Clark Evans reopened the eternal namespaces
discussion by asking why the W3C abandoned the processing instruction approach in favor of
the attributes
approach. The answers suggest mostly politics.
Andrew Layman's claim that processing
instructions don't support scoping was refuted by David Brownell, though
Frank Boumphrey suggested that
opinions on scoping and "a feeling that PIs were 'broken'"
were responsible.
David Megginson, in an email that missed the archives
(but which is partially quoted in
this post
by the author) suggested that 'influential parties' who
wanted to keep PIs from showing up in older browsers were
to blame, though he explicitly excused the usual
suspects.
David Brownell suggested that
'a certain person (or persons) disliked PIs extremely', and
that this was the true cause.
Arjun Ray, while noting and disputing the scoping
issues, pointed out that ordinary readers
have no access to the debates where the namespace spec was
framed, and noted a line in those archives -
"The
making of laws, and of sausages, should be hidden from
children".
Rick Jelliffe joined the discussion with an
exploration of the scoping issues, and concern that
namespaces have added yet another level of scoping to XML,
as well as creating "attributes that are not attributes" on
top of XML declarations being "PIs that are deemed not
PIs".
The namespaces debate goes on. Despite Tim Bray's claim that "the namespace-oblivious
world is just no longer interesting", many developers still
seem to be figuring out - and questioning - the impact of
the Namespaces
in XML recommendation, almost a year after its
publication.