While trying to set an image sequence last night I ran into a possible bug in the image capture module. I was trying to capture M31 and was doing a image preview in the capture module to check that the camera was framing the target properly. This is what happens:

1. Move the telescope to target and plate solve (focus was already checked).
2. In the Image Module capture a preview to check framing.
3. On the first or second preview (usually the second) the Image Module will display "Meridian Flip" instead of the usual image capture message (see screen capture below). The need for a meridian flip is about four hours away so this should not occur. I do have the meridian flip if HA > 0.1 set in the mount module.
4. At this point I have to exit the software and reboot to recover.
5. I repeated the same thing on a Raspberry Pi 3B+ running the nightly build and a Tinkerboard running the stable release, both running the latest 3.3.3 as updated that day.
6. The system time , location and offset, etc. were all set correctly.
7. This ended my imaging attempts for the night

I have attached logs and a screen capture for details. Thanks for any help or suggestions.

Thanks for the report! Could you please check whether you have the log file that fits to the screenshot? The log that you attached starts at 00:38 and I cannot find anything unusual.

- Wolfgang

Hi Wolfgang,
The screenshot was taken when I first tested that evening, I then had to exit the software, reboot the system and repeat the test because I did not have the logs enabled at the first test. After starting everything up I enabled logs and ran the steps as outlined in my first post and the problem repeated perfectly as before. The log file is the correct file for the failure on the second run. I was able to repeat this same thing last night as well. Not sure I have logs for last night but will check later today.

I found an older SD card with KStars 3.1.0 and then ran that version last night after the 3.3.3 version failed with the "Meridian Flip" message again. The 3.1.0 version didn't exhibit this failure and I was able to set a sequence for the rest of the night on M31 which finished successfully.

If it is clear tonight I will repeat the steps above on 3.3.3 with logs enabled again and will attach those as well. Hopefully something will show.

Thanks for your help and to all the team for this great software!

Before you loose again precious time at night, maybe you could try to reproduce it with simulators. That's how I start with debugging.

Yes correct, I plan to test with the simulators, however I will be tied up for most of the day and will likely be late when I get a chance to test so I was just going straight to the scope. That being said I will first test with the simulators anyway as I am curious if it fails with the simulators.

Indeed you are right about losing precious night sky time, which has been all to rare here this year. In fact this series of clear skies is the first I have been able to get out for the past several months.

I have been able to easily repeat this bug using the simulators on both my Raspberry Pi 3B+ (running the latest 3.3.3 nightly builds) as well as my current Kubuntu desktop machine running the latest 3.3.3 release build. After starting KStars and Ekos and connecting to the "Simulators" profile do the following:

1. Set the time in KStars for 12:30am ( I do this to match the time that M 31 is high enough to clear my trees for imaging - over 30 degrees). This may not be needed however I am setting the system to reproduce my situation.
2. Slew the telescope simulator to M 31.
3. In the Image Capture Module capture a preview by clicking the "Capture Preview " icon
4. The "Meridian Flip" message appears at the exposure progress area with the spinning "busy" icon just as in the screen shot from my first post.
5. Note - this even occurs when I set an imaging sequence of M 31 and use the Scheduler running from a fresh reboot or start up and no other action taken. The "meridian Flip" message appears with the system just hanging and waiting and no exposure occurs. This is why my attempts to image M 31 over the last several nights have failed. They simply won't run past this point.
6. I have also noticed I can stop the preview or sequence by hitting the stop button in the imaging module but I still need a reboot to get the system working properly afterwards.

I have attached part of a log file with the activity log of when I ran this test with simulators last night on my Raspberry Pi. At this point I have easily repeated this on a Raspberry Pi3B+ running the latest nightly build (3.3.3), on an Asus Tinkerboard running the latest release build (3.3.3), and on my Kubuntu desktop system running the latest 3.3.3 release build.

OK, now I could reproduce the case. It looks like it is crucial to set the time manually. If I use a target that is with current time post meridian AND set the time so that with that time it is before meridian, the mount changes to "MF planned". No clue, why.

Could you reproduce it with having the time set to current time?

- Wolfgang

This might be a case where the code checks both the system time and the simulation time. All time checks should use the KStars singleton instance for homogeneity.

-Eric

Every time I had this problem the time in KStars was exactly the same as the system time. My first test with the simulators was the night before while I was in the field with my system. The "Meridian Flip" bug occurred with my real hardware as before so I rebooted the Pi3B+ and then tested with the simulators which failed exactly the same way. I just set the time at 12:30 AM when running the test with the simulators on my desktop computer the next morning to match the time when I was in the field with my system. In any case the bug is fully repeatable which, in my experience, makes the art of debugging a bit easier.

I have been able to move to other targets, such as M 20 with no problem. By way of troubleshooting I also tested with a star such as o Andromedae about an hour or so earlier in the evening as this had my scope pointing in a similar position as it would if pointed to the Andromeda Galaxy later on. In this case the "Meridian Flip" message also occurred. As it stands I have spent several hours testing (with several lost nights) and trying to find a workaround to the problem.

I hope this helps clarify what is happening.

I found it!

When calculating the current HA, I did not ensure that the HA is between +12 and -12. This has not been discovered yet since small RA values were below the horizon. With M31 having an RA of ~ 00:45, this creates problems.

As a fix, please deactivate the meridian flip option for targets with small RA values. The fixing patch is on its way.

- Wolfgang

This is excellent! Well done and thank you for your help!

I'll look for the fix in the nightly build some time soon. Until then I'll follow your suggestion and deactivate meridian flip for targets in this region.

The fix is meanwhile part of the latest KStars v3.3.4 release. If you want to have it immediately, please check out the latest version from github and build it from the sources.