Re: HTML3 <OL inherit> gone for good?

Brian Behlendorf (brian@wired.com)
Mon, 13 Mar 1995 15:37:11 +0500


On Mon, 13 Mar 1995, lilley wrote:
> "Nothing more than" - I agree with you. Style sheets as well, yes, sure.
>
> It is when people seem to suggest dropping tags and attributes and doing
> style sheets instead, as a mutually exclusive option, that I start to worry.

Browser authors, and document creators, *HAVE* to accept the fact that
unstandard extensions are experiments and are not guaranteed to be in any
final spec. Even <P ALIGN=CENTER>. Without that underpinning there's no
point to the standardization process other than to ratify first-to-market
solutions.

> Mainly I worry about the time for style sheet technology to be pervasive,
> and whether we have that time before alternative, commercial approaches
> become pervasive and the whole standardisation discussion becomes moot.

I'm very worried about that as well - which is why we need to see as much
effort put into style sheets as possible, given that there's a finite
number of resources in the world spent on WWW development. If we split
it between a solution that works in the short term and a solution that
works in the long term, I prefer the latter.

I wouldn't be so passionate about this if there wasn't a working
implementation of style sheets right now - Arena 0.96. If you have an X
platform around, give it a try - colors, alignment, lots of stuff. The style
sheet format is still experimental, yes, but we could reach closure on that
sooner if more people developed test implementations instead of "cool
tricks".

Finally, style sheets are a proven technology in the SGML world, so we
have a wealth of experience to draw from there.

> Sure. Although saying, "10 million people are using this, so maybe there is
> a need for it" seems pretty sensible. Saying, "10 million people do it this way,
> but that leads to a problem, so doing it this slightly different way that works
> now" also seems quite reasonable. [For example, the <center> tag]

Fortunately writing a script which will convert a Netscaped document
to one using style sheets probably won't be too hard to do - once there's
a consensus on the format for style sheets I'd be willing to do this
(though, in the same way converting plain text to HTML can result in
inefficient HTML, converting attribute-based presentation information to
style sheets may result in inefficient style sheets, though the converse
would be really easy).

> Saying, "10 million people do it this way, and they must stop, now, while we
> explore and develop a new technology and maybe have it ready in a year or so"
> seems naive and unworkable to me. I expect it does to you, too.

Agreed there - however I don't feel style sheets are really that
obscure. I would like to compare the amount of C code it takes to
implement a centering attribute on each body tag to implementing it as an
attribute in a style sheet - I have a feeling it'll be really close, if
not less for the latter case. Again, there is a working implementation,
and soon code in the public domain for this.

> If stylesheets are mandatory for all HTML 3 browsers, and this causes a
> substantial lag in time-to-market of a high quality and stable implementation
> on each major platform, we may very well find that HTML 3.0 is treated with
> polite disinterest because there may be, by that time, well established
> alternative methods of achieving the same thing. That would make me very sad.

I think they have been declared as not mandatory, but recommended for
browsers wishing to implement HTML 3.0 professionally. This matches the
orthogonality of style sheets and the markup well - a browser which
doesn't use style sheets can still fully validate the content as HTML,
since the DTD doesn't need to be changed for every new attribute. It
keeps the DTD simple, and allows for a nice extensional system for
presentation issues (which are the most likely to require extensions
anyways).

> > This is all about how future development should move
>
> OK, but perhaps we differ with our definitions of "future"
>
> I take "future" to mean from right here and now, not from August 94. Then,
> existing browsers did not implement centred text, or loweralpha list items.
> Now, they do. We have to take some account of that, and provide a reasonably
> easy migration path to a better way.

No, here's the distinction - browsers which *incorporate* extensions to
HTML which may or may not be part of a standard, can do things like
centered text. ALIGN=CENTER is an example of an easy migration for a
function many people use right now, has been in the HTML+ proposals
forever, and thus should be around as a legacy tag, but technically
server which dish up pages with <P align=center> are lying when they
claim the content to be text/html - it's much closer to text/html;
level=3.0, or better yet text/x-Mozilla-html.

> Incidentally, if some document comes along with pink on lime green headers
> in the style sheet, and you happen to have poor eyesight, is there any way
> to make your needs as a viewer over-ride the style sheet so the text is at
> least legible? Or has than not been settled yet? I recall talk about cascading
> style sheets but I don't recall that discussion reaching any consensus.

Yup, in fact that's one of many reasons why style sheets are much more
powerful than per-tag attributes, the level of control can be at a higher
level. As much as we at Wired would like to force you to see flourescent
pink letters on a flourescent green background... ;)

It seems we agree on all the basic issues - I just want to make a
couple issues clear in the hope that the Right People are listening.

Brian

--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@hotwired.com brian@hyperreal.com http://www.hotwired.com/Staff/brian/