RE: XML Processing instruction proposal

[Peter Nilsson <pnidv96@xxxxxxxxxxxxx>]

On Mon, 15 Feb 1999, Didier PH Martin wrote:

> <YourComment>
> > Here is my proposal submitted to you for discussion. this is concerning
> XML
> > rendered with DSSSL style sheets and it is expressed as a set of rules.
> >
> [...]
> > b) The browser make this implicit assumption:
> > 	- if not media property is mentioned it is by default the screen and the
> > native rendering formatting objects are HTML+CSS based.
> > Example: <?xml-stylesheet href="MyScript.dsl" type="text/dsssl"?>
> > 	- If media="screen, html" then the browser is using the SGML
> transformation
> > part of the DSSSL engine and display the resultant SGML markups. This is
> > corresponding to the Jade's flag "-t sgml".
> > Example:<?xml-stylesheet href="MyScript.dsl" text="text/dsssl"
> > media="screen,html"?>
> 
> I think the default medium should be "screen" and then the DSSL Online
> flow objects (scroll, link etc.) should be assumed. The SGML
> transformation you are referring to is non-standard Jade extensions
> (indeed they are useful) and using them by default is therefore not a good
> idea.
> </yourComment>
> 
> <Reply>
> So, on the first point you agree with the proposal (i.e that the default is
> set to screen)

Yes. This is a good idea for online browsing.

> The second point. You are talking about Jade and I don't intend to modify
> Jade as it is used today (as a console exe). My proposal concern DSSSL and
> document browsers (so either a modified version of Jade or a new CSS, XSL or
> DSSSL implementation). This proposal implies both IE and Mozilla but could
> also affect any other document browser for XML documents.

But when you say that the medium "screen, html" is equivalent Jade's
option "-t sgml", you imply that the nonstandard flow objects for SGML
transformation (document-type, element, empty-element etc.) that Jade
implements should be the default set of flow objects (right?). And what I
tried to say is that the default medium should be "screen" (a graphical
color display), but the default flow objects should be them specified in
DSSSLO (a subset of the DSSSL flow objects). The reason is that they are
standardized and provide a simple way to specify the formatting of a
document without going through HTML.

[...]
> render SGML documents). These extensions make sense only in this context and
> are based on the work on some W3C workgroups. XML specs are driven by W3C,
> so I tried to uniformize the style sheet PI with the latest W3C
> recommendations/proposals.

And I have to say that uniformizing with W3C specs is of course a very
good idea.

> > - if media="screen, rtf" then the browser display the result with a RTF
> > player associated to the rtf extension. For example, this could be word or
> > word viewer.
> > Example:<?xml-stylesheet href="MyScript.dsl" text="text/dsssl"
> > media="screen,rtf"?>
> >
> 
> <YourComment>
> I don't see why one should specify that RTF should be used for formatting
> the document. In my opinion, the process of formatting is implementation
> dependent and will vary between different browsers. What I am trying to
> say is that if one has a native DSSSL browser, it would be very
> inefficient to do the formatting via RTF.
> </YourComment>
> 
> <Reply>
> Of course, if the browser can only render in a single format, this is
> overkill. If the browser is more versatile and allow rendition in different
> formats like for instance RTF or PDF, the last provision works for more
> versatile browsers. To allows different format also have the advantage to
> adapt to the user's need if a format like RTF or PDF (or any other format
> supported by the browser)is needed for any good reasons the user have. A
[...]

But as I understand it, the media pseudo-attribute is used to choose
between different style sheets for different media. But the output format
is irrelevant in choosing the style sheet. For instance, you wouldn't have
to make one style sheet for generating RTF, another for PDF and a third
for TeX. The flow objects are the same (paragraph, table, line-field
etc.). Then the output format shouldn't be specified as a "medium", with
one exception coming up to mind: SGML transformation, because it uses
another set of flow objects or even the DSSSL transformation language.


Regards,
/Peter Nilsson

--
'(?P . (?e . (?t . (?e . (?r)))))


 DSSSList info and archive:  http://www.mulberrytech.com/dsssl/dssslist