On Fri, 17 Jan 2003, Joseph A Nagy Jr Enterprises wrote:
>> Yes, a lot of the images in the CSS2.1 spec are illustrative only. The
>> examples starting in section 9.8 are especially good examples of this --
>> those are completely not to scale.
>> The reason we (the working group) don't like to make pixel perfect
>> examples is that when we do, we always find that we end up doing them 1
>> pixel wrong, or a slightly wrong colour, etc, and then implementors come
>> back and say "the spec is all wrong! the spec is all wrong! we tried to do
>> like the diagram but it disagrees with the spec!" etc.
>> Making the pictures completely wrong gets around this. :-)
> I wouldn't be proud of making the pictures completely wrong. What use is
> an illustrative picture if the illustrated effect only works a certain
> way or in certain browsers?
The illustrated effect should work in exactly the same way in all
browsers, if they all follow the spec. But none will look exactly like the
picture in the spec.
For example, take the example in section 9.8.1 of CSS2:
All it wants to illustrate is how two inlines with different colours, one
after the other, are simply on the same line, so that in the next section,
it could show how relative positioning would move the text up on the line,
and in the section after that, how a float could move the text aside.
Creating a reference rendering for those cases is HARD. You can get close
to it, but if you get too close, as I mentioned in my last message, you
get people who start to consider the image normative rather than referring
to the text.
In addition, fixing errors in images like this is a real pain. If the
working group decides that the line box model should be based on the font
box height instead of only the line-height, we have to redo every single
picture that involves text. That is probably more work than would be
required to go out and fix every single UA to do the new thing!
Ian Hickson )\._.,--....,'``. fL
"meow" /, _.. \ _\ ;`._ ,.