ietf-nntp Issue: number range rollover

[Jack De Winter]

At 05:24 PM 12/31/96 GMT, chrisy at easynet.net wrote:
>On Sat, 28 Dec 1996 22:49:00 +0000 (GMT), USENET news manager
><newsmaster at ucs.cam.ac.uk> wrote:
>
>>(1) If anyone's reading news via NFS or direct local spool access on the
>>server (yes, it still happens!), the news server has no way of knowing 
>>when clients have stopped using the old numbers.
>[etc]
>
>You should probably stipulate that "this is bad", or have some other
>mechanism for indicating to file-based clients that the numbering has
>changed. Readers are going to have to be adapted anyway...
>
>...so why not rewrite all the readers with an agreed maximum HWM and
>rollover to 0 like a FIFO buffer. It's obvious when this is the case -
>the HWM is less than the LWM. If IBM could code it into a 16kb PC BIOS
>ROM for 16-byte buffered keyboard handling I'm sure you can implement
>it with 32-bit numbers in a news system. No renumbering required [tm].
>It's all a question of the maximum HWM being agreed on.
>
>This then requires little other adaption of the existing systems,
>server and client.

This system would work if we went with the article numbering system
of empty groups being first <= last, count = 0 approach.  In that
case, rollover could be detected as a special case and handled
separately.  It would work with the first = last + 1 approach,
but possibly not as cleanly.

regards,
Jack
-------------------------------------------------
Jack De Winter - Wildbear Consulting, Inc.
(519) 576-3873		http://www.wildbear.on.ca/

Author of SLMail(95/NT) (http://www.seattlelab.com/) and other great products.