Add OData- prefix back to EntityId header

The following is feedback received from the IANA OData Header registration.  The summary of the feed back is that the EntityId and Isolation header names may be too generic

See also https://protect-us.mimecast.com/s/6cHPCQWX4XfkDzX5iPE-q-?domain=mnot.net

From: Ted Hardie <ted.ietf@gmail.com>
Date: Wed, Aug 29, 2018 at 2:58 PM
Subject: Re: [Ietf-message-headers] Registration request for EntityId, Isolation, OData-MaxVersion, and OData-Version
To: Chet Ensign <chet.ensign@oasis-open.org>
Cc: ietf-message-headers@ietf.org, "Mark Biamonte (Progress)" <Mark.Biamonte@progress.com>

Hi Chet,

I'm a little confused on one point.  The Isolation header states that it was called OData-Isolation in version 4.0, but the linked ABNF document says:

isolation  = [ "OData-" ] "Isolation" ":" OWS "snapshot"

That seems to indicate that both ODate-Isolation and Isolation are expected to be valid for this token.  The ABNF for EntityID is similar:

entityid   = [ "OData-" ] "EntityID" ":" OWS IRI-in-header

Can you clarify the intent here?  Are both expressions of this token expected to continue?

In general, "Isolation" seems to be a very general concept, and the reservation of it, unadorned, for a single token use ("snapshot") is somewhat surprising compared to the more-obviously scoped OData-isolation.  EntityID seems similar board in possible usage outside the OData specification. 

regards,

Ted Hardie

Hi Mark,

On Thu, Aug 30, 2018 at 6:49 PM, Mark Biamonte <Mark.Biamonte@progress.com> wrote:

Hi Ted,

I am a member of the OData Technical Committee (TC).  Chet submitted the request for OASIS on behalf of the OData TC.  You are correct, in OData 4.0 the Headers were titled OData-Isolation and OData-EntityID.  In the latest OData 4.01 spec the header name was changed to just Isolation and EntityID.  The OData- prefixed versions are still supported for backwards compatibility. 

[TH] Okay; I was probing to find out if this superseded the previous header, but it sounds like you expect both to be used at least during the compatibility period.

This change was made based on feedback for users and implementors of the OData standard.  I gather your preference would be that we use the OData-prefixed version for the registrations.  If so I will need to discuss that with the TC and get back to you.

[TH] While that would be my personal preference, it is simply that:  my opinion as an individual.  I'm primarily concerned that other, unregistered or pre-registration uses of these more general concepts may occur.  While recording them in the registry will help avoid collisions, there is always some risk and the more general the term, the higher the risk.

regards,

Ted Hardie

> On 1 Sep 2018, at 2:54 am, Ted Hardie <ted.ietf@gmail.com> wrote:
>
>> This change was made based on feedback for users and implementors of the OData standard. I gather your preference would be that we use the OData-prefixed version for the registrations. If so I will need to discuss that with the TC and get back to you.
>
>
> While that would be my personal preference, it is simply that: my opinion as an individual. I'm primarily concerned that other, unregistered or pre-registration uses of these more general concepts may occur. While recording them in the registry will help avoid collisions, there is always some risk and the more general the term, the higher the risk.
>

Speaking personally, I'd second that concern. It's least confusing when headers that are specific to an application (in this case, OData) include that application's name in their field name.

Cheers,

–
Mark Nottingham https://www.mnot.net/