Re: [apps-discuss] draft-kucherawy-rfc5451bis

[Scott Kitterman <[e-mail]>]

I tried out "auth"="pass" and my implementation chokes on it.  I'd strongly 
prefer not to have to change it and having a single canonical form is 
definitely better.  So my answer to Alessandro's question about how many 
parsers would have to be updated is "at least one".

Scott K

On Sunday, March 24, 2013 01:04:07 PM Alessandro Vesely wrote:
> There's nothing intrinsically wrong with "value", albeit it is sometimes
> a nuisance not to have a canonical form.  The point is how many A-R
> parsers will need to be updated to comply with it, since "dot-atom"
> didn't admit quotes.
> 
> Another point is about extensions.  I proposed "Keyword" because it
> matches all the registered methods, and thus looks like what may seem to
> be the spirit of the spec.  If 5451bis will say "value", then the
> appointed expert will have no reason to oppose methods with long names
> and properties, comprising spaces and punctuation like some sort of
> filenames which require composition assistance in order to be spelled
> correctly.  Is that what we want?
> 
> On Sat 23/Mar/2013 21:31:22 +0100 Murray S. Kucherawy wrote:
> > What would be wrong with "dkim"="pass"?
> > 
> > The main difference between "token" and "value" is that the latter
> > permits quoted strings.  I couldn't think of a good reason to proscribe
> > those, and it's not like it makes parsing any more difficult; things
> > that can parse "value" have been around for years.
> > 
> > -MSK
> > 
> > 
> > On Sat, Mar 23, 2013 at 11:56 AM, Alessandro Vesely <[e-mail]
> > 
> > <mailto:[e-mail]>> wrote:
> >      On Fri 22/Mar/2013 20:22:43 +0100 Murray S. Kucherawy wrote:
> >     > I've revised the grammar in the next draft to avoid this
> >     
> >     ambiguity.  Let
> >     
> >     > me know if it works.  Essentially, "dot-atom" has been replaced by
> >     > "value", which disallows the case you've illustrated here.
> >     
> >     Hm... that way it becomes possible to have "dkim"="pass".  I'd beg for
> >     "token", which is somewhat simpler, but it allows dots.  Wouldn't it
> >     be
> >     possible to have a grammar that matches the actual use?  I mean, e.g.:
> >     
> >     authserv-id = domain-name
> >     
> >          method = Keyword [ [CFWS] "/" [CFWS] version ]
> >          result = Keyword
> >        
> >        property = Keyword
> >     
> >     "Keyword" is defined in RFC 5321, as well as "domain-name".  (It seems
> >     to be useless to recall the definition from RFC 6376, as the domain
> >     literal alternative that was in RFC 2821 has gone away.)
> >     _______________________________________________
> >     apps-discuss mailing list
> >     [e-mail] <mailto:[e-mail]>
> >     https://www.ietf.org/mailman/listinfo/apps-discuss
> > 
> > _______________________________________________
> > apps-discuss mailing list
> > [e-mail]
> > https://www.ietf.org/mailman/listinfo/apps-discuss
> 
> _______________________________________________
> apps-discuss mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/apps-discuss