Re: Are Named sequences always going to be graphemes?

From: Asmus Freytag <asmusf_at_ix.netcom.com>
Date: Thu, 21 Jun 2012 20:05:54 -0700

On 6/21/2012 7:51 PM, Karl Williamson wrote:
> On 06/21/2012 01:45 PM, Asmus Freytag wrote:
>>
>>>>
>>>
>>> OK. Will they always be in NFC?
>>>
>>>
>>>
>>>
>>
>> To apply Ken's dictume to this case:
>>
>> That seems like a straitjacket looking for an unwilling wearer. ;-)
>>
>>
>> Unless it's excluded from the start, anytime you limit it, when the time
>> comes you need something like that, you have to invent a new
>> property/mechanism.
>>
>> These restrictions seem not necessary right now....
>>
>> A./
>>
>>
>
>
> Let me make sure I was understood. I meant, can someone count on the
> file NamedSequences.txt having all its data in NFC, as it currently
> is, or is this yet another potential inconsistency that an implementer
> has to contend with.
>

Here is the state rationale for having named sequences:

    "In some limited circumstances it is necessary to also provide a
    name for such sequences. The primary example is the need to have an
    identifier for a sequence to correlate with an identifier in another
    standard, for which a cross-mapping to Unicode is desired."

If there's ever a standard that's not in NFC (and some are not) then
being able to name a specific non-NFC sequence may indeed be useful.

Sometimes, trying to make the life of the implementer easier cannot be
the goal.

A./

PS: as it's well-defined how to create an NFC equivalent to any
character sequence, you could still use the names in an orderly fashion
to match normalized text.
Received on Thu Jun 21 2012 - 22:09:03 CDT

This archive was generated by hypermail 2.2.0 : Thu Jun 21 2012 - 22:09:03 CDT