Yoh Deadfall

> All in all, I'm for the way, when on the next major release we deprecate the previous API and it's removed with the next one. It's possible only in...

I understand, but can't see how it's possible at all. In COM all interfaced are resolved by IIDs, but in .NET by names and it's not possible to have two...

Good that we all share the same thoughts, thank you all for taking a look! But are all of you are fine with `NpgsqlTimetz` constructors? Wouldn't it be better to...

> Is it OK if I hold off reviewing for now, until you bring this PR to a ready-to-review state? Or do you think it's useful to give this a...

Hey, @Emill, I saw that you have made developer notes on PostgreSQL types and mentioned that theoretical limits of date/time types are wider than practical. Here you can the real...

> I'll try to look at all this at some point. Then let's push .NET platform folks on dotnet/runtime#19522, dotnet/runtime#27665 and dotnet/runtime#27687.

Yeah, but we have weighty arguments why this methods should be added. So while inventing a new EF function, we could help the community promoting the issues.

@ajcvickers, it means [PostgreSQL composite types](https://www.postgresql.org/docs/11/rowtypes.html) which Npgsql is able to project to CLR types. [See test](https://github.com/npgsql/npgsql/blob/dev/test/Npgsql.Tests/Types/CompositeTests.cs) for an example.

Well, my suggestion here is to improve serialization `DateTime` by using a custom converter first which isn't a problem, but it might be worth t introduce an Elastic specific type...

For Mac there's AV Foundation which bindings are available through `Xamarin.Mac`.