up on the same pixel.

You somehow seem to be forgetting that I am fumbling with Celestia's code just half a year less than you do...(although you have been doing so much more than me ). I sure know about the pixel pile up problem and soon or later there will be the proper improvements found. But --my God-- this is NOT the type of corrections to address as a first comment in a preview!

In particular, what you were telling me was just NOT the solution of that familiar issue!

Chris wrote:

I am certainly ready for criticism, but after having spent a good deal of time with the stuff and knowing what I am talking about, I do expect that the criticising person is sufficiently prepared...

This is insulting. Of course I was prepared; my method may not be perfect, but given the constraints we have to deal with (8-bit per channel frame buffers, etc.), it's likely good enough.

It was not meant to be insulting. Sorry if it was. Yet, you were certainly prepared concerning your OWN argument, but it was ALSO obvious that you were unaware about the crucial point related to King's cluster modelling: The statistical star sprite generation needs to be done in 3d according to the 3d King distribution, while your type of matching refers to the 2d skyplane. Hence the matching requires to project the generated 3d distribution (depending on King's standard parameters) to 2d via a nasty integral. I have never read a word from you about this issue... So...

But if you've got a globular cluster rendering working and the clusters look good at all sizes and distances, then just share the code patch already. Actually, even if the rendering isn't perfect, you should still share the code. Or if you're unwilling to participate in the development of globular clusters, then say so, and we'll get the task assigned to others.

That's not quite the right tone I would say...Under such premises you are certainly free to assign the development of globular clusters to others, if you prefer... Then this was it, I suppose.

Fridger

You know...
I absolutely love both of you gentlemen, but wouldn't we all be better served if these heady discussions were done via PM's?
I only understand about 1/10th of what you are saying here, but I am certainly not immune to the notes of discord and lack of harmony prevailing here.

Come on guys, you are staining my Universe.
Let's have a bit of cheer and start over somewhere else, eh? Hell, send me your addresses via PM's, and I'll personally buy you BOTH a bottle of your favorite
noxious drinking liquid. Remember that we ALL love Celestia. Do we not?

I'll shut up now, but please keep in mind my offer. Hell, take a vacation for a while.