Smit Patel

From Query side, I am not entirely sure that it translates correctly. Nav expansion may be chopping it off. or we may generate join back. Or we may not generate...

(#29105) Every time I work in scaffolding area, I miss this so much!

I think best way to handle this with the current pattern/infra we have is to have a class which is combination of https://github.com/dotnet/efcore/blob/release/7.0/test/EFCore.SqlServer.FunctionalTests/Scaffolding/SqlServerDatabaseModelFactoryTest.cs https://github.com/dotnet/efcore/blob/release/7.0/test/EFCore.Design.Tests/Scaffolding/Internal/RelationalScaffoldingModelFactoryTest.cs https://github.com/dotnet/efcore/blob/release/7.0/test/EFCore.Design.Tests/Scaffolding/Internal/CSharpEntityTypeGeneratorTest.cs So we can test e2e...

Cosmos has `DateTimeToTicks` which returns ticks since unix epoch though not sure if Cosmos can handle DateTimeOffset as a type. Add translation if it is supported.

Should we mark property shadow in scaffolding in any case? How would we determine the heuristic when a user want a property to be skipped from CLR model? Wouldn't it...

@roji - I believe you investigated some perf work around rownumber. Can you answer why generating RowNumber is more efficient than CROSS APPLY?

@ajcvickers - I propose punting this issue.

> just a quick guess, but shouldn't the WHERE [t].[row]

@itan-mcp - Your last post is not related to this issue as there is no RowNumber. Grouping by scalar subquery generating invalid SQL is being tracked at https://github.com/aspnet/EntityFrameworkCore/issues/19027

Same fix as https://github.com/dotnet/efcore/issues/28247 but in entity equality rather than materialization.