Let's talk about this briefly. I am preparing an alternative solution that uses the full functionality of insert for a new update. This fixes other issues such as, for example, updating external values. While the solution in this PR addresses the issue, the forthcoming new feature will fix this and all other issues with update. We may choose to keep _update for continuity but will likely remove it in future versions.

Would you please merge #757? Then it's good to go.

On my fork I implemented a new update that uses the functionality of insert. The new update function I wrote also prohibits updating of entries with autopopulated dependent entries (this behavior can also be suppressed). Please, let me know if this is of general interest.

Awesome stuff, @gucky92! We have also been working on this with a PR to address it imminent. @dimitri-yatsenko can you update if the new update1 will have features mentioned above?