Re: thoughts on non-compliant browsers

> Formalizing the issues in CSS would give the browser

Formalising issues is OK, but CSS doesn't attempt to provide pixel
perfect rendition, so must deliberately leave a lot of options
free to the implementor, to encourage them to find solutions for
different sorts of user and different technologies.

> - (And even) should leading increase proportionately with
>   text scaling? If my memory serves me well, paper page
>   designers tend to use a smaller percentage leading for
>   larger font sizes.

The CSS model of leading is indirect, in the form of line-height,
and this is a problem, at least with the IE 5.5 implementation
of font-size related accessibility features, although the problem
is actually partly the result of bad design practice.

As well as specifying microscopic absolute font sizes, designers have
taken to specifying corresponding absolute line heights.  If you turn
off the font sizes in IE 5.5 Windows, reasonable size fonts end up
with leadings which are too small for legibility or even negative.
Designers should, of course, specify line heights in ems.

> - The browser manufacturers, who have to do the best they can
>   to accommodate the (sometimes unpredictable) decisions
>   of the designers and users, within the limits of the
>   fonts available, and without introducing so many options

The business model is important.  Some important browser developers have
business models that are based on the sale of authoring tools and back
end database systems, and therefore react to designer requirements more
strongly than to reader requirements.  Even the transition from
Mosaic to Netscape withdrew user control to be more attractive to
designers.

>   An easy way to turn off scaling andreturn to the
>   "designer" values is essential.

This is only necessary because it can be the only way of making a 
badly designed page readable, but ease of turning it on is also
necessary for similar reasons.  I certainly think that, for GUI
browsers, it should be a tool bar item.

> 
> - CSS adds a property for which I can't think of a better
>   name than "scale-basis" and which has at least two values
>   that mean:
>   .  "Scale with text"
>   .  "Scale with page"

In most cases, scale with page as an author hint would indicate that
they are using the wrong medium and should  be using a page description
language, like PDF or SVG.

> - "scale-basis: text" if "background-repeat: no-repeat"
> - "scale-basis: page" otherwise

scale-basis: page should never be the default if the resulting page
would require horizontal scrolling.

Received on Thursday, 11 March 2004 02:31:16 UTC