[Murray Altheim]
>Aha! The culprit: 'XML processing applications'. I think where the confusion
>lies here is with the lack of differentiation between processor and
application.
>We are defining an XML _processor_, which in all cases preserves whitespace
>and hands it on to the _application_. An application's handling of whitespace
>will be entirely dependent upon the needs of the application. For example,
>'XML as a data format' might normalize or even eliminate all whitespace,
>whereas 'XML as a document markup' may rely on some type of default handling
>under certain circumstances, or rely entirely on stylesheets. A browsers, as a
>specific case of application, will have different WS handling than a database
>engine, and different again than an XML text-based editor.
>
>If the processor faithfully passes all WS to the application, the application
>can generate character-accurate offsets for links, etc. with no problems
>due to WS data loss. I see no problems in the current spec, although I must
>agree with Tim and others that XML-SPACE="DEFAULT" seems to have no discernable
>meaning in this context.
>
>Does that help at all?
>
The "processor" is the XML parser and the "XML applictation" is the
editor, browser, spell checker, indexer etc. Okay. My concern is how these apps
will *interoperate* in the face of application specific WS conventions. To
do the right thing they need to faithfully reproduce the WS. I think
this is a hard problem. I await with interest some code examples
that illustrate XML->XML interoperability.
Am I completely off base in thinking that WS makes for some hairy issues
in XML->XML applications? What apps have been written that read/write XML? How
have they handled WS integrity? Are patterns emerging that can usefully
become part of
XML-DEV lore?