[ietf-nntp] :bytes metadata

[Ken Murchison]

Charles Lindsey wrote:

> In <3FE30A49.6070906 at oceana.com> Ken Murchison <ken at oceana.com> writes:
> 
> 
>>Agreed.  But if we're just going to state that its a <= approximation, 
>>we probably don't need to get into the storage format.  If what we're 
>>saying is true, then the client shouldn't/won't care whether its stored 
>>in wire format, canonical format, base64 encoded or AES encrypted, as 
>>long as its gets a reasonable estimate of the canonical size.
> 
> 
> I think you would get a rather poor estimate using base64, but I don't
> think our protocol gives that option (or not in a manner visible to the
> user).

I was just using that to illustrate a point, I wouldn't imagine that the 
article would be stored in any form other canonical or wire format.

FWIW, I think that the server should be required to report the :bytes of 
the resulting canonical form.  But it sounds like some servers are 
either reluctant or incapable of doing so, so I proposed that the server 
can return an estimate of the size as long as its >= the size of the 
canonical form (provided that clients don't care if the :bytes is 
completely accurate).

-- 
Kenneth Murchison     Oceana Matrix Ltd.
Software Engineer     21 Princeton Place
716-662-8973 x26      Orchard Park, NY 14127
--PGP Public Key--    http://www.oceana.com/~ken/ksm.pgp