From: Brad Templeton (brad@templetons.com)
Date: Wed Jun 02 1999 - 14:15:06 CDT
On Wed, Jun 02, 1999 at 08:08:46PM +0200, Brad Knowles wrote:
> At 10:46 AM -0700 6/2/99, Brad Templeton wrote:
>
> > This only has value when the
> > news database and news transport are independent, which exists but is
> > not that common. However, it's impossible with headers in the body.
>
> Not true. Just store the bodyparts recursively, after parsing the MIME.
I think you must have misread the text you removed. The thing that is
"impossible" is avoiding parsing the body! So one can hardly say, "of course
that is possible, just parse the body."
> In fact, if you do this, this is a *major* win in terms of disk
> space, because while there may be large numbers of copies of messages
> posted with the same body but varying headers, if they all really do
> have exactly the same body, then that body need only be stored once.
Huh? Large number of messages with identical bodies? You mean control
messages with blank bodies? Spam?
>
> Of course, all this is totally off-topic and we should table
> further discussion.
The way to attain that goal is not to reply. It's a sad rule of debate
that if you want to end a debate you volunteer not to have the last word!
However, while I agree we are not about discussing the mechanisms about
signing, the question still in the air is does the spec say that news
transports and injectors MUST not alter static headers in any way, and
MUST put any headers they will add at the top, so that we have the following
ordering
a) Local
b) Variant
c) Added, static
d) Original poster/injector (all static)