Jim if you can make the file available that would help.

Cerro Torre, Jasem said it is probably a build issue, you might try deleting your kstars build folder and just rebuild it and see if that helps

Sure thing. Please find attached. I don't think it is the initial file that is loaded that is the problem. That image is solved and the scope moves to the location of the solve. It is upon this next capture/solve to find out how close it is that things go sideways.

It can be reproduced with the simulator as well.

Jim

PS. @knro would it be possible to add .bz2 to the allowed file extensions? .tar.bz2 is there, but it rejects just .bz2. You could then replace .tar.bz2 if you want.

Hi Jim, I tried it both in KStars and in the tester program and it solved right away both times. It didn't have to use any parallelization either since it had the correct scale and position information. It solved and slewed right away. So I'm not sure?

Thanks for checking, Rob. It solves and slews for me as well. It is after the slew when it takes another image to compare that it seems to have the problem. Do you have auto-WCS turned on in the options? I opened another thread where when fitsviewer opens after capture, if auto-WCS is on, it crashes to the desktop in the same way. I think now, after looking at gdb, that they may be related. That thread is here: indilib.org/forum/development/8067-fitsv...rash-to-desktop.html

Yep, I do have Auto - WCS turned on.

Spent a little more time digging into this tonight. My local package install of astrometry is redundant with stellarsolver, so I removed that thinking perhaps there was a conflict, but kstars must only be using the stellarsolver included version of astrometry as removing the system package had no effect.

I then moved on to looking through the backtrace from gdb, and it aborts on line 24 of kdtree.c, and fails a validation with the error "Data, N, or D is zero". The program appears to abort when an fprintf has invalid parameters.

This can be seen directly in the backtrace as well. Any idea what Data, N, or D are, and why they would be set for your config but not mine? It comes from blind/verify.c line 92. A quick glance seems to indicate these may have something to do with scaling?

I do have an idea of how you can find out more detail.

Since StellarSolver uses an internal build of astrometry.net, it has the ability to use some nice logging features. If you go to the KStars logging, you can select the astrometry logging level. You can try setting it to a setting that is not Log_NONE, such as Log_Verbose or all Just be sure if you set it to one of the higher settings, be sure to tell it to save the output to a file because if you set the logging too high, the volume of messages coming from astrometry can be quite extreme and can overwhelm kstars with messages. So by default it saves to a file, and I would recommend saving it somewhere you can find it.

Also you were asking about whether StellarSolver can use the local astrometry.net. The answer is yes absolutely. You can select that in the StellarSolver options.

Thanks Rob, see private messages please.

A couple more questions. Do you have gsc installed? If yes, do things still work if you uninstall it? I need to create a copr repository for gsc, as I currently do not have it, and it is not available for Fedora in the standard repositories. It seems kstars must work okay without it though, or there would be other issues with the Fedora production packages.
It also appears that the version of gsc in use is old, but that is a different topic.

So GSC is just used by the CCD Simulator. If you uninstall GSC, the simulator will have no stars. So if you are planning to do testing with the simulators, you need GSC. But if you are doing real astrophotos, you don't need GSC.

That was my understanding as well. It seems kstars will still allow you to set up and try to use the simulator without gsc installed. This in turn is likely where the crash comes into play, as no stars are passed to the solver. The fact that gsc and/or no stars exist, should probably be caught and an error message logged.

Thoughts?

It might be good to check for gsc when the CCD simulator is added to a profile too. @knro ?