W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > May to August 1997

Re: multipart/byteranges

From: Jeffrey Mogul <mogul@pa.dec.com>
Date: Mon, 04 Aug 97 10:27:20 MDT
Message-Id: <9708041727.AA03739@acetes.pa.dec.com>
To: Yaron Goland <yarong@microsoft.com>
Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com, Rajeev Dujari <rajeevd@microsoft.com>
    A question has come up, is it legal for a server to send a
    multipart/byteranges with a single range? This apparently causes some
    very ugly code on the client side if it is legal. The following is an
    excerpt from section 19.2 of RFC 2068:
    
    19.2 Internet Media Type multipart/byteranges
    
       When an HTTP message includes the content of multiple ranges (for
       example, a response to a request for multiple non-overlapping
       ranges), these are transmitted as a multipart MIME message. The
       multipart media type for this purpose is called
       "multipart/byteranges".
    
    My reading of this section is that a multipart/byteranges can only be
    sent if there are byte rangeS but there has still been some question.
    
    1) Is it correct that multipart/byteranges can only be used if there are
    multiple, as in more than one, byteranges?

    2) Either way, can we put in a sentence to clarify the matter?

I believe that I wrote this section, and it certainly never occurred
to me that handling a "multipart/byteranges" with only one part would
cause implementation difficulties.  After all, if the client is
prepared to deal with N parts in a multipart type, what makes N = 1
harder than, say, N = 2 or N = 13?

Can you perhaps explain a bit more about why this makes the client
code so ugly?  I'm not strongly opposed to changing the HTTP spec to
prohibit 1-part "multipart/byteranges" values, but I think doing
so would add complexity to the spec, and isn't really consistent
with the robustness principle.  Also, I'm certainly not a MIME
expert, and I don't know if MIME has a rule that either explicitly
requires or prohibits N = 1 in multipart types.

Have other client implementors found this hard? easy? didn't realize
until now that it might be a real problem?

-Jeff
Received on Monday, 4 August 1997 10:32:31 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:32:50 EDT