Associating DSSSL style sheets with documents

Eve L. Maler elm at
Mon Mar 17 19:52:16 GMT 1997

At 03:40 PM 3/15/97 +0700, James Clark wrote:
>At 13:10 14/03/97 -0500, Eve L. Maler wrote:
>>This is interesting: Should an XML effort determine a PI that should be
>>usable in general by SGML documents? 
>I wasn't proposing that *XML* define such a PI.  All I was just suggesting
>was that people who have DSSSL engines implement it (preferably making the
>name of the PI configurable).

Oh, I see.

>>I tend to think that the "authority"
>>that invents/maintains the format of the PI should be identified, and "XML"
>>sort of fits the bill, similarly to <?SO.  We did say that the first name
>>token in a PI functions as a sort of notation.  It would be weird for an
>>XML spec to specify <?stylesheet .
>>I've also been beating the drum on the WG list about how our PIs should
>>have "GIs" as well as "attribute specs," so I'd prefer to see <?XML
>>stylesheet att1="val1" att2="val2"... ?>.  This way, "<?XML" targets the PI
>>so that it will be processed by an XML-aware processor, and the rest
>>identifies the semantics of the instruction.
>I disagree.  XML requires that all PIs start with a name, and says that this
>name is normally the name of a declared notation.  So I think PIs should
>look like
><?name att1="val1" att2="val2" ...?>

I'm not sure how your second sentence follows.  Why not have XML as the
notation (that is, XML-handling processors should operate on this PI) and
still have a "GI" that indicates the subclass of XML PI?  (But see below

>(Note that the currently-defined XML PI fits this pattern not the one you
>suggest.)  The authority should come from the public identifier on the
>notation declaration for name.  Since XML reserves all names beginning with
>XML-, I would think that an XML-defined PI should look like:
><?XML-stylesheet type="text/dsssl" href="foo.dsl"?>

This is a good point.  In that case, then the XML PI at the top should
start with "<?XML-prolog" or something.  I don't care which way it goes,
but it should be consistent to the extent possible.  Also, I'd like to be
able to describe the PI "template" and any intended constraints on the
values by showing a mock ELEMENT/ATTLIST declaration, and if it's always
<?XML att="value" att="value">, then you can't easily distinguish among the
PIs by type.


xml-dev: A list for W3C XML Developers
Archived as:
To unsubscribe, send to majordomo at the following message;
unsubscribe xml-dev
List coordinator, Henry Rzepa (rzepa at

More information about the Xml-dev mailing list