Re: [art] Last Call for Comments: draft-nottingham-rfc7320bis (BCP 190 update)

[Adam Roach <[e-mail]>]

Ah, okay. So it sounds like you're suggesting that Section 3 should add 
new content to talk about additional techniques, beyond the three it 
already describes.

Can you propose some text?

/a

On 10/16/19 7:16 PM, Larry Masinter wrote:
> As I understand it, people wanting to do what used to be separate protocols
> now define the network interface as a kind of REST API.
>
>
> https://en.wikipedia.org/wiki/Hypertext_Application_Language
>
>
> HAL is the name for a style of Interface where the relationship between one
> resource and other resources that are related are not defined by asking the
> client to do string manipulation on the URLs (like you might by defining a
> base and doing relative relations) but rather explicitly returning
> "hypermedia objects" that contain explicit URLs of the related resources.
>
> JSON-HAL is just one such building block.  When I mentioned it on the ART
> list, Carsten Bormann noted that the CoRE working group was going to adopt a
> derivative called CoRAL.
>   (draft-hartke-t2trg-coral-02 contained a note about JSON HAL that was
> dropped in later versions.)
>
> The reason for including any of this in rfc7320bis/BCP-190 is to point
> people at this rather extensive body of work by people who were trying to
> solve the problem that BCP-190 was set up to warn people away from.
> BCP-190 started as an internet draft "get off my lawn" but as is, it gives
> scant advice for how to avoid the practice and still accomplish the goals.
>
> It just seemed odd to update BCP 190 without giving more of a hint about
> what seemed to be a widespread method of avoiding doing excessive and
> unwieldy URL mangling in network protocols.
>
>
>>> As a data point, for the applications in the CoRE working group, we
>>> have mostly been able to avoid BCP190-style arguments by using
>>> /.well-known, mainly because simple devices only tend to have one
>>> service offered directly under / and because IoT device platforms
>>> tend to provide the application developer full control over the URI
> space.
>>> /.well-known/core provides a discovery mechanism for the entry point
>>> URIs actually offered by a server.
>>> For where this is not enough, the WG has just last week adopted CoRAL
>>> (not yet resubmitted as draft-ietf, so you can find it at
>>> draft-hartke-t2trg-coral) as our idea of a hypermedia format like HAL.
>>> Up to now, we tried to make everything work with RFC 6690 link
>>> format, but that has too many idiosyncrasies that started to get in
>>> the way of a long-term way forward.
>
>
> https://it.toolbox.com/blogs/craigborysowich/overview-of-api-hypermedia-type
> s-093018
> https://www.fenews.co.uk/press-releases/33067-detailed-guide-api-technical-a
> nd-data-standards-v2-2019
>