Comments on Byte range draft
Ted Hardie (hardie@merlot.arc.nasa.gov)
Thu, 9 Nov 1995 15:50:18 -0800 (PST)
In the range specification section it says:
* In the case that the second integer is smaller than the first
one, an empty range is returned.
Since the specification says just above that point that:
* The first integer must always be less than or equal to the second
one.
I would suggest that in cases where the second integer is smaller than
the first, an error message indicating a malformed header should be
returned (rather than an empty range). An empty range doesn't tell
the requester what went wrong; an error message might.
In the section on the return of multiple ranges as multipart MIME
messages, the draft says "A server may send also a single byte range
as a multipart message." Why should a server send a single byte
range in a multipart message, and would that break any clients expecting,
not unreasonably, multiple parts to a multi-part message?
I also note that the draft gives the response Range header as
Range: bytes x-y/z . Is z in use to handle situations where a
new version has occurred between the clients request of one byte
range and another, and, if so, isn't that already handled by the
other headers (you note that the if-modified-since works with
this)? I also assume that the Last-modified header and if-modified
since header will generally be applied to the document as a whole,
not to the individual byte-ranges, but I'm afraid I didn't find
the draft to be too clear on this point.
Regards,
Ted Hardie
NAIC