State should not depend on the API server

[David Cheney]

Thanks William,

On a related note, I'm also planning on moving api/params to
apiserver/params; these are the server's parameters, not the client's.

On Mon, Sep 1, 2014 at 5:12 PM, William Reade
<william.reade at canonical.com> wrote:
> On Mon, Sep 1, 2014 at 8:03 AM, John Meinel <john at arbash-meinel.com> wrote:
>> FWIW I'd favor 3 layers, though it does mean you have to do copying between
>> structs that would likely otherwise be almost identical. A State layer for
>> saving in the DB, an API layer for communication, and a Model layer for use
>> in the Agents/Clients.
>
> Yes please, in general: one set of types per layer boundary.
> Independent of what Dave's doing, which is necessary regardless, I
> agree with what you're saying: *except* that I think we really have to
> consider the API layer to be *two* layers.
>
> That is to say: if you can change some type in api/params and
> everything still works, you are Doing It Wrong. We cannot depend on
> servers and clients always running the same version -- so, every time
> you thus change server/client in a single motion, you're almost
> certainly introducing more or less subtle incompatibilities.
>
> So, I would be very pleased if we would stop using the same
> definitions (ie api/params) on both sides of the wire -- it's one of
> those things that's nice and easy when everyone's running the same
> version, but an active trap as soon as multiple versions exist (as
> they do).
>
> All the struct-copying is kinda boring, but the layering violations
> are straight-up evil.
>
> Dave, thanks very much for doing this.
>
> Cheers
> William