Celestia 1.7.0 Development Thread

[pirogronian]

@Janus

alpha will get you a match on 'α', then cen narrows it to A:B:C, hit 'A' leaving you with 'α Cen A' showing as a match.
When you hit enter to select it, nothing happens.

As far as I remember, selecting completion items is done by hitting TAB, not Enter. Mayby previous implementation allowed for ignoring case comparisions. It also can be done now.

Edit: I just commited small "fix". It just takes first completion, if entered text doesn't match to anything.

[Janus]

Speaking strictly for myself, 'case' is the single largest P.I.T.A. in char~string handling.
Adding utf/unicode/mbcs/??? only makes it worse.

I personally dislike {And more really strong cuss words} case sensitivity in variables, file names, etc.
I have no end of problems with it, and would not miss it if it were to disappear.
At the same time, I freely acknowledge that there are times and places where it is appropriate, just not in the above.

In this case, pardon the pun, it is not just case sensitivity, but equivalency that is the problem.
The text search/match/display routines need to be shorthand aware.
'Alpha' needs to be the same as 'alpha', ALF, 'α', & ALPHA, transparently.
The same with the rest of whatever shorthand is relevant.

I personally prefer not to see the greek shorthand, 'α' and its cousins are frustrating.
They require me to translate in my head while reading, very distracting.
It breaks the thought flow I use, though it may be a part of other people's.
Perhaps a switch to go between shorthand, and plain text.

You might also want to implement a utf/unicode/mbcs compatible input so people can type those awkward messes in as well.
Right now you can't even copy/paste them.

Just some thoughts.


Janus.

[onetwothree]

"alf cen a", "alpha cen a" or <tab> instead of the last "a" select the proper star with the latest changes in PR #249

Added after 3 minutes 16 seconds:

'Alpha' needs to be the same as 'alpha', ALF, 'α', & ALPHA, transparently.

That's how thing are going now (at least in a search console).

Right now you can't even copy/paste them.

That's a different problem. Contributions are welcome

[john71]

Unfortunately the same bugs are still in the current version:

1.) the Earth is black, I mean it is not illuminated. If I add more illumination, then the darkness disappears, but there is the same level of illumination all around the planet, which is obviously not normal.

2.) a gas giant orbiting a triple star barycenter lost all of its moons, as it happened with the earlier version. They are just gone. No trace of them in the qt exe version. In the simple win exe version the orbits are visible, but the moons with Earth-like clouds and normalmaps are gone, they became invisible. It is strange, because the spaceships are still orbiting the invisible Earth-like moons, so they (the moons) at least exist in the win exe version.

Added after 2 minutes 13 seconds:
By the way, the triple star barycenter orbits another barycenter...

[onetwothree]

How do you remove light sources?

[john71]

Well, I did not remove any of it.

[onetwothree]

But your problem is caused by absence of light sources.

[cartrite]

In the beginning of this thread, there was mention of porting to opengl es. Then to Android. From what I read, a lot of the code that can't be used in opengl es has already been depreciated in opengl 3. Like glbegin/end, everything in between, etc. Are there any plans on how this will be done? And preserving desktop versions?

[pirogronian]

@cartrite
In short: probably no. As far as I know, there is only onetwothree, who maintains this project in general and he is rather busy. I perform some crucial internal changes, but have no experience with opengl nor with Android. Others are focused ather on data content, not source code. So, currently there is no one who can work on it. Contributors are welcome.

[cartrite]

I don't know much about it either. But from what I am reading, it looks like a nightmare.

Added after 1 minute 2 seconds:
Hope I'm wrong.

[onetwothree]

Android is a "good to have" port, not a necessary one:

When OpenGL ES is ready port and we have enough human power and enthusiasm then we'll start porting for Android.

But OpenGL ES 2 is a must. Basically it's very close to OpenGL 2.1, so when we have OpenGL 2.1 code without any deprecated features it will be trivial to add an ES version. Anyway I don't think that this will be done in 1.7 cycle. But we need to remove immediate mode because today its support in some video drivers is in bad shape (at least with mesa's radeonsi Celestia (both native and under wine) is ~60% slower than with proprietary amd driver on win7 (40 vs 70fps), my cpu was mostly busy with memory copying. Modern engines prefer glsl pipelines so new drivers don't care about optimizations for the fixed pipeline.

