I18N-ISSUE-150: (Encoding) User agents must not support any other encodings or labels.

The text now says:

"The table below lists all encodings and their labels user
agents must support. User agents must not support any other
encodings or labels."

> Raised by: Richard Ishida
> On product: Encoding-prep
>...
> We should discuss the ramifications of the second sentence.

I understand what I think is the intent here and, as intent,
agree.  The more of the world that settles on Unicode encoded in
UTF-8, the better off we will all be.

But a strong normative statement like the one in the second
sentence seems problematic to me.  Consider two cases (there may
be others):

(1) Consider that web pages may exhibit or discuss historical
scholarly topics, not just contemporary materials and
conversations and that page images are generally less desirable
than encoded text when the latter is feasible.  No suppose there
is a script that is not yet encoded into Unicode but for which a
scholarly community has devised an encoding.  Almost certainly
we would not want to ban that community and any browser they had
modified to work with their encoding from the web and the
network, at the above statement appears to do.  We could try to
convince them to map their encoding into Unicode private-use
space but, not only would that pose all of the problems of
private-use space in a global environment (probably including
having to invent encoding indicators), but, if they had any
expectation at all that their script would eventually be
incorporated into Unicode in a reasonable way, expecting them to
convert twice (from their encoding to Unicode private-use space
and then from those private-use codes to final code points) is
profoundly unrealistic.

(2) The web is often used to support interfaces to other
protocols, with email probably being the most common.  Some of
those protocols allow non-UTF-8 encodings; some uses of them
even require others for interoperability with particular
(non-web) systems.  Banning encoding-sensitive interfaces to
those protocols and systems could be extremely disruptive,
damaging to interoperability, or taken as an assertion by WAHTWG
and W3C that such interfaces and configurations just don't
belong on the web -- an assertion that would be ignored except
for the credibility problems it would cause for both bodies.

The second sentence should be converted into what the IETF would
call a "strong SHOULD", e.g,, replacing it with something like: 

	"Unless it is required to support very unusual
	situations such as scripts for which no acceptable
	Unicode encoding exists or applications that impose
	their own requirements on web interfaces, applications
	should not support any other encodings or labels." 
	

thanks,
    john



because, if conformance is implausible

Received on Thursday, 17 July 2014 16:07:00 UTC