I would really like to see the initial focus position adjusted for temperature. This can be easily incorporated by getting two numbers from the user: 1) slope of focus least square fit line 2) intercept of the line. These two numbers define precisely the best focus position as a function of temperature. This would allow the maintenance of an excellent focus position completely automated for a long period of time. Auto focusing would not even be required, except perhaps from time to time to confirm that the best fit line is optimum.
If additionally, a maximum deviation of the focus position from this calculated value could be given, then I would have no reservations about running the autofocus routine on some regular interval during each night's session. The reason I do not do so now is that, although the new auto-focus routine works wonderfully well, most of the time, it sometimes fails. When it fails, it leaves my focus position badly out of focus and that whole stretch of imaging time is toast. There are various reasons why it might fail, perhaps a brief cloud or wind gusts come along. My RC12 is particularly prone to auto-focus failures, so I never ever run the auto-focus unless I am attending it. It works so much better to just let the temperature compensation just do its thing.
This would in no way change anything for those folks who are very happy with the way it is running now. It would just be a new optional feature that could be turned on.