Re: XML Blueberry

[John Cowan <jcowan@reutershealth.com>]

David Brownell wrote:


> I think that's the key point.  It creates "islands of (non)interoperability",
> fragmenting a landscape that really deserves stabilization instead of
> more earth tremors.


Not that much of a tremor, I think.


> As it stands,
> the XML 1.0 spec is effectively independent of changes from the
> Unicode consortium, but still leverages Unicode where it's most
> essential (representation of text, not markup).


Correct.


> That hypothetical situation differs in at least one key respect from
> this real one with IBM.  Macintosh users have always had access
> to ASCII, while it seems this IBM line-end is a legacy from the days
> that IBM fought ASCII because it was too open, and threated to
> decimate their cardpunch/terminal/... margins by facilitating the
> creation of interoperable commodity infrastructure.


ASCII or EBCDIC is not the issue.  The question is, of the 65 available
control characters in 8-bit character sets, which one or ones are
you going to use for the logical "line end" function?  Mainframes
use NEL, other systems use CR or LF or CR/LF.  NEL has been around
for a long time even in the ASCII world: the ISO 2022 7-bit
equivalent of 0x85, which is 0x1B 0x44 (aka ESC D) has been supported
by ANSI X3.64 terminals since VT100 days.

-- 
There is / one art             || John Cowan <jcowan@reutershealth.com>
no more / no less              || http://www.reutershealth.com
to do / all things             || http://www.ccil.org/~cowan
with art- / lessness           \\ -- Piet Hein