Re: [Ltru] [psg.com #1061] eliminate (or proscribe)Private Use Tags

["Randy Presuhn" <randy_presuhn at mindspring.com>]

Hi -

> From: "Dylan N. Pierce" <dylanpierce at megared.net.mx>
> To: <ltru at ietf.org>
> Sent: Friday, July 08, 2005 4:52 PM
> Subject: Re: [Ltru] [psg.com #1061] eliminate (or proscribe)Private Use Tags
>
> >  (4) "Private-use subtags require agreement between parties
> >   intending to use them. They MUST NOT be distributed beyond the scope
> >   of the relevant agreements. Subtags intended for wider
> >   distribution MUST be registered. Developers of private-use subtags are
> >   warned that, despite this prohibition, private-use subtags might
> >   "leak" beyond the scope of the agreement defining them.  This means that
> >   applications making use of private-use tags need to be able to cope with
> >   conflicting (same character sequence but differing semantics) and unknown
> >   private-use tags.
>
> I have no problem with any of the changes you made here; just one point
> on an omission which I'm willing to abandon immediately if my case is
> unconvincing. Regarding the line in my original:
>
> "There MUST NOT be consequences for any application which chooses to
> ignore private use tags."
>
> This, in my mind, was a protection against the MS-JVM phenomenon, where
> an organization with wide product distribution can implement its own
> private refinements, and then, simply through product saturation, win a
> position with products that don't quite "work" without their
> refinements. To make sure that even an enormous company or organization
> can't, simply through product share, implement a fait accompli
> adaptation of the standard.
>
> If that possibility doesn't bother anyone, I willingly abandon the
> petition on that line.
...

To clarify...
As a technical contributor, I had dropped the "consequences" sentence
because I couldn't figure out exactly what it was supposed to mean,
particularly in the RFC 2119 interpretation of the "MUST NOT" keyword.
There isn't anything we can really do about a private-use subtag that
enters the wider realm of discourse.  Those familiar with SMTP and NNTP
headers may be able to think of similar situations from the past.  :-)
It's one of the inherent risks of having this kind of mechanism, but it's
already a legacy problem.

Randy




_______________________________________________
Ltru mailing list
Ltru at lists.ietf.org
https://www1.ietf.org/mailman/listinfo/ltru