Re: WGLC: p5 editorial nits

[Julian Reschke <[e-mail]>]

On 2013-03-28 20:36, Ken Murchison wrote:
> Hi All,
>
> Sec 1, para 2:  The phrase "obsoleting those parts previously defined in
> [RFC2616]" seems unnecessary since the document boilerplate already
> states as such.  Also, parts 4 and 6 which similarly describe optional
> facilities don't bother with such text.
>
> Sec 2.1, postscript to suffix-byte-range-spec ABNF:  The sentence
> immediately following the description of how to handle a representation
> shorter than the suffix-length begins with "For example..." .  One would
> think that the example would demonstrate over-sized suffix-length case,
> but it doesn't.  I don't think such an example is needed, but I think
> the wording is misleading.  I would suggest removing the "For
> example..." sentence and replacing it with "Examples of
> byte-ranges-specifier values:" or "Additional examples of
> byte-ranges-specifier values:" or "Examples of byte-ranges-specifier
> values using suffix-byte-range-spec:" like is done for the discussion of
> byte-range-spec.
>
> Sec 4.4, Note: suggest changing "... many implementations will simply
> respond with 200 (OK) ..." -> "... many implementations will simply
> respond with the entire selected representation in a 200 (OK) response ..."

-> <http://trac.tools.ietf.org/wg/httpbis/trac/changeset/2217>

> General implementation question that I didn't find answered in the
> document:  Can a server that receives a request for multiple ranges
> reply with only a single part corresponding to the first satisfiable
> range?  Or can it coalesce all satisfiable ranges into a single part
> regardless of the overlaps/gaps?

Good question. A perfect client will understand whatever, as the message 
is self-describing. In practice, clients might get confused if they 
don't get exactly what they asked for.

Best regards, Julian