Hiccup every few seconds

[gzotti]

Expected Behaviour

Smooth operation

Actual Behaviour

Describe or maybe attach a screenshot?

Every few seconds, the simulation stalls for half a second or so.

Stellarium.25.0+.2025-01-31.11-31-38.mp4

Steps to reproduce

Run it. Switch on altazimuthal grid and observe any body move over the sky in closeup. Also the selection marker stops.

System

• Stellarium version: master (6f21b4a)
• Operating system: Windows 11
• Graphics Card: Geforce Mobile 3700Ti
• Screen type (if applicable): 3440x1440, 10%

Logfile

Nothing suspicious.

[alex-w]

@gzotti did you see that problem in our latest beta?

[gzotti]

No, I did not check beta installers recently. If one of you has an idea what happened I'd be happy, if not, I will bisect maybe on Sunday.

[alex-w]

I cannot reproduce the issue in linux. Is it possible that Stellarium did get some downloads/updates at the moment of recording video?

[gzotti]

It hiccups every few seconds over a minute (was too annoying to watch for all of 10 minutes...). There is no progress bar visible. What would cause that, in a daylight sky? On a slow PC DSO textures take some time to load, but I won't see it here, Will keep observing.

[alex-w]

It hiccups every few seconds over a minute (was too annoying to watch for all of 10 minutes...). There is no progress bar visible. What would cause that, in a daylight sky? On a slow PC DSO textures take some time to load, but I won't see it here, Will keep observing.

Some heavy computations (a caching issue)?

[alex-w]

@gzotti would you check it again in FOV 6-10 degrees?

[gzotti]

OK, I am stopping review now and start bisecting this.

[alex-w]

Wait! Do you using VSOP or DE ephemeris?

[gzotti]

DE440. And the hiccup seems to be indeed related to the sun being selected. Indeed there are spikes accessing my secondary drive. What are you caching here? You see SSD activity monitor on the right. The spikes are from when Sun was centered. With Venus, zero problems.

[alex-w]

Hmm… a barycentric computation issue? Would you look at that too @henrysky?

[gzotti]

That is something I have to look at in detail. What were the reasons to switch to, and what are the implications for using barycentric computations? Won't we have to use a different sub-flavour of VSOP87 for that? We always had assumed heliocentric computations, the sun is implicitly at 0/0/0 and there may be silent assumptions that may have been broken now. The barycenter of the Solar system is AFAIK inside the sun. Why do we need the switch when dealing with stellar proper motions or parallaxes? How much are stellar coordinates affected when switching SSO computations to barycentric?

[gzotti]

I just see the disk load is related to Sun infostring "Additional information". Which is weird, given how little is added. Wow, this is even in 24.4!

[gzotti]

Uh, dc25540. Will keep me busy!

[henrysky]

That is something I have to look at in detail. What were the reasons to switch to...

Both parallax and aberration are supposed to use barycentric position and velocity and the fix originally seems easy enough for me. The differences between barycentric and heliocentric are very small as you said so probably does not matter if switched back to heliocentric.

I have a commit here to undo the barycenter calculation here just in case if it fixes the issue (henrysky/stellarium@d1d6a16).

[gzotti]

Thanks. I'll have a look once I fixed my own mess...