More usefully, the editor could present the DTDs to the user as a serie=
s of
stacked palettes, each containing a list of the elements defined within=
each DTD. When the user selects an element from a palette for inclusio=
n in
the target document, the editor could present the tags and attributes
associated with the element, and hence guide the user in constructing a=
syntactically correct document that conforms to the DTDs.
Let's up the ante a little. A palette of entities would be more useful=
if:
* Each entity were represented by an icon that suggests its function
* Each entity popped up a one-liner outlining its function whenever the=
mouse hovered over it for a while
* Each entity was backed by complete help documentation (in XML, of cou=
rse)
To do this stuff well, the editor would need access to more than just t=
he
plain unvarnished DTD. It would need extra meta data to be associated
with the DTD, but only used at document composition time. Browsers and=
other rendering programs would not need to access this extra informatio=
n
when they render the final document, and indeed it would slow them down=
unnecessarily to do so. It may make sense to exploit XML's (proposed)
powerful hyperlink facilities to associate compose-time meta data with
DTDs. All of the design-time meta data required to help the user under=
stand
and exploit the DTD could be made available in this way.
If this meta data is made available through hyperlinks then it may be a=
good
idea to establish a convention now, while it's still early enough, as t=
o how
such compose time meta data will be classified, and to encourage the
builders of browsers and other rendering engines to omit these designat=
ed
hyperlinks from the popup menus they present to their users should the =
user
click on the associated hot-spot; or at least to make this omission the=
default action, overrideable in the browser's option settings.
It seems likely to me that a number of different software developers wi=
ll
build XML editors that make use of associated compose-time meta data
such as I have described above, and that each will choose to format thi=
s
information in a different way, and that DTD builders will be faced wit=
h
the dilemma of which meta data format they should use, and that the val=
ue
of all DTDs will be diminished by the fact that different XML editors w=
ill
work best with different formats of meta data.
Can we try to pre-empt this problem before it hits us by debating and
proposing a standard format for compose-time meta data?
Trevor Turton
=