Re: Extraneous nesting in QAPI schema

[Christophe de Dinechin]

> On 16 Dec 2019, at 17:59, Markus Armbruster <address@hidden> wrote:
> 
> Kevin suggested to investigate a more generic flattening solutions.
> 
> Of course, flattening is only possible as long as there are no name
> clashes.

Or you could reverse the name sequence and allow
the “upper” layers to disambiguate.

backend.data.addr.type -> type.addr.data.backend
other.data.addr.data.type -> type.data.addr.data.other

then “type” would match both, could be disambiguated with
type.other or type.backend, or even type.o or type.b.

That way, you could have some generic way to disambiguate
for the rare cases you need it.