[
Lists Home |
Date Index |
Thread Index
]
Liam Quin wrote:
> ...
>
>Some possibilities in the future might include, for example, a format
>or framework for interchange at the infoset level, or even a direct
>encoding that can be declared in an XML encoding declaration,
> <?xml version="1.1" encoding="w3c-binary-thingie-1.0"?>
>
>
I already use:
<?xml version="1.1" encoding="x-esxml0.1"?>
at the beginning of my format... This is completely legal, however the
standard does talk about this as a 'character encoding' so it is
stretching it to actually substitute the structure.
On the other hand, it also says that other than the two required
encodings, UTF-8 and UTF-16, a parser should just fail on other encodings.
Certainly this is how a 'parser', i.e. an access library, would
disambiguate input.
I would also note that every element can have it's own encoding. This
meshes perfectly with my plan to support whatever encoding is used for
text in a particular element, individually for each element. The
interesting question is how to reasonably handle two different types of
binary payload. b64 encoded sections are the major case, with support
for other MIME encodings and direct storage of binary scalars also
needed by some applications.
As I stated, the issue of binary payloads is different from the issue of
binary structure. Binary, or rather efficient-in-one-or-more-aspects,
structure is needed first. Binary payloads can be done as application
convention and may only need industry/use-level standardization, just as
they could be done now as b64 or hex payloads. (Loading or storing a
fixed size hex payload can be done in a macro/inline with nothing but
fast shifts, ands, and array reference, a win for ieee floats for instance.)
>(in which case no revision to XML is needed; this is sneaky, rather
>like Tim Bray's character entity proposal a while ago, and the name
>"binary XML encoding" would be appropriate)
>
>There's nothing in the XML Specification that requires specific code
>points to be transmitted for pointy brackets. You can transmit an
>EBCDIC-encoded file if you like. The XML Processor is responsible
>for mapping to Unicode code points.
>
>As Rich Salz points out, there is not complete consensus.
>
>Best,
>
>Liam
>
>
sdw
--
swilliams@hpti.com http://www.hpti.com Personal: sdw@lig.net http://sdw.st
Stephen D. Williams 703-724-0118W 703-995-0407Fax 20147-4622 AIM: sdw
|
