ietf-nntp Standardization of LIST OVERVIEW.FMT

Andrew Gierth andrew at erlenstar.demon.co.uk
Sat Apr 5 07:53:11 PST 2003


>>>>> "Matthias" == Matthias Andree <matthias.andree at gmx.de> writes:

 Matthias> I'd rather do the opposite and make the protocol really
 Matthias> 8-bit clean no matter what,

this has been discussed in the past.

The protocol cannot be made binary-clean as it stands because the
article is required to end with a CRLF _which is part of the article
data_. Fixing this requires pretty sweeping changes.

The number of sites that will convert bare LF to CR LF is still
extremely large, and outlawing it would have the effect of outlawing
the traditional spool format, which is pretty much a non-starter.

 Matthias> One thing to keep in mind is that implementations WILL HAVE
 Matthias> to check that the line ends at CR LF and *ONLY* at CR LF,
 Matthias> not at LF or CR. Some implementations will goof this up for
 Matthias> sure, but these are IMHO broken and have always been, as CR
 Matthias> LF has always been the "wire format" for lines.

There is a known issue with the sequences <non-CR> <LF> <.> <LF> and
<non-CR> <LF> <.> <CR> <LF>; some servers (incorrectly) recognize
these as terminators and others do not, which allows NNTP commands to
be embedded in the article. In response to this there are now servers
which will drop any article containing either of the above.

So my preference is still to treat articles containing bare LF as
ill-formed.

-- 
Andrew.



More information about the ietf-nntp mailing list