ietf-nntp Wildmats

[Russ Allbery]

Tim Roberts <troberts at bigfoot.com> writes:

> If "\" is not treated as an escape character in a set - does that
> exclude the use of "\s" in a set?

Yes.

> I'm also assuming the UTF-8 encoding is done PRIOR to set processing
> (otherwise the same applies to UTF-8 characters)..

Per our current writeup, \u isn't allowed in sets either.

We could introduce \ as an escape character into a set, but that's an even
more substantial departure from existing wildmat syntax than we've already
made.  That changes the meaning of [\-~], for example, and previously
valid wildmats like [/\] become invalid.

It's worth bearing in mind that complicated character sets are a fairly
infrequently used feature at present, although I know some clients use
simple ones for case insensitivity.  With UTF-8, you can just put the
literal UTF-8 characters in the set.

I'll also throw this out:  At some point, you end up complicating wildmat
so much to make it more useful to PAT that it may be a better idea to just
introduce a completely new extension called MATCH or SEARCH or something
that uses regular expressions instead (which are more powerful than
wildmats).

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