Question: any practical differences between 2 types of declarations in DbContext?

[zelenij]

Adding new types to DbContext, your intro suggests this type of syntax

    [<DefaultValue>] val mutable countries : DbSet<Country>
    member __.Countries with get() = __.countries and set v = __.countries <- v

Wouldn't it be the same to use this syntax from practical point of view?

    member val Countries: DbSet<Country> = null with get, set

As far as migrations generation, both seem to work fine...

[simon-reynolds]

Doesn't seem to make a difference but a plain member val is definitely more succinct
Thank you for this, I'll update the docs and scaffolding generation to switch to a plain member val

[GunnerGuyven]

When I take this approach, the entity members of DBContext remain null. The longer definition works as expected. Am I missing some step?

type MyDataContext() =
    inherit DbContext()

    member val MyExampleTable: DbSet<MyTableType> = null with get,set

    [<DefaultValue>]
    val mutable private myExampleTable2: DbSet<MyTableType2>

    member __.MyExampleTable2
        with get () = __.myExampleTable2
        and set v = __.myExampleTable2 <- v

use db = new MyDataContext()
db.MyExampleTable2.First() |> ignore    // works
db.MyExampleTable.First() |> ignore      // null exception

[simon-reynolds]

No, it does seem that this approach does result in the members remaining null at runtime
I'll revert these changes in the documentation

[zelenij]

Looks like this has been raised in the past: fsharp/fslang-suggestions#103