p1: Purely Editorial Feedback

This is strictly editorial feedback on the latest p1.

* 2.2 "... requirements that an automated action be confirmed by the user before proceeding can be met via advance configuration choices..."   s/can/might/  (don't imply that it's a closed set)

* 2.2 Add text to indicate that the chain of intermediaries isn't necessarily fixed; i.e., that while it goes through "C" this time, the next request might go direct to origin, or through "D", or...

* 2.3 'A "gateway"... is a receiving agent that acts a a layer above some other server(s) and translates the received requests to the underlying server's protocol.'   "layer" "receiving agent" and "underlying" are awkward here. Suggest:

"""
A "gateway" (a.k.a., "reverse proxy") is a server that acts as an origin server, but translates received requests and forwards them to another server or servers, using any protocol (possibly HTTP).
"""

* 2.3 "MUST implement the Connection and Via header fields for both connections." --> "... header fields for both inbound and outbound connections."

* 2.7.1 "Other protocols might also be used..." -> "Other transport protocols might also be used..."

* 4.3 "For chained requests..." -> "For requests from an intermediary..." 

* 5.2 "If the client has a response cache and the request semantics can be satisfied by a cache ([Part6]), then the request is usually directed to the cache first." --> "If the client has a HTTP cache [Part6] and the request can be satisfied by it, then the request is usually directed there first."  (simplify, simplify)

* 5.7.2 "A transforming proxy MUST preserve the message payload..."  "MUST NOT modify" would be clearer here.

* 6. "HTTP only presumes a reliable transport with in-order delivery of requests and the corresponding responses."   This isn't really well-stated; the important thing is that the data transport itself is in-order, because requests and responses themselves can be chunked into multiple messages.  Suggest: 

"HTTP only presumes a reliable, bi-directional transport with in-order delivery."

* 6. "Most severs are designed to maintain thousands of concurrent connections.."   s/thousands/many/

* 6. "Most clients maintain multiple connections in parallel..." --> "Clients MAY maintain multiple connections in parallel..."

* 6.7 uses unregistered upgrade tokens in the example; this should be noted.

* 6.7 There should be a full example of an Upgrade header in a request and in a response.

* 8.3 There should be an appropriate reference to RFC6585 here for 431 Request Header Fields Too Large.

* A.2 Is there any particular ordering here? Perhaps they should be ordered by section?  (same for other parts)

Thanks,


--
Mark Nottingham   http://www.mnot.net/

Received on Saturday, 20 April 2013 03:25:36 UTC