W3C home > Mailing lists > Public > www-style@w3.org > July 2003

Re: Fwd: [CSS21] body { padding: 8px; } HTML40 sample stylesheet

From: Jukka K. Korpela <jkorpela@cs.tut.fi>
Date: Fri, 25 Jul 2003 21:49:30 +0300 (EEST)
To: www-style@w3.org
Message-ID: <Pine.GSO.4.50.0307252107070.13689-100000@korppi.cs.tut.fi>

On Fri, 25 Jul 2003, Tom Broxton wrote:

> Given that the descussion was regarding the default style sheet of for
> screen based media,  I think it is relevant and important that UA
> authors are encouraged to display web pages in such a way that they
> have a generally consistent user-interface.

I think we are discussing sample style sheets in general. And although
aural properties are practically irrelevant at present, the existing
sample style sheets are not limited to screen media. There's nothing in
them that says that they would not apply to print media too, for example;
in fact they have some specific rules for print media only.

Do you really mean user interface, or do you actually mean document
rendering? They are quite different things. And would the consistency,
whatever that means in detail, be important to users or to authors?
For which reasons?

> We should not wish for
> a world in which the default rendering for a page was an uninterrupted
> stream of text with no distinction for headings, paragraphs or
> hyperlinks.

But does this mean that a specification needs to contain an appendix of
obscure status, saying that <ul> elements have 40 pixels left indentation?

> It may well be that allot of the conventions that we have settled are
> merely based on the way Mosaic did it, I think the web would be allot
> less popular if each subsequent UA author had gone their own way on how
> to display a page of information.

Would it? Would it really have spoiled the Web, or would it spoil the Web,
if browsers by default applied some decent typographic rules? Did the Web
crash when the default background color was changed from gray to white?
(Actually, the sample style sheet in CSS1 has body {background: white}.
I don't think this contributed much to the change, but if it is important
to have consistent default rendering, why hasn't CSS2 sample style sheet
have anything about background color?)

> It makes sense that headings are more prominent than body text and that
> there is some consistency in the way that hyperlinks are displayed on
> they web.

Surely. But that relates to the general principles, not specific styling,
and it should be said in HTML specifications. Surely the principles should
apply when no CSS is used - they should apply even stronger, when no style
sheets are in use, and a browser just renders a document by fixed rules.

Besides, the sample style sheets violate the specific principles you
mention. In the CSS1 sample h5 and h6 were set to font sizes smaller than
copy texts, and CSS2 is even worse for h6, font-size: .67em. (Bolding
_may_ affect the rendering. But 67% of basic font size still does not make
text more prominent than normal text.

Regarding links, colors are probably the most important presentational
feature in most browsers, and underlining comes next. The CSS2 sample
style sheet says nothing about link colors. This is perhaps better than in
CSS1, which sets visited links red and active links lime - luckily browser
vendors ignored this recommendation.

> The sample stylesheets are a way of describing this status quo.

They don't actually describe any status quo (except in some details), and
there's no reason to recommend preserving the status quo. Regarding
descriptions of browser behavior, they would be useful, but such things
hardly belong to specifications even as appendices, especially since the
target is moving fast. Besides, browser style sheets should be published
(and preferably documented by explaining the decisions) by browser
vendors, instead of being reverse engineered.

I recently compared, in detail, the sample style sheets in the CSS1 and
CSS2 specifications and the CSS 2.1 draft. And I can say that they are
inconsistent with each other, and they neither describe actual browser
behavior nor serve as good examples - as a whole. They are still useful
reading _to a very critical mind_ (or if explained critically), since
they partly make explicit some browser features, partly show what oddities
browser default style sheets _might_ contain. But they should become
history now. If there are some rendering features that should be
recommended to be implemented in browsers by default, they should be
given in HTML specifications, not CSS.

-- 
Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/
Received on Friday, 25 July 2003 14:49:32 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:22 GMT