Celestia 1.7.0 Development Thread

[onetwothree]

Hi celestians,

Some time ago I decided to created another fork of Celestia because I don't see that Alexell's fork has any future.

Here're the reasons:

* Code quality. Alexell merges random patches from random people without any attempt to check their quality and even necessity. Sometimes it's so low then (pardon me please) better call it crap. Alexell's own code is also terrible. I can show examples if somebody wants proves.
* But the latter doesn't matter because he disappear since the 4th of August. And he's the only person with merge privileges so for 3 moths there is no activity in the main branch. We have 16 pull requests, some are useless or unfinished but 3 of them are important bugfixes, it's absolutely unacceptable to call yourself "a project lead" and ignore your duty.
* Random patches from random people is the only development model. If you want your project to succeed you must have a project vision at least midterm.

So that I started with original codebase, but I imported it properly so we can see original authors' names and emails, we have old branches alive (for 1.6.1 for example). Then I imported useful patches from Alexell's repository fixing them if need. Besides incorrect patches I did not import controversial changes (audio support, new data catalogs structure).

My midterm plans for user visible changes are:
* Modern font rendering. No more bitmap fonts, only TrueType! Support for RTL languages (hebrew, arab, persian). In progrerss.
* Not only Qt5 but other UI will be supported.
* Better star rendering. Help needed.
* Make interface more friendly to languages other than English. In progrerss.
* Plug-ins. We really should not put everything into the core. Such features like audio support or scripting languages should be implemented as plug-ins.
* Remove old rendering paths. 2018th is near its end, so why should we support ancient OpenGL 1 or ARB vertex programs? Even Radeon r300 cards manufactured 13 years ago support OpenGL 2.1. So let's remove anything but OpenGL 2.1. Done.
* Port Celestia to OpenGL ES 2.0 so it will run on low power hardware which doesn't have desktop GL (Rasberry PI & other ARM based HW).
* When OpenGL ES is ready port and we have enough human power and enthusiasm then we'll start porting for Android.

Future from a developer view:
* Modernize the codebase C++11. It's much better than previous standards. In progress.
* Remove deprecated OpenGL features (immediate mode, fixed pipeline). We should have only modern GLSL based rendering. In progress.
* Remove autotools (Done) and QMake. Add CMake instead. In progress.
* Build Celestia as a library + small applications with GUI. In progress.

The current state:
* Only OpenGL 2.1 based renderer.
* Old data directories (data, extras, extras-standard). Alexell's change was not accepted by everybody even on this forum. Let's discuss this topic and make a proper solution.
* No audio support, just dummy cel/celx functions. Because in Alexell's fork it's done incorrectly and audio support should be implemented as a plug-in. But if Pirogronian makes a good OpenAL based implementation I'll use it temporary until a have a decision on plug-in architecture.
* Bugfixes and polishing of Qt5 GUI.
* Finished transition to high quality vector math library Eigen started 10 years ago.


How to help.
If you're a developer then join an organization on GitHub. I have rather high code quality requirements but I will help newcomers. If you don't have coding skills there are a lot of areas where you can help: help localize user interface, update star catalogs and other databases, create new textures and models, test new builds.

I really need help with a new star texture to implement a better star rendering. I can code but I have no drawing skills. Please contact me I you want to help.

[Joey P.]

Really?

[onetwothree]

Really. Do you want to help with texture drawing?

[Askaniy]

It's really great! I can help with textures and object data.

[onetwothree]

Hi Askaniy!

For stars we need a following texture:
- The texture has just a single 16-bit floating point channel; color is not "baked in" to the texture
- The center is a Moffat distribution; the halos are based on actual astrophotos
- It's nice if it has star flares, but it's optional at least for the 1st version.

If we do it properly then we'll have nice stars like in iOS application Sky Guide being developed by Chris Laurel now.

[Askaniy]

Ok. I will ask some clarifying questions in private messages later.
Please note that English is not my native language.


Added after 1 minute 3 seconds:
Alexell wrote 6 days ago that he completely stops developing the Celestia. Google Translation of his text blog:

Termination of activities in certain areas

Spoiler

Greetings to readers and guests of my blog. Today we will sort through my activities.

