ietf-nntp Draft summary of IETF 48 meeting

[Andrew Gierth]

>>>>> "Russ" == Russ Allbery <rra at stanford.edu> writes:

 >>> (note - this is not what INN did last time I looked, it just replaced
 >>> CR,LF,TAB one-for-one with spaces.)

 >> Yes, and current INN still does this behavior.  I've looked past
 >> versions of INN since 1.4.  I see only 1.4 replaces LF and TAB.  All
 >> others behave like above.

 Russ> Which does raise the question of which is correct... is there a
 Russ> specification anywhere?  I assume that the behavior Andrew is
 Russ> describing is from the reference implementation?

No, I'm describing what we currently have in the spec.

I am not convinced of the correctness of the reference implementation
in this context (when conjuring up XOVER or XHDR data from an article,
it does not seem to do anything about tabs, and converts folds into a
single space). It doesn't have XPAT at all, so there's no comparison
to make there, and neither does it generate overview data itself in
normal circumstances (leaving that to the underlying server).

INN's behaviour is in line with its predecessors (other than perhaps
in its handling of CRs), as I understand it from the somewhat sketchy
documentation available (e.g. newsoverview.3 in C News). The
collapsing of non-printing sequences appears to have been introduced
on this list; the wording change that avoided collapsing printable
whitespace sequences was in response to an objection from me.

-- 
Andrew.