ietf-nntp NNTP doubling of starting periods

Keith Moore moore at cs.utk.edu
Thu Nov 14 15:21:32 PST 1996


> RFC 977 specifies the following for lines that start with a period (.)
> > [...]
> Is this still in use? It seems unnecessary to distinguish lines that
> contain *only* a period (etx) from text lines that start with a period. Any
> parser should be able to recognize that <cr><lf>.<cr><lf> is the end of the
> data stream and treat lines that start with a period as normal text lines.
>
> I tried sending lines that started with 1-5 periods on a line through INN
> and it passed them unaltered as the second paragraph suggests it should.
> 
> If this is not used should it be phased out for the sake of
> transparency? 

The current scheme provides transparency.  Senders add a dot to every
line that begins with a dot.  Receivers remove the initial dot from
any line that begins with a dot.  It's simple, it works, and anyone
can understand how to do it right.

Changing the spec can only lead to confusion and breakage.
It's not broken, so don't fix it.

Keith







More information about the ietf-nntp mailing list