[EnMasse] Upcoming REST API changes

[Gordon Sim]

On 08/05/18 11:32, Gordon Sim wrote:
> I guess my question would be what the aims are.

Just to clarify, I get that one idea is to allow the API to be 
implemented in different ways. However we are also discussing the 
constraints on the possible APIs from backing them with CRDs. So partly 
I mean the aim of using CRDs and partly also the envisaged use case for 
the API itself.

> Is it to allow kubectl 
> to be used intuitively in place of a custom tool?

If this is the case, then perhaps requiring valid kubernetes names would 
not be an issue since the user would be familiar with that constraint?

This is also perhaps true for people working with the yaml definitions 
directly, which I think is all users of the REST API, right? So perhaps 
there is no issue in requiring kubernetes valid names in the API?

The console could still hide that detail if/when used though obviously 
it gets harder to ensure uniqueness if combining user defined names and 
auto-mangled names.

> Is it for some better 
> form of interaction at scale? Is it to allow namespacing of 
> address-spaces? Or combination of these?