From: Greg Berigan (gberigan@cse.unl.edu)
Date: Tue Sep 16 1997 - 00:00:02 CDT
<curtis_whalen@geocities.com> wrote:
>Greg Berigan <gberigan@cse.unl.edu>
>>Seth Breidbart <sethb@panix.com> wrote:
>>>> User-Agent = "User-Agent" ":" SP product 1*(SP product | SP
>>>> comment)
>>> Shouldn't that 1* be 0*? One product alone is enough.
>> Hmm, probably. Otherwise it makes either two products or a comment
>> mandatory. Yeah, make that "probably" a "definitely".
> User-Agent: GrandsonNewsReader32/0.05 CNNTPPoster/0.05
>
> Would this be legal?
Yes. The 0* means "zero or more instances of". Theoretically you could
have an infinite number of products mentioned. Still, the additions would
stop with the POSTING AGENT. INJECTING AGENTs can't modify it nor generate
one if not present (no addition of " INN" or the like to it). That would
be like proxies modifying it on the web. It isn't intended to be a total
audit trail of what software handles the article and is completely
optional. I'd even hesitate allowing inews to amend it.
>I see this type of thing in the HTTP equivalent with the libwww library
>provided by the W3 Consortium.
>That's why I'm asking.
Only if CNNTPPoster generated the header. Generally, the last application
is what generates it, possibly second to last, and must know all software
used up to the point it provides. For example, inews doesn't know or care
who uses it, be it tin or trn or slrn or whatever and shouldn't generate
it. Tin and slrn though could know they used pico or vi or emacs and be
able to include that information, or even include inews themselves.
Specific versions are optional.