All my friends, acquaintances, as well as those who are just familiar with my pseudonym know that I have a lot of different skills: computers, programs, creating music and working with audio, video editing, some Photoshop, the Internet and its device, routers, programming (C ++, PHP), domains, SSL, hosting, creating websites, forums, some experience with Linux servers and much more related to computers and the Internet.

At the time of writing this post, I mostly earn my living by working as a network administrator in an IT company and I understand networks, setting up L2 equipment, telephony, GPON, and so on.

Over the past 3 years I have done a lot of work: I launched my own eternal online service on VKontakte, a service for determining the manufacturer and location of Wi-Fi routers at their MAC address, as well as a convenient IP calculator, several HTML5 games, I had my CS: Source server for half a year He wrote several useful articles on Linux and MikroTik, and also launched his private VPN in France.

Separately, it is worth mentioning that I reached an agreement with the founder and developer of the Celestia space simulator Chris Laurel who left the project, and the project went under my management, the project repository migrated to GitHub, Celestia now has a new modern website, an updated forum design and everyone can help in developing this wonderful programs on GitHub.

I especially do not consider work as my activity, it is only work, and in my free time for the last 3 years I have been torn between all areas of my activity, trying to do everything everywhere, my head constantly buzzed with tension. This summer I realized that this approach is completely unacceptable because it will not allow me to fully reveal myself in this or that sphere. The way out is obvious - to choose for development any one field of activity.

I will be cunning if I say that I am not interested in programming and working on various projects and services. Therefore, to be honest, I am still interested. However, I will have to leave one of the most difficult and demanding projects completely, because its development requires a lot of time and concentration, and I want to focus on another. I had to make a difficult decision - to leave the Celestia project, and also to stop wasting time on creating programs, services and working in any projects. I will still look after my sites and services so that everything works correctly. Sometimes maybe I will even make small changes. But no more than that. As for Celestia, I will still continue to pay for the domain name so that the site with the forum will continue to exist, but I will have to lay down the administrative functions (including on GitHub) to someone else.

Sometimes I will make small videos on my favourite computer games and upload them on my Youtube, but it will be easy for a variety and maintenance of video editing skills.

In addition to all of the above, I previously provided assistance to four novice vocalists. I discontinue cooperation with two of them, and cooperation with two remains open since this sphere is nevertheless close to my main one. All notified in person.

So, there comes a new era of my development. Almost all my free time now I will spend on creating modern electronic music, other musical experiments and the development of my creative skills.

Good mood to you all:)

[pirogronian]

Hi @onetwothree!
Your initiative is very interedting and I would be happy to participate.

I have some questions. What is main reason for plugins? Reducing binary executable size? Lowering memory overhead? Faster startup? Other reason?

And some "visions": my primary goal of Celestia developing is making objects (stars, planets, dso) dynamically loadable/unloadable via scripts. This can be very substantial for making extensions.

But after all, I like solution for extensions of pioneer space game. It loads datafrom user directory prior to system dir data. All user data with same name replaces original one. This eliminate need for messing in system dirs with root priviliges (in Linux).

Please, write what You think anout my propositions.

Ps: i have been noticed recently that @Alexell is willing to give me the Celestia Project administratior privileges. However, after reading this thread, I'm willing to give this honor to You. So, are you willing to take that responsibility?

[Art Blos]

onetwothree < Why did you become active just now, after so many months? Alexell has long time ago spat on the project, and so much time has been lost. It is, of course, great to be found by someone who is ready to improve in the source code. But your unexpected appearance is at the very moment when people did not know about Alexell’s official departure, leads to various suspicions.

[Janus]

It is nice to hear from someone with a plan, and what appears to be the time and knowledge to do something with it.

I gave up on using the current code base when it was decided to start merging QT stuff into the core.
Audio support is something I never understood the need or use for.
It would make more sense to me to be able to call an external module for sound if it is needed.

I am happy to help where I can, but I am not a real C/C++ programmer.
Part of the reason I work on Celestia is learning, I also like to make as many of my own tools as I can.
I use VS2013 for celestia, and am trying to acquire VS2015, I can not use VS2017 as it trashes my system setup.
I lack the knowledge to make much progress beyond that.
Though I have managed to get the support libraries compiling up to VS2013, and most also compiling under Mingw from QT via batch file.
I have not been able to get celestia itself compiled via mingw yet.

