W3C home > Mailing lists > Public > www-style@w3.org > April 1997

Re: indents

From: Jon Bosak <bosak@atlantic-83.Eng.Sun.COM>
Date: Mon, 14 Apr 1997 23:54:17 -0700
Message-Id: <199704150654.XAA05377@boethius.eng.sun.com>
To: www-style@w3.org
[Greg Kostello:]

| Are you suggesting that browser vendors adopt DSSSL (or some
| derivative), as well as implement CSS?

Yes.  As this thread clearly shows, they will end up doing this in
effect anyway.  The only question is whether it will be done well or
badly.

| Certainly there is nothing that precludes any vendor from inplementing
| both style systems, but I thought that war had been waged and CSS
| won. You are welcome to disagree :).

CSS and DSSSL are not competing "style systems."  CSS is a declarative
language for assigning formatting characteristics to particular
elements.  DSSSL is a functional language for constructing formatting
specifications of arbitrary complexity.

DSSSL can be used to make roughly the same kind of formatting
specification that CSS can (though more verbosely), but it can also be
used to create radically different outputs from the same set of inputs
-- to generate an index from a document, or change a listing ordered
by last name into one ordered by first name, or create a bibliography
from all the works cited in a document, to take a few examples at
random.

CSS is much better suited to the compact delivery of simple format
specifications than DSSSL, and it is much easier to learn.  But DSSSL
can readily solve problems that are completely beyond the reach of
CSS.  Think of the alternating paragraph problem again.  Sure, you
could build this specific treatment into CSS -- but you would never
stop needing to add more of them.

| For now anyway, CSS has been adopted by MS and NS, so I can envision
| that both will want to extend CSS for future needs. It would certainly
| seem wise to borrow (steal) from other style systems where it makes
| sense.

It doesn't make sense in this case.  CSS is great for what it does
(and will be very widely used), but it is not an appropriate base for
further development much beyond its current scope.  Extending CSS into
the domain covered by DSSSL is not impossible, but doing it in any
remotely graceful or coherent way certainly is.

| It seems that DSSSL has a much more sophisticated selector system.

The difference is far deeper than that.  DSSSL is a programming
language; CSS is not.  And the DSSSL flow object model is incomparably
richer than the simple box model on which CSS is built.

| My guess is that the inventors of CSS probably looked at the DSSSL and
| designed a implementation that was simpler and hence more doable.

Your guess is incorrect.  The basic architecture of CSS was in place
before its designers examined DSSSL.  (Which is not to imply that they
would have designed CSS differently if they had known about DSSSL; CSS
syntax flows naturally from its design requirements.)

| Now that the bait has been set and the fish have bitten, perhaps its
| time to get mS and NS to go after another prize. After investing in
| CSS, it is unclear to me that there is an advantage to supporting
| another style system as opposed to just extending CSS. Comments?

The question presupposes that DSSSL and CSS are just alternative
syntaxes for doing basically the same kinds of things.  They are not.
DSSSL and CSS are very different tools with very different potential
users.  We need both of them.

Jon
Received on Tuesday, 15 April 1997 02:55:04 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:53:49 GMT