Re: Combining latin small letters with diacritics

Le 5 mars 2012 21:50, Ken Whistler <kenw_at_sybase.com> a écrit :
> On 3/5/2012 12:17 PM, Benjamin M Scarborough wrote:
>>
>> On Mon, Mar 5, 2012 at 19:09, Michael Everson wrote:
>>>
>>> No, because both the combining-a and the combining-diaeresis are bound to
>>> the base letter; the combining diaeresis is not bound to the combining-a.
>>
>> Just like the proposed U+1ABB COMBINING PARENTHESIS ABOVE will be bound to
>> the base letter, not the combining mark that it's parenthesizing. Oops.
>
> That's not an "oops". It is intentional. Michael's wording is misleading,
> because combining
> marks in any case are not "bound" to base letters. The standard goes to some
> length
> to provide terminology for this. See D61a Dependence and D61b Graphical
> application,
> in Chapter 3 of the standard.
>
> Neither of those concepts prevents (or requires) the interpretation of
> sequences of
> combining marks as having strict *semantic* binding in any one particular
> way.
> That is important, not only for cases like the German dialectological
> parentheses,
> which are handled most efficiently (and intuitively) in the way proposed for
> their
> encoding, but also for much more widespread cases where a combining mark
> may be applied a single character, but where the intent of the modification
> applies notionally, semantically, or phonologically to entities larger than
> that single
> character. The most obvious case of this is tone marking in transliteration,
> where
> a vocalic nucleus may be represented by a sequence of letters, but the tone
> mark
> is graphically "applied" to just one of them.

Sometimes I hate the too restrictive character encoding model that
cannot adapt simply to simple situations and concepts, and create
artificial distinctions (disunifications) that were not needed.

And TeX still has a long way to live as its model is much more
productive, and easier to implement and undestand, with far less
exceptions like those above (whose rationales were insufficiently
thought at their origin, and for which you don't want to simply create
an enhanced framework with a compatibility mapping scheme).
Received on Mon Mar 05 2012 - 15:22:53 CST