You mention other UIs, I would like to suggest wxwidgets.
There is already a wxlua system, which might make a good base.
I am tinkering with making a wxwidgets based GUI for celestia using codeblocks.
In part to get away from VS, in part because it is multiplatform and portable.
I have multiple copies of it on my system, and they remain isolated, which is nice, VS is a mess.
Making a gui from scratch is hard.

I know little about opengl, most of the tutorials are missing what I really need, which is just life.
I have no idea what you mean by C++ 11, I can only hope VS2013 supports it.

I personally avoid cmake, the projects it makes always refer back to itself during compile, which I do not like.

I do not care about the directory setup as I seldom use addons.

The current codebase uses eigen 3.3 if I remember correctly, which I believe is current.

As I said, happy to help if I can, though I doubt I will be much help.
I do not use things like svn or git, so all I will be able to do is offer snippets or modded files.
I maintain my own fork which I use for making pictures and videos like my solar system tour.
If any of that code is useful to you, help yourself, it is offered freely.

Good luck.


Janus.

[onetwothree]

Ok. I will ask some clarifying questions in private messages later.
Please note that English is not my native language.

No problem. I see you are from Russia, I can speak russian.

Alexell wrote 6 days ago that he completely stops developing the Celestia. Google Translation of his text blog:
Termination of activities in certain areas

That's expected. If you're doing many different things you're doing nothing. So it's better to revise activities and leave only important ones.

I have some questions. What is main reason for plugins? Reducing binary executable size? Lowering memory overhead? Faster startup? Other reason?

Other reasons
1) I don't think that every single feature should be in core. Especially rarely used and low quality.
2) Independent development. If you have a bug in your plug-in you fix it and release it. If its functions are in core then you wait until a new release is ready.
3) We can provide compiled plug-ins with different dependencies, i.g. provide celx linked against lua5.1, 5.2, 5.3 or luajit.
4) New features can be developed independently so users can try it early without compiling from git.

And some "visions": my primary goal of Celestia developing is making objects (stars, planets, dso) dynamically loadable/unloadable via scripts. This can be very substantial for making extensions.

Looks good for me

But after all, I like solution for extensions of pioneer space game. It loads datafrom user directory prior to system dir data. All user data with same name replaces original one. This eliminate need for messing in system dirs with root priviliges (in Linux).

I was using env vars for this. But this should be achieved with a proper .celestia.cfg

Ps: i have been noticed recently that @Alexell is willing to give me the Celestia Project administratior privileges. However, after reading this thread, I'm willing to give this honor to You. So, are you willing to take that responsibility?

I'd continue to work with my fork, though it had an intension to share patches with Celestia Project. I also want a different organizational approach. Celestia Project is just an other account which can grant merge privileges to other accounts. I want an organization, so it's known who can merge commits, who are decision makers, etc.

I also prefer to have a linear commit history, so I do "rebase and merge" instead of just "merge pr".

Why did you become active just now, after so many months? Alexell has long time ago spat on the project, and so much time has been lost. It is, of course, great to be found by someone who is ready to improve in the source code. But your unexpected appearance is at the very moment when people did not know about Alexell’s official departure, leads to various suspicions.

Sounds like a conspiracy theory You can check commit dates to see that I'm involved in Celestia development since the May. Actually I decided to start a new independent fork 3 months ago when I saw that Alexell merges pull requests without any understanding what they change.

It is nice to hear from someone with a plan, and what appears to be the time and knowledge to do something with it.

Frankly speaking I have a lot to learn so in my current you won't find anything like "changing the way objects are rendered" or "adding gravity lenses" .

I am happy to help where I can, but I am not a real C/C++ programmer.

Neither me. But if doctors do kernel hacking why can't we develop Celestia?

You mention other UIs, I would like to suggest wxwidgets.
There is already a wxlua system, which might make a good base.
I am tinkering with making a wxwidgets based GUI for celestia using codeblocks.
In part to get away from VS, in part because it is multiplatform and portable.

I familiar with WX only as a user. But if you're going to work on this I'll be happy to merge your work.

I have multiple copies of it on my system, and they remain isolated, which is nice, VS is a mess.

Git with branches and independent workspaces is nice. It's much better the several copies. And if you sync with a remote server you won't lose your work in case of an incident.

