From: Asmus Freytag (email@example.com)
Date: Tue Jul 17 2007 - 15:01:14 CDT
On 7/17/2007 4:50 AM, Kent Karlsson wrote:
> Kenneth Whistler wrote:
>> I think you are missing the point here. The domain of
>> conjunct ligatures in Devanagari is the aksara. You parse for
>> aksara boundaries and don't attempt to map into ligature space
>> across those boundaries. That is rather different from how
>> ligatures work for Latin or for Arabic, for that matter.
> I do hope that no rendering system PREVENT ligation across
> aksara boundaries. There just happens not to be a need for
> ligation across aksara boundaries. But preventing such
> ligatures seems unnecessary.
I'd suspect that allowing full generality comes at a cost, as usual. I
don't think that either developers, or their customers, should be
expected or forced to pay for the support of unnecessary or unused features.
Sometimes a general solution can be faster, more robust and not more
expensive that one with design limitations. That would be the exception
to the general rule, but unless there is such an exceptional case here,
what I wrote is very much at issue here. I suspect that in most
architectures, adding that support would add to the cost.
The majority of your arguments seem more based on your personal take on
what the ideal world should be, than based on what's actually required
to deliver the best support for users of the various writing systems,
and what can (and should) be left out, to make such support affordable
to develop and test.
I'm usually quite sympathetic to the argument that rendering software
(and character sets) should not be designed to support just the lowest
common denominator, but that users benefit from having easy access to
support for lesser-used features, be it for minority languages, or
special notations or typographical conventions. In many cases that
argues for having such support built-in by default (or in the case of
characters, it argues for a universal character set). However, I
definitely part company with your absolutist position in this discussion.
This archive was generated by hypermail 2.1.5 : Tue Jul 17 2007 - 15:03:02 CDT