Re: [precis] local case mapping

[Peter Saint-Andre <[e-mail]>]

On 10/18/13 1:41 AM, Takahiro Nemoto wrote:
> I would like to address some of the comments that I have received
> from Peter, Alexey, and Andrew in this email.
> 
> First, I would like to do some revision to the document according to
> Peter's suggestions. Also, I am planning to split the "References"
> section into "Normative References" and "Informative References". 

That's all good.

> One
> other thing, as Andrew has pointed out, I am thinking of local case
> mapping as followings:
> 
> Purpose: The purpose of local case mapping is to increase the
> probability of matching-result from the comparison between uppercase
> and lowercase characters,  targeting language-dependent characters.

Works for me.

The text in the I-D is now:

   The purpose of local case mapping is to increase the probability of
   matching-result from the comparison between uppercase and lowercase
   characters, targeting locale and locale and context-dependent
   characters.

I stumbled over the last clause. I think it would be a bit clearer as
follows:

   targeting characters whose mapping depends on locale or on locale
   and context.

> Regarding the final sigma,
> 
> Currently, the handling order of PRECIS Framework is local case
> mapping to case mapping. Because of this, when Σ=U+03A3 is handled by
> local case mapping that apply the context-sensitive (i.e.,
> "language-insensitive") mappings from SpecialCasing.txt, it becomes
> ς=03C2, in the end, when ς=03C2 is case mapped, it will still result
> in σ=03C3.
> 
> For this reason, final sigma is not one of the targets for local case
> mapping.

OK, I understand that logic. However, it implies that if a protocol or
PRECIS profile uses local case mapping then it will benefit from the
nice language-sensitive mappings in SpecialCasing.txt but it can't
handle Greek final sigma in a context-sensitive way.

I *think* that's probably fine for comparison purposes. To use my
previous example, a nickname of "ΦΙΛΟΣ ΜΟΙ" would be case folded
to "φιλοσ μοι" (not "φιλος μοι"). That comparison would be applied
consistently, without attention to context (i.e., we wouldn't get
"φιλοσμοι" if there's no space but "φιλος μοι" if there is a space).

This seems to be an acceptable approach, I just wanted us to be clear
that this is what we're doing. :-)

> Also, in the next I-D, I would like to add "local case mapping can be
> selected only when case mapping is selected using the PRECIS
> Framework profile" and "casefolding in this document means full
> casefolding described in the Casefolding.txt file".

That's sensible.

Peter

--
Peter Saint-Andre
https://stpeter.im/