I-D Action:draft-ietf-idnabis-mappings-00.txt

Elisabeth Blanconil eblanconil at gmail.com
Wed Jul 1 12:17:29 CEST 2009


2009/7/1 Mark Davis ⌛ <mark at macchiato.com>

> Well, I agree that "loss of information" can sound drastic, but I was just
> repeating John's words.


Unfortunately, this is exactly what it is.Let understand, Unicode is
slightly entropic because sometimes you can print the same thing with
different Unicode codepoints. This is a small problem. This can be corrected
through tables.

The real problem is to make IDNA itself an entropic process. This means to
make it a non end to end process (end to end process strives to be
information neutral). Mapping at protocol level prevents any metadata to be
exchanged to carry the information that has been lost in the mapping
process. The necssary presentation layer does not exist in the basic
internet protocol pile.

This is why mapping can only be carried at a place where a complementary
process can send the necessary negentropy to restore the initial entry.The
Internet does not supporting it: IDNA has been designed has been assigned
the job. It can then be carried at application layer, but it must be carried
both way, adding entropy first and restoring the correspoding negentropy.

This is why the Charter says that mapping must not occur at protocol level.

As a result it means that your propositions should be totally orthogonal to
IDNA and IETF. They belong to the user application and can be used today.
They may also be incorporated into the usage pseudo-network layer
(stabilized common usage applications). Your best bet to make sure the
Internet supports IDNA correctly is that you can freely change your
specification and everything will continue to run smooth;This is network
neutrality.

Perhaps less scary is just to say that case mapping (whether upper, lower,
> or folding) is not a reversible transformation: "McDonald", once transformed
> to "MCDONALD" or "mcdonald" cannot be restored without additional
> information. In that respect, it is similar to width transformations or
> others under discussion here, except that as I pointed out, it typically
> loses *more* semantic information than the others do.


Far more scarry. McDonald is a registered trade mark. MCDONALD and mconald
are probably not subject to the same constraints. This shows that billions
of $ are at stake.

Elisabeth Blanconil



>
>
> Mark
>
>
> On Tue, Jun 30, 2009 at 18:36, Elisabeth Blanconil <eblanconil at gmail.com>wrote:
>
>> 2009/6/30 "Martin J. Dürst" <duerst at it.aoyama.ac.jp>
>>
>>>  A case mapping is also a 'loss of information', but one that people
>>> clearly want.
>>>
>>
>> Could we not phrase this differently ? Case mapping may be considered only
>> if it does not represent a loss of information. Otherwise should we not name
>> it "case and entropy mapping" to explain users where entropy occurs.
>>
>> Elisabeth Blanconil
>>
>>
>>
>>
>>
>>
>>
>> _______________________________________________
>> Idna-update mailing list
>> Idna-update at alvestrand.no
>> http://www.alvestrand.no/mailman/listinfo/idna-update
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.alvestrand.no/pipermail/idna-update/attachments/20090701/77239332/attachment.htm 


More information about the Idna-update mailing list