W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2019

Re: New Version Notification for draft-thomson-http-hx-uri-00.txt

From: Lucas Pardue <lucaspardue.24.7@gmail.com>
Date: Thu, 7 Mar 2019 23:48:54 +0000
Message-ID: <CALGR9oaDG3JLJsOMEkA+zad6r298xnSq5UE=aP=MfNPB9qpGmw@mail.gmail.com>
To: Martin Thomson <mt@lowentropy.net>
Cc: HTTP Working Group <ietf-http-wg@w3.org>
Ok so this is half-baked, and I didn't think about how to express bodies in
the document, but I imagined a Hierarchical Document consisting of hx
links. These are a series of steps the client wishes to perform against an
API. This articulates a set of conditinal actions which are dependent on
the previous step. So for instance, with HTTP/2

1) create object as if POST
2) let location be hx:///1/a/h/location?201
3) if location, modifying object at location as if POST
4) let hx:///3/s/
5) if status 2xx
6) modifying object at location as if POST
7) let status be hx:///5/s
8) if status 2xx
9) POST pizza order

A server that processes the document and completes all conditions will
return 2xx, any failure will lead to 4xx or 5xx. A client submits the
document and waits for the "final response".

Now, having written that, I see that I am unclear how hx and http(s) scheme
are supposed to be mixed with a single HTTP connection, or if that is
indeed the intent at all.

Lucas

On Thu, 7 Mar 2019, 00:14 Martin Thomson, <mt@lowentropy.net> wrote:

>
>
> On Thu, Mar 7, 2019, at 10:59, Lucas Pardue wrote:
> > The thought that occurred to me was was to embed the hx and hxr URIs in
> > a document that is uploaded as an atomic chain to a server.
> >
> > Is that what you're discussing here?
>
> Say more?  If you are talking about a single request, that might not be so
> advantageous.  This was really about managing dependency chains between
> requests without paying round trips.
>
>
Received on Thursday, 7 March 2019 23:49:24 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 7 March 2019 23:49:26 UTC