W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2008

RE: issue 85 - range unit extensions

From: Yves Lafon <ylafon@w3.org>
Date: Wed, 27 Aug 2008 08:15:32 -0400 (EDT)
To: Martin Duerst <duerst@it.aoyama.ac.jp>
cc: Larry Masinter <LMM@acm.org>, 'Lisa Dusseault' <lisa@osafoundation.org>, 'Julian Reschke' <julian.reschke@gmx.de>, 'Mark Nottingham' <mnot@mnot.net>, 'HTTP Working Group' <ietf-http-wg@w3.org>
Message-ID: <Pine.LNX.4.64.0808270805460.9860@ubzre.j3.bet>

On Wed, 27 Aug 2008, Martin Duerst wrote:

>>>> On the other hand, I don't understand the reasons that the
>>>> JPEG 2000 committee rejected using HTTP range retrieval for
>>>> the J2 protocol, instead adopting query parameters in the URLs.
>>>> In this case, the range retrieval might look like it was
>>>> media independent even though J2 defined it specifically for
>>>> JPEG 2000 encoded images.
>
>>> I have no clue, but it could be that one argument for not using
>>> ranges is that you cannot expose them in an URI. This means that
>>> instead of sending your colleague a mail saying "Look at this
>>> interesting snippet: [URI here]" you have to say "Look at this
>>> video [URI here], the interesting stuff is at xx:yy:zz into the
>>> video".
>>
>> Well, when fragments are used to identify part of an HTML document, the 
>> fragment is not sent at the HTTP level, although in the case of HTML 
>> the whole document is always fetched, but in the case of pictures or 
>> video, being able to access a part of the image using ranges derived 
>> form the URI instead of dowloading the whole document before finding 
>> the fragment would be far more efficient (like accessing XMP fragment 
>> inside a picture, 10s out of a 1h movie in fullHD).
>
> Yves, you are talking about fragment ids, which have the disadvantages
> you mention, but Larry says they are using query parts, which are
> sent to the server.

Well, to me it is not a disadvantage, as it clearly identifies the 
partial content of the resource you want to access without creating a new 
resource (like what happens using query parameters).
It just puts the logic of querying the fragment in the client (via the use 
of HTTP range) instead of asking the server to do it via query parameters.

Both solutions (query parameters or fragment) allows identification of the 
same binary content. I was just replying to the argument for not using 
ranges that it is not (easily) possible to surface ranges in the URI 
space.

-- 
Baroula que barouleras, au tiéu toujou t'entourneras.

         ~~Yves
Received on Wednesday, 27 August 2008 12:16:07 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:50:54 GMT