Re: More thoughts on multiple style specifications (was Re: SGML to HTML with jade?)

["James Clark" <jjc@xxxxxxxxxx>]

> A rule of thumb is forming in my head,
> "if it's big enough to stand alone---if it's complete without additional
> stuff---make it a separate style specification,

That's a good rule to apply.

> otherwise make it an
> entity".

Another issue is whether you might want to override any of its definitions
or construction rules.  If you have some DSSSL code with a definition, you
can't override that definition within a style-specification that includes
the DSSSL code as an entity, but you can override it within a
style-specification that uses your DSSSL code as a separate style
specification.

Maybe this is pointing towards a need for a new feature in the DSSSL
architecture.  Suppose in addition to a use attribute on a
style-specification, there was a, say, "combine" attribute whose value was
a list of IDREFS like "use", but which differed from "use" in that the
referenced style-specifications would be considered as belonging to the
same part as the referencing style-specifications, so that construction
rules in the referencing style-specification wouldn't override construction
rules in the referenced style-specification.

James


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