ietf-nntp Draft summary of IETF 48 meeting

[Russ Allbery]

Charles Lindsey <chl at clw.cs.man.ac.uk> writes:

> OK, I more or less understand PAT now, except for that bit. Yes, it is
> not pretty, but neither are the alternatives. It would be worse to allow
> "2 spaces" between the wildmats to mean "match exactly 2 spaces here".

I guess I don't understand why that would be worse.

> How do you match a Subject line with a fold in the middle? And so on.

You'd have to use ?? to match the CRLF.  Hm.  That's probably another good
argument for \-escapes for characters.  But that's actually a bigger
problem; in my experience a lot of clients mangle folded headers in a lot
of ways, so you're likely to end up with a lot of variations on the
folding space.  On the other hand, an escape to match any amount of
whitespace just feels like a very non-wildmat sort of thing.

> 1. What do existing implementations of XPAT do (the spec seems to
> suggest that they match arbitrary whitespace)?

INN collapses all spaces down to a single space, which then matches a
literal single space.

> 2. What else would you suggest? The only real alternative seems to be to
> introduce /s or /040 or something (but how do you match a fold then)?

With a generic \-escape, you could match the fold provided that no one
changed it.

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