Perhaps it is best for everyone to understand what side
I am on. I side with Cowan but completely understand
Elliotte's position and support that too. We need to accommodate
the variants. We need to reckon the costs. That is as clear as I can state it.
In some cases and situations one finds oneself engineering for, a core
product that is customizable is the solution. Cases where local
diversity is real, is not amenable to homogenization, customization
is to be preferred, and in fact, is a source of business. That the
ancient mediterranean community agreed on demotic enabled
trade. Yet they did not give up local custom or language; they
adapted to a common language. XML is that. A common
language was created and applied by subsetting an international
standard for creating languages. The international standard provides a
declaration file as the ultimate means of customization. Therefore,
in cases where there are local requirements in conflict with the
global language declaration, the option of enabling the local system to create
its own variant for its own purpose is preserved. Costs must be
reckoned for that and it is wise to reckon these in advance. But
the cost of permanently removing the option to further a cause
of global interoperability under the aegis of international companies
is to my thinking, horrible.
As Tim Bray said, one looks at the options and has some
discomfort with each. I have no problem with
making a choice. I choose diversity because the alternative
is to tell groups "they cannot play". The IBM NEL doesn't
bother me much. The Japanese language issues bother
me a lot.
It is a world wide web, not a bottle of homogenized milk.
If the W3C for business reasons decides not to support
this, being a business consortium, that is their decision
to make. However, I do think those affected adversely
should remember they have an escape hatch provided
by ISO whose customers being the governments of the
world have considered local needs wisely. Again, there
is a cost to reckoned in all such deicsions, and the greatest
cost may be interoperability as you note.
So, optimally, the task is to preserve interoperability and enable
diversity, not one or the other.