RE: Custom Ranges

What I remember was that I pushed for custom ranges and that there was
a lot of push-back from people who thought it was too much complexity.

I think the idea 'sort of' got into the spec, but not fully fleshed out.

I agree that it belongs in the issue list, to either clarify how to
use custom ranges (with a range unit registry, for example) or else
to remove the feature.

Larry



-----Original Message-----
From: ietf-http-wg-request@w3.org [mailto:ietf-http-wg-request@w3.org] On Behalf Of Yaron Goland
Sent: Wednesday, August 29, 2007 2:31 PM
To: Kornel Lesinski; ietf-http-wg@w3.org
Subject: RE: Custom Ranges

I ran into this exact same problem. In the end we choose another approach but the lack of a definition in RFC 2616 on how to extend ranges was a problem. Another big problem is that certain servers (who shall remain nameless) seem to believe that the range values must be integers. I don't see anything in RFC 2616 that seems to require that. But RFC 2616 does seem ambiguous on the subject. This would seem like a good issue for the HTTPBis issue list.

        Thanks,

                Yaron

> -----Original Message-----
> From: ietf-http-wg-request@w3.org [mailto:ietf-http-wg-request@w3.org]
> On Behalf Of Kornel Lesinski
> Sent: Friday, August 24, 2007 4:15 PM
> To: ietf-http-wg@w3.org
> Subject: Custom Ranges
>
>
>
> I'm writing application that would benefit greatly from being able to
> use
> Range requests with custom units and range definitions instead of
> bytes.
>
> The RFC 2616 seems to suggest such possibility in 3.12 Range Units:
> there's a "other-range-unit" defined.
>
> However definition of Content-Range uses "ranges-specifier" and Range
> uses
> "content-range-spec", which both seem to allow only byte ranges.
>
> In such case, is there any use for "other-range-unit" in Accept-Ranges?
>
> If I wish to use custom ranges, should I stay clear of Range/Content-
> Range
> and status 206? What is the best alternative? (custom headers + Vary?
> Piggybacking my range spec on E-Tags?)
>
> --
> regards, Kornel Lesinski

Received on Friday, 31 August 2007 15:41:06 UTC