BTW, I noticed the 'git checkout ..." example I gave you was from an experimental branch, so the ID string for the submit before the median fix was wrong.
You probably noticed that, but just in case, here is the log list for the head of the tree right now:

> git log --oneline | head -20
6ce69709a Check for errors in WCS transformation
11f68d971 Correct bug in median computation
0d418a390 Make sure all image files permissions end up 644
b9dbfa720 v3.4.1 release
5632a36a1 Do not calculate download time or send image when upload mode is LOCAL
1d72ab065 Adding test for Ekos start/stop.
...

So, to go to the submit before the median fix you may want to try
git checkout 0d418a390

Then you could run
git status

and you'd see:
HEAD detached at 0d418a390
nothing to commit, working tree clean

and then go back to the recent with
git checkout master

and to verify, you'd run
git status

and you should see

On branch master
Your branch is up to date with 'origin/master'.
nothing to commit, working tree clean

Thanks,
Hy

Thanks. Please don't hesitate to ask for help.
Please also double check to make sure you're really running the older binary (I've made that mistake before).

Hy

I noticed the ID thing and chose the Release 3.4.1 ID. That was two or three "checkouts" before your fix.

I found the reason for my problem. It was my offset (still set to the default value of 8 ). This appears to be too low. With an offset of 32 I get a reasonable Bias frame. I NEVER took any Bias frames before and that's why I never realized an offset of 8 is too low. A light frame taken at 0.000032s also comes up empty. (Now I'm wondering how the Bias frame that I uploaded (offset=12) can be OK.) Anyway...

Thank you very much for your work and for explaining checkouts to me in particular! I guess this makes me a more useful code tester. In any event, I'm looking forward to using my new ability right here!

Thanks for testing, and glad to hear there's no issue with the new code.
I take it that you get good median values now

BTW, wrt Bias, there are some interesting conversations about NOT using the absolute lowest exposure time (and in this respect, I was tricked by the indi thing you pointed out to me about it ignoring the exposure value).
See, for instance, www.cloudynights.com/topic/579728-flat-f...-value/#entry7920805 where Rista recommends not using any bias exposure on the ASI1600 any lower than .3s.
May not apply to your camera, but ...