[ietf-nntp] :bytes metadata

[Russ Allbery]

Clive D W Feather <clive at demon.net> writes:

> It seems to me that the most likely use for :bytes is to reserve space
> to hold a copy of the article. For that, it's more important not to have
> too small a number than too big a one.

The only thing that it's used for in practice so far as I know is
killfiling and rough size estimates.  Do you have any reason to believe
that it's used for anything more important than that?

I think we're putting too much effort into something that clients don't
actually care about.

> What I've currently got, *knowing* that it's insufficient, is:

>     The :bytes metadata item for an article is a decimal integer.  It
>     SHOULD equal the number of octets in the entire article - headers,
>     body, and separating empty line (counting a CRLF pair as two octets,
>     and not including either the "." CRLF terminating the response nor
>     any "." added for "byte-stuffing" purposes).

I'd really like to see some concrete reason why we need to say more than
that before putting a lot of additional effort into this.

> The parenthetical comment above shows some reasons for variation, and
> this provides some reasonable room for servers without making it useless
> for clients. In particular, counting CRLF as one octet, or including all
> the stuffed dots, will fit within that limit, provided that the
> variation in headers plus the number of header lines is no more than
> 999.

> If any server can't keep within that limit, I suggest they need to rethink
> their strategy.

You're making the assumption that servers should change their architecture
to make sure that :bytes is roughly accurate.  If nothing really cares
about that value, that's a bad assumption.

BTW, I don't seem to be getting any e-mail from the mailing list right
now.  I'm going to try to figure out why that is.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>