Re: Color: auto, or colour fallbacks

> It would say "If the author's color is the same as the background or
**** visited color, use some other color (specified where?).

As I argued in the orignal article, I don't think that the algorithm
needs to be part of the specification; this is a tool for users not
authors, even though it benefits authors by not having their colour
styles completely cut off.  Different users will have different needs
meaning that this is an area that should be left open to innovation
by user agent developers.

> This doesn't quite sound like auto so much as "contrast." Further the

"auto" was the first thought, but I've no real objection to another

**** problem easily becomes complex. For example, what if the background-color
**** is barely blue and the link color blue? Is that in violation of contrast

As I see it that should be an issue between the user agent and the user.
It might be pre-set by the user agent or the user might have a control that
sets the colour space (to allow for colour vision defects) and level of
contrast, considered sufficient.

One could add attributes and make the algorithm explicit, but I don't see
a benefit.  I do see a benefit in making the existence of the feature 
explicit as it will be a trigger to that innovation, and will make really
professional designers aware that users are likely to selectively override
their colours.

**** rules here? Most users would find the links unreadable, yet, some may
**** prefer this color scheme (at the very least the author).
**** below 70%. User Agents are encouraged to offer a "force contrast"

The contrast is in a multi-dimensional space which is not the same
for all people.

**** option that, when rendering text at less-than 70% contrast, brightens

This doesn't allow the user to opt into deliberately equal colours.
It also assumes that one is solving for a combination of only two colours,
whereas one is often solving for at least four.

**** and darkens the text and background as needed. User Agents could offer
**** this as a checkbox, as a minimum contrast slider bar, etc."

In reality, for most users, a "user style sheet" means a combination of
checkboxes, etc.    !important rules in user style sheets are all really
like this; they could all be said to be outside the CSS world.

Could you please note that, whilst most GUI mail programs, except Mozilla
in compliant mode, misrender them, long lines in MIME email are long lines
and should be displayed horizontally scrolled, not word wrapped.  I've
arbitrarily broken yours in the above.  The only core MIME format that
allows word wrapping is the rich text format, that was never well supported
by user agents.

Received on Tuesday, 8 July 2003 02:51:04 UTC