W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > September to December 1995

Re: draft-luotonen-http-url-byterange-00.txt

From: Mark Brown <mbrown@openmarket.com>
Date: Wed, 27 Sep 1995 15:39:42 -0400
Message-Id: <199509271939.PAA12215@breckenridge.openmarket.com>
To: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Cc: mbrown@openmarket.com

In the course of implementing the bytes parameter in Open Market's
WebServer I noticed the following things about the draft:

* "separated by a colon" should read "separated by a semicolon"

* The draft is silent on how the server should respond to a
   very large integer parameter, e.g.

		http://host/dir/foo;bytes=-123456789012345678901234567890

   This might be treated as a syntax error or might be treated as the
   largest integer that the server can represent.

* The draft says that large second numbers should be treated as referring
   to the end of the document.  The draft also says that the first
   number must always be less than or equal to the second number.
   Suppose that dir/foo is a 1-byte file and the request is

                http://host/dir/foo;bytes=9-10

   Using the first rule the server transforms the request into

                http://host/dir/foo;bytes=9-0

   Is this now an error according to the second rule?  Or should
   the server give some arbitrary response, e.g. return the last
   byte of the file?

* My reading is that the characters of the bytes parameter, including
   the semicolon, equal sign, minus, and comma, do not need to
   be unescaped before interpretation.  While this is really a URL
   encoding issue rather than a bytes parameter issue, it wouldn't hurt
   to clarify this point in the draft.

    Mark R. Brown
    Open Market Inc.
    245 First Street
    Cambridge MA 01242
Received on Wednesday, 27 September 1995 12:42:34 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:33 EDT