Implementation Questions

Just read [1] (enjoyed it too ;-). Some questions:

1) Section [1] > "Namespaces" says "this specification does not define
an information set for documents which use relative URI references in
namespace declarations". Why was this adopted while base URIs can be
relative? (just looking why and how relative [base URI]s can be useful
and *not* why the REC does not handle documents with relative namespace
URIs).

2) Does the REC favor the use of the "Unknown" property value to mark
fragments that may need further processing to produce information items
(or extensions to those) and how should these fragments be
marked/recognized? Would an implementation break the REC if it used
namespace declarations to spot "roots" in the documents and process
those fragments having a [document element] for each one?

3) According to [3] and [4] and the whole inconsistencies issue with
synthetic infosets, why did the WG favored this to be resolved in
"proprietary" way in each implementation? From what I understand it
would be of greater use and simplicity to include a mechanism in the
spec that will resolve the issue (can't this actually be done by
extending the infoset and documenting the relevant information items?)
instead of telling each implementation to describe the inconsistencies
internally. Since the Infoset is abstract by nature and it even includes
something like the [all declarations processed] property, it can expand
to (for example) hold references (just by using properties on the
[document element]) to namespaces that should be in scope but are not
due to a synthesis.


[1] http://www.w3.org/TR/xml-infoset/
[2] http://www.w3.org/TR/xml-infoset/#intro
[3] http://www.w3.org/TR/xml-infoset/#intro.synthetic
[4 http://www.w3.org/TR/xml-infoset/#conformance


Kindest regards,

Manos

Received on Thursday, 10 January 2002 07:08:23 UTC