Coverage decreased (-0.008%) to 88.574% when pulling e6295d0 on dimitri-yatsenko:dev into 0c9fac0 on datajoint:master.

I'm not entirely sure if I agree about adding this, at least without some discussions. I completely understand where this is coming from, but I want to make sure we flesh out all imaginable cases before we put in what would be a fairly strong restriction.

This feature is introduced due to a user issue when they did not understand how auto-populate worked. They first inserted into master and then called populate to have the parts populated. It took me a while to catch the bug because I did not imagine someone inserting into a computed table.

This feature is important for new users to understand how to properly use DataJoint.

If the user must insert directly, they can set allow_direct_insert=True to bypass the restriction.

Ok makes sense and ability to bypass was something I was going to request. Ok - I'd still like to think about this before I will fully accept. I also would prefer that you would create an issue describing this so that this PR can be considered as a fix for that issue.

It is a fix of issue #511

Ah I oversaw that.