On Sun, Feb 23, 2003 at 08:31:00PM -0800, Dan Kohn wrote:
Of course, this should be defined as a C-T-E. Since Section 5.2.2 of
RFC 2046 restricts message/partial to be 7bit only, it would also seem
worthwhile to specify an application/news-partial with nearly identical
semantics, except for allowing use of 8bit content. You would need to
specify whether non-8BitMIME gateways were allowed to re-encode the
As I have indicated before, there is no purpose for message/partial
in USENET news, and thus no reason to do a news/partial form.
There are two reasons you might want to break a large file up into
components:
a) You are going to transmit the file over a link that has
technical problems carrying a large file, and so must
disassemble and reassemble it to fit within the technical limits.
b) You wish to bypass a policy restriction on the size of files.
Well, we're hardly here to encourage reason (b). In general, people
setting policy restrictions on file sizes want to know the full size
of any article, and want to decide to accept or reject it for policy
reasons based on that. They don't want to be deliberately mislead
about the size -- or have the size be available but hard to figure
out, because one must process all the parts to get it, usually defeating
the resource-saving purpose of the policy.
So that leaves (a). This is a link issue. When it comes to regular
USENET, there are few if any technical reasons for a limit below
a gigabyte with today's equipment. Every protocol, ever tool,
we have has no reason to be unable to handle an article of any size
up to the file size limit of the host operating system. Older tools
may have had limits, of course, but there is no reason for new ones
to. Of course some OSs limit files to 2GB.
However, individual links may have problems. It is the wrong design
to expect the poster and the newsreader to know and solve the problems
of the large number of intervening links. Packet disassembly and
reassembly is the job of the router even in IP, the most end to end
of protocols. Wisely so, as the sender doesn't know the vagaries of
the right chunk size for any given link.
So it's time to start telling the protocols that carry USENET to
do any disassembly and reassembly needed if they can't handle large
files, and to make that invisible to the format and the higher level
protocols.