Samuel Colvin

Good point, I'll do that now. It'll also match V2.

thanks again for this.

We should probably be consistent with `dict` and `Dict`, but more generally pydantic can't really make any guarantee about copying everything so editing a model could never modify it's input...

I understand why you might think that, but in the vast majority of situations it's not what people want and would be unnecessarily slow. Example ```py data = yaml_or_toml_or_json_or_pickle_or_whatever.load(raw_input_data_bytes) m...

Sorry for the slow reply. This is a feature request not a bug. We'll try to support this in V2, in theory shouldn't be too hard.

Completely agree with @tiangolo. I'll be ignore this until someone comes back and provides a constructive approach for solving it.

`exclude` on `from_orm` would be fine by me, but I think it's more of a work around than an actual fix.

This is fixed in pydantic-core, recursion cycles will be safely caught and raised in pydantic v2. See https://github.com/samuelcolvin/pydantic-core/pull/134.

I've changed the label from bug to feature request since we don't make any guarantee about this working the other way at present. The reason for this is: https://github.com/samuelcolvin/pydantic/blob/645e5fe6a0c74c95c24410571e9bb804af3eb677/pydantic/main.py#L857-L869 This...

This is fixed in v2 I think. I'll add a unit test to confirm that order is maintained. I doubt it will get changed in v1, but hopefully v2 will...