Martin Adámek

> from what i understood, since those methods bypass the identity map, it would execute immediately vs waiting for flush? Nope, those are basically shortcuts for QB, would you expect...

I don't really understand what you are asking for here, could you provide more verbose example? Like runtime filtering that would work the same way as SQL where query? I...

`assign` does not work this way, it does not compare things, just sets new values based on metadata (it's of course a bit more complicated, but the point is I...

This is unfortunately much more complicated. It requires changes on hydration level and they would be breaking, or we would have to have the hydration for such happening twice, as...

Not sure if I like this enough to make it default, will need to think about it a bit more (but it's definitely an interesting idea). I'd call this a...

> i started using em.create/em.assign mostly because it was more ergonomic to define the relations without having to introduce additional api calls/lines (especially for ingesting user data, eg: graphql) for...

I am quite sure I saw such snippets either here in issues or on slack, and even if I haven't, I still consider this a breaking change - previously such...

You don't have the right EM instance from inside new entities. I know you are using `em.create()` but that is optional, with constructors we don't have a way to get...

Having the patched prototype seems like the best way to me.

FYI in crawlee the default storage is again using JSON files (and keeps things in memory rather than in sqlite database), so you should be able to alter things manually...