- From: Larry Masinter <masinter@parc.xerox.com>
- Date: Mon, 13 Nov 1995 13:33:22 PST
- To: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
If you want to encode byte ranges into URLs, you should also encode
the unique ID of the actual data stream.
So, if I start to GET http://slow.host.dom/images/logo.gif HTTP/1.0
and then cancel the transfer, and then later on want to get the rest
of the data stream, I shouldn't use
"http://slow.host.dom/images/logo.gif;bytes=500-"
but rather
"http://slow.host.dom/cid:0102345@slow.host.dom#bytes=500-"
where
0102345@slow.host.dom
is the content-id URL for the content that was originally being
delivered with /images/logo.gif in the first place.
That is, the URL for partial content needs to identify the exact
content from which the remaining bytes are to be extracted.
Note that "#" is currently illegal in a URL as it is the separator
between the URL and the client-side selector of the URI. In this case,
you could propose that when the server is willing to retrieve partial
content that the server could perform what is logically a client-side
extraction, it's allowed when applied to a uniquely identified
content.
Received on Monday, 13 November 1995 13:38:12 UTC