Celestia 1.7.0 Development Thread

[vpontin]

Interesting. I think for it happen a full transition to x64 will be obligatory. And other efficiency changes to data file consuming to load so much data.

[Lafuente_Astronomy]

Interesting. I think for it happen a full transition to x64 will be obligatory. And other efficiency changes to data file consuming to load so much data.

Agreed. If I remember, that's one of the aims of developing Celestia 1.7.0. I think in the very first post of this thread, onetwothree said something like that. I'll try to look

Added after 2 minutes 11 seconds:
Actually, he didn't say anything about transitioning the program to x64. But I do think that's an objective

[cartrite]

OK, these are the solar system bugs:

red earth.png

black moon.png


cel://Follow/Sol:Earth/2019-01-18T06:38:08.54844? ... 579132347&lm=2198&tsrc=0&ver=4


cel://Follow/Sol:Earth:Moon/2019-01-18T06:40:45.2 ... 579132347&lm=2198&tsrc=0&ver=4

Selden,

The Earth is red, the Moon is totally black from every angle.

I suspect it is not an ambient light problem, it is a texture problem.

There are no reflective surfaces.

Comparison: fictional planet with maximum ambient light in Alexell's version of Celestia

ambient.png

The same planet with the new build

white surface.jpg

I suppose I don't have to use any files from the Celestia 1.6.1 version...?

Because I use the provided 7z file as it is.

At least one image (white surface) from this post looks like what I was experiencing with normalmaps, should be fixed now. See viewtopic.php?p=142691#p142691

[john71]

Would the developers be so kind to give us a new Celestia exe file to try?

[onetwothree]

You can easily compile it yourself

[john71]

[pirogronian]

@onetwothree
@john71
I just red something about corss-compiling using cmake, but stucked in... Otherwise, maybe I'll could produce windows binaries from time to time.

[onetwothree]

The problem with crosscompiling is that you should have windows qt binaries as you need to patch qt libs using winqtdeploy application. It's possible to crosscompile qt libs but not tools.

[Janus]

Here, have some non official slightly tweaked binaries.
The display limit is set for 1GigaLY, and the debug error for albedo is removed, all it does is litter {developer} debug logs anyway.
As usual for my stuff, VS2015, fully static linked, stand alone.

I have no intention of bothering with the QT version unless there is a specific reason to do so.

If you are also wanting the updated ssc data, try looking in My Downloads for commit 5436.
The 7z archive is ~52MB.


Janus.

[onetwothree]

A new win snapshot is published to our bintray repository

Changes are:
- bugfix for specular maps, john71 could you check the bugs you reported earlier?
- duplicated entries for dwarf planets were removed
- location type for ring labels was changed from Other to Ring.

Added after 1 minute 27 seconds:

and the debug error for albedo is removed, all it does is litter {developer} debug logs anyway.

Could you share what have you changed?

[Janus]

@onetwothree

/src/celengine/solarsys.cpp:818

I simply commented it out.
Basically all older addons or extras use albedo directly.
Straight deprecation, even with a warning like that, breaks backwards compatibility.
Just let it "fail" in a compatible manner.
Even if you manage to force everyone to rewrite all the entries, you would also be forcing people to guess which is the proper path.
Their source may not have albedo, or whatever, broken down to the level you want, so they can't be as accurate or precise as you want.


Janus.

[Lafuente_Astronomy]

Janus

I noticed that your fork commits 5423 beyond prevents most of the Constellation diagrams from being shown. This could be problem as it is probable that your fork, in attempting to allow new stuff, has tampered with the system that allows the Constellation diagrams to appear in the first place:

As you see here:

There's only one diagram line shown in the Constellation of Monoceros

Is there a way to solve this? Thanks in advance

Added after 1 minute 25 seconds:
Note: I used your latest commit, 5436

[Janus]

Not me, this is a text handling regression.

During the greek stuff in name handling "upgrades" recently, changes were made to how labels were parsed.

Now neither 'Alpha Mon', nor 'alpha mon', nor any other purely text label matches, 'α Mon', whether from the keyboard, or in a name search.
This applies to any greek letter in a name, I have not tested for other languages, but it wouldn't surprise me.

Try hitting enter and selecting 'alpha centauri', good luck.
It wants to match, it shows a match, but fails to select.

Janus.

[Lafuente_Astronomy]

So, that means I'll have to revamp the entire asterisms.dat file just to make it work? Alright then

[Janus]

It was the change in text handling that broke it.
Not enough regression testing was done.
Matching should be done at search, not at load, breaks backwards compatibility.
Take a look at the Feb 8th greek commits to see how it was changed.

It should display the greek replacements, not store them, transforming search to match as needed.
Always store the names in the database in as close to plain text as you can get.
Simply do the opposite of the the greek changes did, and replace inserted greek with its plain text eqiuivalent before search.
Showing 'α' as part of a name is fine, but it isn't going to consistently match across languages, and it also fails to match alpha, or Alpha.
Shouldn't the PO or PO2 files be able to handle exact replacements like that?

I am tempted to undo it just for testing purposes.
In fact, I think I will once I get caught up, current job is very frustrating, every time I think I have fixed a problem, it is just another symptom.
It feels like the real problem is saying, I'm sorry; but the princess is another castle, over and over.


Janus.

Edited for clarity.

[pirogronian]

@Janus
@Lafuente_Astronomy
Just tested: selecting, both from kayboard and from browser, works, both for greek and non-greek letter names. However, constelation diagrams are indeed invisible, except two lines with Epsilon Mon A.
Edit: I got it: It doesn't work for entering plain text "as is", just for prepared completions. Fix is on the way.

[john71]

Thanx for everyone!!!

[Lafuente_Astronomy]

@Janus
@Lafuente_Astronomy
Just tested: selecting, both from kayboard and from browser, works, both for greek and non-greek letter names. However, constelation diagrams are indeed invisible, except two lines with Epsilon Mon A.
Edit: I got it: It doesn't work for entering plain text "as is", just for prepared completions. Fix is on the way.

Alright, thanks guys!

Hopefully that would allow the capacity to make asterisms.dat files for other systems so that I can make asterisms as seen from other solar systems out of fun or just reference

[pirogronian]

Here is fix PR, if someone can't wait for merging into master... https://github.com/CelestiaProject/Celestia/pull/249

[Janus]

It may be just me, but this still doesn't work.

Whatever all was done, ALF ~= Alpha ~= 'α', and even text searches using enter fail to match, even though they appear to.
What you type is narrowed to a match, then the selection does not change when you hit enter again.

Try searching for alpha centauri.
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.
Same problem as reading the asterisms.dat file, only with user input.


Janus.