Re: [Ltru] UTF-8 In Registry Considered Harmful, or Doom! Doom! I Cry Doom!

I have some qualms along the same line. I don't think we should expect IANA
to do anything but paste some text into a position in the file. (Frankly, I
think even that is too much -- it would be better if we just handed them a
replacement file which we've reviewed, but people didn't like that idea.) So
processing NCRs and extracting UTF-8 is too much.

The only process for producing UTF-8 that seems tenable to me is where the
text is sent as an email file attachment, where the attachment is in UTF-8.

Mark

On 3/20/07, John Cowan <cowan@ccil.org> wrote:
>
> I wish to oppose in the strongest terms the proposal to change the
> registration process in RFC 4646bis to publish the registry in UTF-8.
> This is *not* a matter of a one-time conversion from an ASCII I-D to a
> UTF-8 registry.  It also affects every addition and update thereafter.
>
> The essential link in adding and modifying subtags is an email sent from
> the Language Subtag Reviewer to IANA requesting the change.  Now either
> (1) that email will contain NCRs or (2) it will not.  In case (1), a
> *continuing* capacity to convert the NCRs to UTF-8 on the part of IANA is
> assumed, something they seemingly do not have today.  Are we confident
> that this will reliably be performed as part of registry updating?
> We are creating a whole new risk for ourselves.  I foresee a time when
> the registry contains a mixture of UTF-8 and NCR representations.
>
> In case (2), the email contains the characters directly.  Then either
> (2a) it is in UTF-8, or (2b) it is in some other encoding sufficient
> to represent the particular characters needed.  In either subcase, we
> are taking risks: email is notoriously the most i18n-broken part of the
> Internet, and probably every one of us has received at least one broken
> email of this kind.  In subcase (2b) we also add the risk that IANA (or
> their mail-processing tools) will fail to interpret the encoding, or will
> interpret it wrongly due to problems at the sender's end.  We are, once
> again, creating a whole new risk for ourselves.  I foresee a time when
> the registry contains a mixture of UTF-8 and non-UTF-8 representations.
>
> There is also the matter of the implicit contract which RFC 4646
> created about the registry: that it would be in ASCII with NCRs.
> Are we *convinced* that we have *no choice* but to break this contract?
> The current system may not produce pretty results, but it *works*.
> Let's not break it gratuitously.
>
> Yours emphatically,
>
> --
> Mos Eisley spaceport.  You will never           John Cowan
> see a more wretched hive of scum and            cowan@ccil.org
> villainy -- unless you watch the                http://www.ccil.org/~cowan
> Jerry Springer Show.   --georgettesworld.com
>
> _______________________________________________
> Ltru mailing list
> Ltru@ietf.org
> https://www1.ietf.org/mailman/listinfo/ltru
>

-- 
Mark