GL ES has less features than its desktop counterpart so the latter won't be removed in favour of the former, don't worry (but actually you can use GL ES on desktop).

I have an idea to extract a render code from celengine so in future it will be possible to have several render libraries, e.g. GL/GLES, Vulkan, or even based on a 3rd party 3D engine.

[Janus]

For anyone who cares, here is commit 5441
Largest change is a better star chooser, it understands greek shorthand much better now.
They made a large improvement in it.

As usual for me, VS2015, static linked, x86 & x64 exes only.

Albedo deprecation warning in log files, deprecated, old style albedo silently defaults to geomalbedo properly.
For amusement value, I set octree to 100GLy, which is larger than the observed universe.

If anyone wants any of my other recent tweaks added in, PM me, no need to pollute this thread with side noise.
I can make my own thread in parallel with this one if enough people want them.


Janus.

[Lafuente_Astronomy]

or anyone who cares, here is commit 5441
Largest change is a better star chooser, it understands greek shorthand much better now.
They made a large improvement in it.

As usual for me, VS2015, static linked, x86 & x64 exes only.

Albedo deprecation warning in log files, deprecated, old style albedo silently defaults to geomalbedo properly.
For amusement value, I set octree to 100GLy, which is larger than the observed universe.

If anyone wants any of my other recent tweaks added in, PM me, no need to pollute this thread with side noise.
I can make my own thread in parallel with this one if enough people want them.


Janus.

Woah! Earlier yesterday, a member of the forum claimed he was able to allow stars at 45 GLY. And today, now we have a 100 GLY limit? Nice! And just to make sure constellation lines work in this commit, I'll go for a test run

Added after 9 minutes 16 seconds:
Well, I did a test run and you really did it! Thanks to the Star chooser changes, it now allows Constellation Diagrams to be shown once again. For comparision, I will show these images:
Before:

After:

And with all those other features you included from other previous forks, especially the increasing of the allowable star distance to 100 GLY I gotta say, you're getting better and better my friend! Keep up the good work!

Added after 4 minutes 39 seconds:
And I agree with you. 100 GLY is equivalent to around 30 Billion Parsecs. Truly greater than the whole Observable Universe indeed.

[Janus]

Ok, here is some more fun.
In light of better text handling, I decided to bring another toy forward.
This is 5441 with just the stuff mentioned before, none of my other toys, except this new one.

When you select something besides earth, the Ra Dec is given based on the simulation clock.
Mostly this will mean real time, though if you change the clock in the program, you will change the displayed RaDec to the value for the new date/Time.

So now you have a tool that will show and tell you where a star, or whatever, is, right now, or at some time.
I suppose you could try to use it for matching star patterns in pictures as well.
Make a script that changes the data/time one day at a time when focused on something, then watch everything around it move.

I hope someone out there enjoys.


Janus.

EDIT: superseded, see #258 for updated version.

[Lafuente_Astronomy]

This is 5441 with just the stuff mentioned before, none of my other toys, except this new one.

What do you specifically mean by "other toys"? Just curious

[Janus]

No betwixt, betwixdex, benear, etc.
As per post#25 in this thread, plus the RaDec thing.

I am considering a forking development thread, just for fun.


Janus.

[Lafuente_Astronomy]

Well, having the RA and Dec for stars is a must in all planetarium apps, and I thank you for that!

I am considering a forking development thread, just for fun.

Looking forward to it, honestly and seriously

[Janus]

And here is the next little tweak.

This is the previous version with one extra feature.
In addition to Ra & Dec, it also shows Dist.
The Dist value uses the same routine as the observer Dist, so it should be familiar.


Janus.

EDIT: Updated, please see #262

[Lafuente_Astronomy]

Well, there's already a distance info for all stars in Celestia. But you may have a reason as to why you included a Distance info.

[Janus]

The dist info normally shown is from the observer.
The dist I added is for how far from earth.
It uses the same routine as observer for adjusting units of distance, DistanceLyToStr.

For example, select alpha centauri, goto it, then zoom out.
The dist value shown in the upper left will change.
The dist value I just added however, will not.

In one of my forks I moved the observer dist to the lower right, much more natural there.
I also moved the text in from the edges to make it more readable on large screen and projectors.
I myself have always found having the observer dist in the upper left confusing.
Sorry, just rambling while waiting on a debug dump to finish.


Janus.