Jason Marshall

This seems like something I could use as well.

I don’t think exposing mutable data was meant to be part of the Config design and API. In fact I’m not happy with the current Buffer behavior referenced at the...

Good, then we are all three in accord (I've had this conversation before with Loren as well). I think there is a hole here, that I would like to make...

> you would have two entries in TableB that point to TableA as latest. From the migrations it looks more like I would have 2 entries in TableB that point...

So it looks like has_one works opposite from what I would have expected, and I need belongs_to, even though that inverts the relationship linguistically from what it is architecturally.

Yeah I'm going to have to generate a repro because my _id fields aren't getting populated at all now. What does 'declared in the wrong order' mean? Why would there...

So I believe my problem is that the `put_assoc` documentation, and much of the conversation around it, pretty much implies that it will take care of the parent-child relationship but...

``` # Parent def changeset(plan, attrs) do plan |> cast(attrs, []) |> cast_assoc(:current) end # Child def changeset(plan_edit, attrs) do plan_edit |> cast(attrs, [:title, :target_date, :started_at, :finished_at]) |> cast_assoc(:description) |>...

Yeah I need to work up a minimal repro case. I'm around the axle at this point.

Alright, this is a new liveview project with just a Thing liveview with a nested Comment field. https://github.com/jdmarshall/ecto_issue mix phx.gen.live wanted to create the latest field as: ``` field :latest_id,...