Re: word processor document format: what parts?

[Jean-Baptiste <jbnivoit ix netcom com>]

Mauro Condarelli wrote:
> 
> [snip (again, this start to be a long thread ;-)]
>
> > > IMHO this shouldn't be too annoying and the user could learn fast to re-use
> > > previously defined styles.
> > This would force us to impose some structure on their document, and
> > *that* could
> > be annoying (people don't like to structure, it forces them to think).
> Sorry, I didn't express myself well enough.
> I do not think we skould try to impose our styles to the users.
> I was referring to the styles *he* defined.
OK, let them sort out the mess they created by themselves :-)
 
> The MS-Word idea of having local and global styles is ok, but I think styles
> should be global by default, so the user will begin with an almost empty
> style sheet, but would incrementally enlarge his styele database
I am not sure this would happen for joe user, though. As was said
before, most
end-users don't define their own stylesheets/structures, that why
templates exist.

> There should be a way to collapse (under user control) several similarly-looking
> styles in a single one. This feature should reside in the ".dot" file, allowing
> for semi-automatic style-collapsing of documents after creation.
That might be hard to do, i think.

> For documents transmitted to other users we should include a copy of the
> ".dot" file in the document itself.
I would say send a copy along, not embedded.
 
> This is going to give rise to interesting problems when the document is
> transmitted to an user who has a ".dot" file conflicting with the copy
> embedded in the document.
That's for sure.
 
> IMHO the key to sort this can of worms is to find a good "distance function"
> between styles, and use it intelligently.
As you pointed out, such a 'comparison' function (or 'distance'
function) would
only rely on the actual structure & styles defined by the template (i.e.
the DTD
plus the XSL stylesheet), not on names (not even names of elements
inside the DTD).
This raises also the interesting question that the comparison function
should only
operate on structural definitions, not on the style! (this is why
structure and
style are separated in SGML/DSSSL and XML/XSL from the start!)
but there could(should?) be another comparison function that works in
parallel on style 
definitions.
Now how would that 'distance function' work? should it first try to
match equivalent DTD? (and what's even more difficult is that in some
instances of 2 DTDs, they could be equivalent, but in other they might
not be,
if for instance some elements are not used) and then try to match the
associated stylesheets?


			jb.