RE: [Ltru] Re: Remove extlang from ABNF?

I think #3 would be more than just slightly sub-optimal. Many people would look just at the statement of the ABNF and go away and do things on that basis. The statement of the ABNF should somehow indicate clearly that extlang is DEAD, never to be resurrected.

We need to be clear about the decision we’re making: we are deciding to abandon extlang *permanently*. Removing extlang from the ABNF will leave something that is more-certainly stable in that particular regard. Of course, it doesn’t make the ABNF as a whole stable; but do keep in mind: we would be removing from the ABNF something that was reserved for possible future use. We’ve decided to abandon that future possibility. That is not a destabilizing change on the same order as destabilizing some aspect of the ABNF that was already in use. (Granted, though, some other parties may be concerned about any form of instability.)

Peter

From: Mark Davis [mailto:mark.davis@icu-project.org]
Sent: Tuesday, December 04, 2007 7:50 PM
To: John Cowan
Cc: Frank Ellermann; ltru@lists.ietf.org
Subject: Re: [Ltru] Re: Remove extlang from ABNF?

It doesn't cause a problem. Any existing tag that does not match the new ABNF was, in fact, invalid under RFC 4646.

(The ABNF is the "convex hull" of the definition of valid language tags; all valid language tags must follow the ABNF, but the vast majority of strings that do fit the ABNF are not valid language tags.)

As far as I can see, we have the following alternatives:

 1.  Remove extlang from the ABNF.
 2.  Leave extlang in the ABNF, but document that it is "permanently reserved" (will never be used, a la ISO).
 3.  Leave extlang in the ABNF, and just leave in the documentation from RFC 4646 that it is not used in the current version.
Here is my take.

 *   #1 is the cleanest (for reasons given earlier).
 *   #2 is functionally identical to #1, but may feel warmer and fuzzier to some people.
 *   #3 we could live with, although it would be slightly suboptimal.
Mark
On Dec 4, 2007 7:36 PM, Mark Davis < mark.davis@icu-project.org<mailto:mark.davis@icu-project.org>> wrote:
Sorry; didn't mean to imply that. I meant that I agree with your discussion of why removing them is not an formal problem.

While I agree that we don't have to remove the syntax, removing them would be more than a nicety: it makes the syntax checking significantly simpler, as well as simplifying any APIs that are used with language tags.

Mark

On Dec 4, 2007 7:30 PM, John Cowan <cowan@ccil.org<mailto:cowan@ccil.org>> wrote:
Mark Davis scripsit:

> I agree completely with John. There are no possible current *valid* tags
> that use extlang -- it was only a possibility that we left open for the
> future. By removing extlang, we do no damage at all to existing data or
> code.
As I said, I don't favor removing them -- there is no gain from doing so
except a theoretical nicety.

--
Well, I have news for our current leaders       John Cowan
and the leaders of tomorrow: the Bill of        cowan@ccil.org<mailto:cowan@ccil.org>
Rights is not a frivolous luxury, in force       http://www.ccil.org/~cowan <http://www.ccil.org/%7Ecowan>
only during times of peace and prosperity.
We don't just push it to the side when the going gets tough.  --Molly Ivins

--
Mark

--
Mark