Re: Working Group Last Call for HTTP Random Access and Live Content

On 9 February 2018 at 12:38, Mark Nottingham <mnot@mnot.net> wrote:
> Hi Folks,
>
> We didn't receive any feedback during the WGLC period, which is causing your chairs a bit of concern.
>
> If you have read this document and believe it should progress as Experimental, please say so on-list (or privately to us, if necessary).
>
> Thanks,
>

For the most part, I can think of a couple of ways it could go wrong,
but they're all pretty screwy to start with, and the worst that
happens is a dodgy client finds that its "very large" resource has
been truncated mid-download.  So I don't think it really needs to be
Experimental just for that.

That said, I don't feel comfortable about the Shift Buffer
Representations section.  AFAIK this document is introducing this
concept, at least as far as breaking the relationship between resource
representation size and byte range, so it should spend more effort
describing/standardising some of the other behaviours.  (Such as: What
happens if I try to resume an interrupted non-range request using a
range request (assuming I didn't make the first request when the
original first byte was still available)? Whose byte "0" is the real
byte "0"?)  It seems like we've opened a little Pandora's box without
providing any advice or guidance.

Editorial:

* No references until Section 2.1? I would have expected an earlier
pointer to RFC7233 (at the least), maybe even in the introduction.

* Throughout, "indeterminate-length" (or "indeterminate length") is
used as an adjective, as in "indeterminate length resource".
Technically is *is* an attributive noun, but it's awkward, especially
when not hyphenated. What about "resource of indeterminate length"?

* From Section 2.1 on there's a bunch of ABNF, but there's no
fore-warning of that anywhere in the document. Is it okay to defer
that entirely to the normative reference [RFC7233]?

* In 2.2: "Very Large Value", with caps, looks like it should have a
definition somewhere ;)  (Also it's not consistently capitalised
throughout the document, one way or the other.)

* "2^^63" is new notation to me. Usually it's either "2^63" or "2**63".

Cheers
-- 
  Matthew Kerwin
  http://matthew.kerwin.net.au/

Received on Friday, 9 February 2018 07:47:17 UTC