You can have any previous version you like. checkout the one immediately before the stellarsolver perhaps?
commit f34bcb7d999662145425ce83173adbf5b6f738bb (HEAD -> master, origin/master, origin/HEA
Date: Fri Mar 11 14:10:39 2022 +0300
Disable event-driven setScopeStatus from INDI::Telescope since it sets m_Status which
prevents the polling method from working
Date: Fri Mar 11 13:09:56 2022 +0300
Fix bug that caused PAA to fail if a driver send HORIZONTAL_COORD as well
Date: Fri Mar 11 02:46:21 2022 +0300
Improve Look n Feel tab
Date: Thu Mar 10 20:56:51 2022 +0000
updating to StellarSolver 2.1
Date: Wed Mar 9 13:36:24 2022 +0100
A rare clear night occurred on 9/3/2022 and we set up to image. Unfortunately all was going fine until we tried to plate solve the first target and KStars crashed. No matter what we tried on the spot there was no retrieving the situation - not at all happy. On checking our records, the previous time we had a good night and everything worked was on 2/3/2022. It didn't take much checking to find that there had been a KStars update on 4/3/2022 to version 3.5.7. The computer in question is a Lenovo notebook with an 11th generation i5-1135G7 processor running Ubuntu Mate 21.10.
The next day we uninstalled KStars 3.5.7 with sudo apt remove kstars-bleeding and installed an old version using sudo apt install kstars. It was found that this version worked without any issues. Further investigation found that kStars version 3.5.7 will not even plate solve using the simulator as per other people's reports.
KStars 3.5.7 is causing a memory segmentation fault when starting the plate solver, see below.
org.kde.kstars.ekos.align: Target updated to JNow RA: "06h 46m 08s" DE: "-16° 44' 49\""
org.kde.kstars.indi: Telescope Simulator : "[INFO] Slewing to RA: 6:46:08 - DEC: -16:44:49 "
org.kde.kstars.ekos.guide: "Mount is moving. Resetting calibration..."
org.kde.kstars.ekos.guide: "Calibration is cleared."
org.kde.kstars.indi: Telescope Simulator : "[INFO] Telescope slew is complete. Tracking... "
org.kde.kstars.ekos.align: "Capturing image..."
Found one coordinate representation.
org.kde.kstars.ekos.align: "Image received."
Segmentation fault (core dumped)
Now for an interesting twist, we have three other machines (that we don't use for actual imaging) that have KStars installed. All three of them had also automatically updated to Version 3.5.7 and amazingly, all three plate solve with the simulator with no issues. Two of these machines are older quad core (one AMD and the other i3) notebooks running Ubuntu 20.04LTS. The other machine is a desktop i5-8600 8th generation processor running Ubuntu Mate 21.10.
Amazingly, it would appear possible that this latest version of KStars has been compiled with a compiler not compatible with current up to date hardware.
As others have stated, there needs to be a simple way to roll back KStars installs if the stable versions have such significant bugs.
I guess, in some respects, it is our own fault for allow automatic updates, as there are many warnings on the Internet regarding not upgrading KStars if it is working on your hardware.
Hopefully this issue can be addressed quickly as we, and other people, have obviously lost significant imaging time.
So I worked very hard for the last three weeks or so in a separate branch of StellarSolver and KStars to perfect StellarSolver 2.0 so that it wouldn't cause any issues when released to the main repositories since it has API changes. And then when it was ready for more widespread testing, we put each in the master branch simultaneously and thought that all was well and so we kept working on perfecting it all. The thought was that if both upgraded at the same time, there should be very little issues with incompatibility since their API's would match. And then, it got put in the PPA so people could start testing it. Unfortunately, when it was pushed to the PPA, there was a significant issue because 3.5.7 was also using the ppa and was built for StellarSolver 1.9. So when StellarSolver got upgraded with its new API settings, it was totally NOT compatible with KStars 3.5.7. So at first, this caused some big problems, but really quickly we made a patch for 3.5.7 to make it compatible. Unfortunately, the testing and refinements were not yet completed yet since it was still in a beta form so it wasn't perfect yet. Since then I have refined things further and worked out most of the bugs. So I am very sorry if you had any crashes or bad behavior due to this. I hope that in the future we can find a away to make the stable version of KStars depend upon the specific version of StellarSolver that it was designed for so that it doesn't start using a version where the API doesn't match. I'm not planning anymore significant API changes for a long time and this was the first time I changed it in well over a year.
KStars 3.5.8 beta should be working very well right now with the latest stellarsolver and my plan is to release a new version of Stellarsolver and make the new KStars depend on that so that you will have a nice stable version of KStars 3.5.8 hopefully in a few days. We are still doing bug fixing and testing, but it should be soon.
Rob, just letting you know that the linux system that was failing just did an update from the repository and the version of KStars now installed is 3.5.7 with a build date of 16/3/2022. Thus far testing with the simulators indicates that the issues have gone as it both plate solves and does a simulated polar alignment with no issues. Hopefully we will get an opportunity to test with actual hardware tomorrow evening as the weather looks promising.