Re: [secdir] SecDir review of draft-loreto-http-bidirectional-05

[Peter Saint-Andre <[e-mail]>]

Thanks for your review, and our apologies for the delayed reply.

On 12/16/10 9:38 AM, Laganier, Julien wrote:
> I have reviewed this document as part of the security directorate's
> ongoing effort to review all IETF documents being processed by the
> IESG.  These comments were written primarily for the benefit of the
> security area directors.  Document editors and WG chairs should treat
> these comments just like any other last call comments.
> 
> The document describes "Known issues and best practices for the Use
> of Long Polling and                    Streaming in Bidirectional
> HTTP", and it has the following abstract:
> 
> There is widespread interest in using the Hypertext Transfer
> Protocol (HTTP) to enable asynchronous or server-initiated
> communication from a server to a client as well as from a client to a
> server.  This document describes the known issues and the best
> practices related to the use of HTTP, as it exists today, to enable
> such "bidirectional HTTP".  The two existing mechanisms, called "HTTP
> long polling" and "HTTP streaming" are described.
> 
> The document is very clear and articulate and I have not found any
> security issues that were not covered appropriately in the Security
> Considerations sections.
> 
> I have two concerns regarding the use of "should", "must" etc.:
> 
> 1. I have found at least one occurrence where a recommendation is
> made using lower cases "recommended" and "should". Should upper cases
> be used instead?

Currently this document does not reference RFC 2119 or use capitalized
keywords. Instead of adding such a reference, I suggest changing that
text to:

   Several experiments have shown success with timeouts as high as 120
   seconds, but generally 30 seconds is a safer value.  Therefore
   vendors of network equipment wishing to be compatible with the HTTP
   long polling mechanism are advised to implement a timeout
   substantially greater than 30 seconds (where "substantially" means
   several times more than the medium network transit time).

> 2. Similarly, parts of the text describes node behavior using lower
> cases "should" and "must". This makes it hard for the reader to
> differentiate between behavior specified in another standard document
> from behavior that can be reasonably expected from a deployed
> implementation. I would suggest that upper case requirements key
> words ("SHOULD", "MUST") be used if the behavior thereby enounced is
> specified within another RFC documents, and that document be cited
> next to the statement.

The sentences you mention indeed simply cite other RFCs. Because the
actual normative text is contained in the referenced RFCs, I suggest
that we remove the lowercase "should" and "must" words from this I-D.

> Nits:
> 
> s/DOS attacks\.[RFC4732]/Denial-of-Service (DoS) attacks [RFC4732]/

Fixed.

Peter

-- 
Peter Saint-Andre
https://stpeter.im/