Michael J. Sullivan

This one sure is weird

I would call this a feature request, yeah. I don't think the multiple inheritance issue is a problem, we would just reject conflicts like we do currently with multiply inherited...

@MrFoxPro's example is being rejected correctly. It is doing an update from within an update trigger, and so should be rejected (and is). The original issue does seem like a...

We still reject strings containing `\(` so we could do the original syntax or the f-string version. I'm partial to f-strings but only because I'm very used to them.

I think the best approach for now is to use global variables to configure them and refer to them in `when` clauses

The difficulty here is that we would have no way of validating that the specified function always* produces unique ids. EdgeDB relies on ids being unique across *all* objects (not...

@vpetrovykh Could you triage this, and send it over to @aljazerzen or @fantix depending on what seems wrong?

Reduced version: ``` select User { deck: { name } order by { asdf := User.deck.name }.asdf } filter .name = 'Alice'; ``` fails with `InternalServerError: more than one row...

I think probably details should be dumped in the default formatting

We *do* do static checking of types and cardinalities. The issue here is that for inserts and updates, we intentionally don't enforce whether it is required statically, because required links...