I have no idea what you mean by C++ 11, I can only hope VS2013 supports it.
I personally avoid cmake, the projects it makes always refer back to itself during compile, which I do not like.

You can check Wikipedia C++11 has nice features like smart pointers, auto keyword, looping over containers like in Ruby or Python, move semantics, new useful functions in stdlib. VS2013 doesn't support all its features but I don't know whether we use unsupported features as I prefer GNU/Linux.

CMake is crap, better systems exist. But we need something that 1) crossplatform (good bye autotools) 2) doesn't require huge dependencies (good bye QMake, Scons and Meson), 3) has wide community. So only CMake satisfies these requirements.

The current codebase uses eigen 3.3 if I remember correctly, which I believe is current.

There were some places where old library was used.

As I said, happy to help if I can, though I doubt I will be much help.

Even if you just compile for Windows and test new builds you'll help a lot.

I do not use things like svn or git, so all I will be able to do is offer snippets or modded files.
I maintain my own fork which I use for making pictures and videos like my solar system tour.
If any of that code is useful to you, help yourself, it is offered freely.

Git is great. You really should start to use it. It will save a lot of your time.

It would be nice to see what you have done. But before merging it we need to fix video capture, at least theora files are broken. Dunno about avi.

[Art Blos]

Sounds like a conspiracy theory You can check commit dates to see that I'm involved in Celestia development since the May. Actually I decided to start a new independent fork 3 months ago when I saw that Alexell merges pull requests without any understanding what they change.

Ok. Thank you for your reply.

I would like to clarify which language is native for you? Do you understand Russian only through a translator?

[onetwothree]

I would like to clarify which language is native for you? Do you understand Russian only through a translator?

Belarusian. My russian is good as native.

[pirogronian]

I have to give a credit to @Alexell for restoring the whole project and bring new people to it. However, technically, some development decisions were indeed questionable and developement itself a bit chaotic. There is my own code, which was accepted by Alexell, which I now consider as a complete mistake (albeit working). I was already afraid, why @onetwothree/375gnu's PRs remains unmerged and why @Alexell seems to not participate at all in code developement. Moreover, I have to appreciate @375gnu's help my own PRs and I liked his own code changes. To me, @Alexell is a great project patron, while @375gnu is perfect developement administrator/leader.

[Janus]

@onetwothree

Trying to clarify here.
I do not use svn or git or any other of its kind, on purpose.
I work with and on snapshots only.
It is not unusual for me to have 5 or more versions of fix(es) or feature(s) I am working on.
Sync serves no purpose for me, I snapshot whole projects to merge or as a fall back.
The combination of tools I use does not work with versioning tools, nor actually, do I.

In my normal work I am fixing other peoples hardware/code/stuff, not making my own.
I send them the files I had to modify to make it work, and they merge it in unto whatever their local work flow is.
I know I did my job properly when no one can see my work because it blends in.

I am not big project programmer, celestia & explorer++ are the largest I have ever spent any time on.
Though I have worked on larger, it was finding a particular bug, or glitch.
But I have several times been able to pull out just what was needed from dependency hell, then reimplement much smaller only containing what they needed.
I also avoid using a feature until I understand it, though I often copy usage patterns from elsewhere in a project.
I break big things into small ones, then work on the pieces.

As for cmake, I mostly agree, though I would be far less polite than you are.
That is why I am trying to port the VS project over to either codeblocks or codelite, though I prefer the former.
Since CB can run on a PI, or a linux box on android, and I have heard from one person trying to make an android version of it, I am leaning that way heavily.
Being able to use exactly the same build files on multiple platforms would be great.
One of the first things i did was make SLN files for each version of VS I had.
Which is why right now I can compile mine on VS2013 & VS2015, with QT 5.4/6/8/10, with no changes, and it names the exe according to what I used.
My opinion of QT is not worth putting into words, I would have to replace my keyboard after using language like that.

As for versions of Eigen used, I have no idea, I am still learning the math.
The same goes for Opengl, so far all of the tutorials seem to contain everything except an actual direct answer to my question.
It is me though, the way I see things interferes, though it makes fixing things easier.
I want/need discrete examples, yet everyone wants to just keep building one thing on top of another, which buries me.
My problem though, and I will deal with it.

