Re: editorial feedback on draft-mbelshe-httpbis-spdy-00

On Sat, Mar 24, 2012 at 7:00 AM, Julian Reschke <julian.reschke@gmx.de>wrote:

> 1. Overview
>
> I find the statement about pipelining a bit ... unbalanced. "Intermediary
> Inference" is no problem at all if you run over SSL, which is something
> SPDY does by default. Unless I'm missing something.


Agreed.


>
> 3.2.1 Request
>
> ":version" - so this essentially positions SPDY as alternate transport
> layer for HTTP; preserving the version number. Out of curiosity; what is it
> needed for? Preserving all information when tunneling? The same question
> applies to ":scheme"...
>

Since it's possible to layer different (future) versions of HTTP on top of
SPDY, don't we need the ":version" header to preserve all information? And
similarly, we can conceivably handle different schemes over SPDY, such as
https (the obvious one), http, ws, wss, etc, so I think including ":scheme"
is important.


>
> 3.2.3 Authentication
>
> a) as far as I understand, RFC 4559 does not define the "NTLM" scheme.
>
> b) in general, it's not clear why this paragraph is here; does
> Authentication work any different than in HTTP? Maybe just point to HTTPbis
> P7?
>
>
> 3.3
>
> "Browsers MUST implement throttles..."
>
> That doesn't seem to be testable. Maybe replace with advice in prose.
>
> 6.2 SETTINGS frame
>
> This seems to be a feature complete orthogonal to the remainder of the
> spec. Maybe just remove it?
>

Are you saying you don't feel it's worthwhile calling out the privacy
consideration of SETTINGS frames? I'm unclear what you mean by being
orthogonal to the remainder of the spec.


>
>
> Editorial Nits:
>

Thanks. Updated some of them at
https://github.com/mbelshe/SPDY-Specification/commit/e940e5d3c1cc930e52bd30b7e03f78573c0e04ad
.


>
> Boilerplate: month name needs to be a full name, such as "February"
>

Done


>
> Abstract: should not contain references (so just remove the "[RFC2616]")
>

Done


>
> (speaking of which this should really reference HTTPbis)
>
> Terminology: "header" -> "header field"
>

Where?


>
> 2.2.1 Control Frames
>
> has "...see Control Frames for the complete list..."; this should be a
> proper reference (to where?)
>

Not done yet.


>
> 2.2.2 Data frames
>
> s/MUST send issue/MUST issue/
>

Done


>
> 3.2.1
>
> Cites RFC 1738 for URI syntax; should cite RFC 3986.
>

Done


>
> 3.3
>
> Example host names should use the names reserved for this purpose.
>

Which ones are those?


>
> s/disc cache/cache/
>

I assume you meant s/disk cache/cache/. Done.


>
> 10.
>
> The TLSNPN reference needs to use the proper reference format for Internet
> Drafts. I also note that the referenced spec has expired a few months ago;
> if this is an integral part of SPDY we need to figure out how to make
> progress on it.
>

Agreed. It languished for awhile, but I think we need to revive the TLSNPN
discussions.