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

Re: Color: auto, or colour fallbacks

From: David Woolley <david@djwhome.demon.co.uk>
Date: Sat, 5 Jul 2003 20:48:16 +0100 (BST)
Message-Id: <200307051948.h65JmGs11349@djwhome.demon.co.uk>
To: www-style@w3.org

> than to force all layouts being 'raped'. So I don't agree, and I guess all
> developers work is obsolete when every user-agent overrides site layout.

That is one of the ground rules of HTML+CSS that authors seem not to
understand.  The user does have ultimate control, by choosing a browser
that fits them and by using !important rules (or the equivalent as
IE Tools | Options | Accessibility | { Ignore author fonts, Ignore author
colours}.  It was actually one of the policy changes between CSS1 and
CSS2 that, whilst a user normal rule is overridden by an author rule, 
a user !important rule has ultimate authority.

> Second, if you begin allowing not only maximum contrast (I guess you mean
> black text on white background) but even alternative and still readible

"maximum" would be better stated as "adequate".  Forcing black on white with
blue :link, purple :visited, etc. is already a standard feature of browsers,
and what users are forced to use if the start running into sites with
low contrasts (or as in my real example, defaults that assume scripting
will make the text visible), or which think that visited links should be
indistinguishable from ordinary links, when the user would really like to
know what they visited, either because they want to make sure they 
cover everything, or because they want to return to the same selection
from a forum.  (I would also say that not making links obvious causes real
problems for those not used to using computers an a regular basis.)

What I was proposing was the option of allowing some of the author's styling
to get through, as long as it didn't conflict with the user's policy.  E.g. 
if the author specifies black text on white background with black :link,
:visited, etc., the user may be prepared to accept the foreground and
background but insist that the browser shows the links, and in particular,
highlights the visited links. 

I'm actually making a suggestion that works in favour of authors, by
not being the current all or nothing solution.  At the moment, if I read
one of your pages in the office, none of your colour scheme will get
through.

To allow a user style sheet to cope with poor contrasts as well as attempts
to disable features (without the poor contrast problem, a user might only
specify auto for :link and :visited, but with poor contrasts), one may need
to refine the specification by saying that user agents SHOULD respect
author colours if they are compatible with the user's policy.  This I think
needs a new property, so that it can inherit in parallel with the auhor's
colours.

> alternatives, you'll IMO get a definition problem -- what is 'enough' contrast? #000
> on #EEE? Or #036 on #9CF, is this okay, since user-agents have to calculate
> acceptable values for maybe 'color: auto'?

I'd rather allow the user agent to provide its own way of setting details
of the policy, although some consistency would be desirable to cope with
shared machines, and internet cafe machines.  What I'm trying to do is
get the concept into the model so that browser developers are aware of
it and competent professional page designers are aware as well.  I think it
would actually be bad if the behaviour was so well defined that authors
could predict it in detail and design for the result; it would also not allow
for the issue that there are multiple kinds of colour blindness, so different
users actually need different detail policies (I'm not colour blink but still
find that the use of colours on many sites detracts from their readahility).
Received on Saturday, 5 July 2003 15:52:30 GMT

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