Things I have worked on r modified are marked.

// 18-11-03 Janus for example.
// The original line(s) preserved for comparison.
// And the explanation of what I am trying to accomplish if it is not obvious.

If I am able to help, this is how I work.

You mentioned fixing recording stuff.;
I have a fork that limits frame rate to recording, so scripts still work as long as the render rate stays above the frame rate.
It is what I used to record my solar system tour.
Being able to make mkv or mp4 files instead of just avi woud be great.
Working with an AVI > 4Gb is a pain, I had to tweak/write my own tool to rerender it, not fun.
Lagarith is fast, but uses a lot of room.

I am hoping to have VS2015 available soon, until then, it is just VS2013.
I have dropped VS2010/12 for this because they lack multiplatform compatible methods of going 64-bit.

Before you mention it, I do not use the community versions of VS anything.
I have no account with M$, Google, Github, SF, or any other such place, nor shall I.

I am running W7 64bit, modded to look and work as much like 2K/XP as I can make it.
My opinion of W8 or the garbage/spyware that came after it is on par with my opinion of QT.
Nor do I like the W7 style of menus, most of the features I actually use to navigate were removed, which is why I did not move to it until I could replace them.


Janus.

P.S. Do you know what is going to become of the forum now that Alexell is dropping out?

[Art Blos]

P.S. Do you know what is going to become of the forum now that Alexell is dropping out?

As is clear from his appeal, he will not stop paying for hosting.

[Joey P.]

I can help with things. Since many of the star sizes in Celestia are inaccurate, here is my updated largest stars addon:

[Janus]

@Art Blos

I do not read russian, and had missed the spoiler popup, thanks for the info.


Janus.

[Croc]

@onetwothree & @ pirogronian

1. how will be corrected the flaws and errors of the Celestia interface, outlined in the topic "Search function in the new Celestia 1.7 64 bit ..." posts ## 6-7, 9-15. https://celestia.space/forum/viewtopic.php?p=137577#p137577
2. What new CELX_Lua_Methods will appear in Celestia 1.7. The developers of the Lua Universal Tools GUI (Lua Edu Tools) have unsolved problems.

[onetwothree]

You mentioned fixing recording stuff.;
I have a fork that limits frame rate to recording, so scripts still work as long as the render rate stays above the frame rate.
It is what I used to record my solar system tour.
Being able to make mkv or mp4 files instead of just avi woud be great.

I have tested windows builds, but in GNU\Linux captured theora ogg files don't have keyframes correctly marked:

[ffmpeg/demuxer] ogg: Broken file, keyframe not correctly marked.

Maybe we should think about using ffmpeg on all platforms.

1. how will be corrected the flaws and errors of the Celestia interface, outlined in the topic "Search function in the new Celestia 1.7 64 bit ..." posts ## 6-7, 9-15. viewtopic.php?p=137577#p137577

First I'd prefer to have a more wide discussion on menu layout. Cause now we have your opinion vs original interface author's opinion. It's better to have a consensus among active community rather then change that menus in every new release because other people will be unsatisfied with them. Some of your proposals look good for, with some other I disagree, for some of them I'm indifferent.

Regarding some other proposals:

* "Goto" button in catalogs browser - we have such entry in context menus, so no need to add the button.
* Sorting by name has been implemented.
* Distance display (rounding) has been fixed.
* Sun's apparent magnitude differ - why should it be fixed it you change distance to the Sun?
* Markers doesn't work yet, code for this function was written but it's commented out, we need to figure out what's broken in it.

2. What new CELX_Lua_Methods will appear in Celestia 1.7. The developers of the Lua Universal Tools GUI (Lua Edu Tools) have unsolved problems.

What functions do they need?

[Alexell]

onetwothree, instead of creating another fork, I suggest that you take control of Celestia’s main repository, because I don’t have time to do this anymore.
Requirements for the quality of the code set yourself. All current merge requests are up to you.

If interested, please write to the PM.

Added after 3 minutes 54 seconds:
I completely leave the project and we definitely need a person who will administer the GitHub repository.

You can discuss the development of Celestia on the forum, on GitHub, and fully control the code of Celestia.
If ever Celestia is ready for release or beta release, you can call me to post Celestia on the site.

P.S. Also, the forum administrator will soon be Croc.