Thanks for your reply. I provide a few more comments inline (removing
areas where we have agreement).

On 4/25/13 7:50 PM, Takahiro Nemoto wrote:
> 
> On 2013/04/13, at 6:18, Peter Saint-Andre <[email protected]
> <mailto:[email protected]>> wrote:

>> 2. I think it would be helpful to mention some examples of delimiters
>> other than FULL STOP.
> We are supposing delimiter mapping for width compatible characters 
> that like '@' in mail address and ':' or '/' in URI. Even though it may
> be non-existent as for now, 
> but I'm thinking that in the future there may be protocols that use '+',
> '-', '<' or '>' as delimiters.

OK. Do you think it would be helpful to mention examples other than FULL
STOP?

>> 3. This text is a bit confusing:
>>
>>   One of the most useful case of delimiter mapping is when FULL STOP
>>   character (U+002E) is a delimiter as well as domain name.
>>
>> I am not a DNS expert, but my understanding is that FULL STOP is not
>> part of the domain name but is a delimiter between the labels of a
>> domain name. I think the sentence quoted above means that in certain
>> protocols FULL STOP is used as a delimiter within protocol strings (such
>> as identifiers) in addition to being used as a delimiter between domain
>> name labels.
> That's right. There is no FULL STOP in DNS packet.
> But when comparing domain name strings outside of DNS protocol, 
> FULL STOPs are used as delimiters and width problem should be considered.
> 
> For example, in web browser, FULL STOPs are considered when 
> inputting words in comparing them with the page history list.

Agreed. I think it would be good to mention that example in the text.

>> 5. I think Appendix C could be moved into Section 4.3.
> Because we thought that it's not proper to move a table 
> that has a unicode version dependency in the section, 
> we place in the Appendix.

That seems reasonable.

>> 6. The big open issue here is the order of mappings. RFC 5895 defines an
>> order of case mapping, width mapping, NFC, and the IDNA protocol. I'm
>> not sure why we would do width mapping before case mapping in PRECIS,
>> because we want to do local case mapping before non-locale-specific case
>> mapping. Therefore I suggest the following order:
>>
>>   1. Width mapping
>>   2. Additional mappings (draft-ietf-precis-mappings)
>>      a. delimiter mapping
>>      b. special mapping
>>      c. local case mapping
>>   3. Case mapping (i.e., non-local-specific)
>>   4. Normalization
>>   5. PRECIS protocol
>>
>> Does anyone see a reason to move width mapping after case mapping (as in
>> RFC 5895)?
> We agree about this order.

Great! I'll update the framework specification accordingly.

>> For the record, I think we need to specify the order of operations in
>> both draft-ietf-precis-mappings and draft-ietf-precis-framework so that
>> there is no ambiguity. Once we have consensus, I will be happy to update
>> the framework accordingly.
> We will, too :-)

Excellent. That will enable us to close one of the only open issues with
the framework. :-)

Peter


_______________________________________________
precis mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/precis

Reply via email to