Re: [Encodings] IANA transition

Hello Larry, others,

On 2014/07/01 10:05, Larry Masinter wrote:
> I take the ‘Encodings’ document as evidence that all is not well in the way the community interacts with IANA, if the IANA character set registry isn’t sufficient to define authoritatively how charsets should be used and to also inform implementors how they ARE used.

We could go back to check the relevant mailing list for the discussions 
around changes to registrations in the IANA character set registry. I'm 
quite sure that wherever there was some "community" proposing a change, 
there was also some "community" asking serious questions or more 
directly opposing the change for good reasons. Or there were just not 
enough comments supporting the change that would have made it prudent 
for the expert reviewer(s) to make the changes without risking to 
severely affect many third parties.

As an example, a proposal to define "US-ASCII" as essentially being 
'windows-1252', if ever made on the relevant list, would sure have 
raised quite a lot of opposition for various reasons.

To give a broader background, the goal of a single series of encodings 
with uniformly interpreted labels is an interesting one. But neither the 
IANA registry nor e.g. the Unicode Consortium nor anybody else according 
to my knowledge have been successful in weeding out minor differences 
among encoding implementations. Because any changes in such 
implementations risk to affect a lot of legacy content, the chance that 
the spec we are currently discussing will achieve that are likewise 
rather slim, even if we declare success when all the major browsers conform.

This is not a problem with IANA or the IETF or the Unicode Consortium or 
the WHATWG or W3C, but a result of the dynamics of the encoding 
technology. Just as browsers have a strong and almost physical tendency 
to become more and more bugwards compatible, because users just change 
from strict to less strict browsers, encoding implementations have a 
very strong tendency to stay the way they are, because annoying current 
users is much more of a problem than annoying potential users.


> And a policy of “just use utf8” would argue that it might be OK to close the charset registry.

Well, it's still in use for those protocols and implementations that use 
stuff other than UTF-8. But given that the pace of additions to the 
registry in recent years is lower than one registration per year, a 
formal close isn't really needed.

> Other registries like those for MIME types and URI schemes having similar problems but not being OK to close to additions.

For many of these registries, the main problem is that in the age of 
github, the (few, but non-zero) steps that an usual registration takes, 
and the time periods involved to give interested parties a chance to 
comment, seem too much work for many people. This hits registries for 
open-ended textual labels like the above the most, because the chance 
for collisions is too low to be perceived as an issue.

Regards,   Martin.

> ---------- Forwarded message ----------
> From: IETF Secretariat <ietf-secretariat@ietf.org<mailto:ietf-secretariat@ietf.org>>
> Date: Mon, Jun 30, 2014 at 5:34 PM
> Subject: New Non-WG Mailing List: ianaplan -- IANA Plan
> To: IETF Announcement List <ietf-announce@ietf.org<mailto:ietf-announce@ietf.org>>
> Cc: ajs@crankycanuck.ca<mailto:ajs@crankycanuck.ca>, ianaplan@ietf.org<mailto:ianaplan@ietf.org>
>
>
> A new IETF non-working group email list has been created.
>
> List address: ianaplan@ietf.org<mailto:ianaplan@ietf.org>
> Archive: http://www.ietf.org/mail-archive/web/ianaplan/
> To subscribe: https://www.ietf.org/mailman/listinfo/ianaplan
>
> Purpose:
>
> This is a list for discussion of the IETF's position on the IANA
> oversight function. In 2014, the NTIA announced that it was planning
> to step out of stewardship of the IANA function, on the condition that
> the relevant communities were able to agree on new stewardship for the
> various IANA registries. NTIA has been clear that the IETF is one of
> the affected communities, and that its views are therefore important
> for NTIA's evaluation of any proposal. ICANN is convening a process
> involving many interested parties, including the IETF. This list will
> be the place for the IETF to discuss the proposals.
>
> For additional information, please contact the list administrators.
>

Received on Wednesday, 2 July 2014 11:45:26 UTC