Re: Float overflowing behavior!

On Mon, 19 Aug 2002 13:56:00 -0400, David wrote:

>On Monday 2002-08-19 17:01 +0400, Vadim Plessky wrote:
[...]
>>...I doubt typical web master will take his (her) time to read
>>*all* W3C specs...
>> On th eother hand, he (she) will understand what width: 100% means.

Most of us have been there at one time or another.
Reading specs takes time, thinking, conclusions to be made, and more
often than not one is bound to make an error in ones own interpretation.

>If CSS worked the other way around, we'd doubtless have complaints from
>irate web authors on www-style like:
>  I specified a width for these images, and it always worked correctly,
>  scaling them exactly by a factor of two.  But then I decided to put a
>  border around the images, and for some stupid reason all the browsers
>  decided to shrink the images.  CSS is stupid.  When I specify a width
>  I mean the width of the content, not the width of the content and the
>  padding and the border.

Sure; but as you know, exactly this part of the css spec is not written
out in big enough letters. I still have the mail saved from when you
wrote a simple explanation about 'width' to me at the time when we where
both on the "samurai" list. That was in 1999, and it was like being
given a BIG clue from simple words.

The 'float' experiments that has been mentioned in other posts here was
designed by me directly after I got sent into the right way on the CSS
box rendering model by you.

(the example "correct rendering" snapshots was made with Opera 3.62 :)

>In other words, there are tradeoffs to be made, and there are arguments
>on both sides of whether 'width' should include the padding and border.

It shall not. The spec is clear, implementations are also mostly clear
(Win-MSIE being the problem child as usual :)

What I miss is the possibility to suggest 'border-width' in percentages.
That would have made it possible to get rid of the extra level of DIV
markup that is needed now to make the 'math' come out precise.

I had the opportunity to ask Haakon directly about this "omission" when
Sue and I was at WWW9 in Amsterdam; his reply was that he and Bert had
been thinking about it as a possibility but decided to leave '%' out for
borders since it would have ended up as so small numbers in most cases.

I'm not sure that this is an argument that holds up today, we have more
strong harware and given the fact that lot's of www instance producers
seems to demand their right to run their unproven code on my computer,
what problems could ever come from a {border-width: 0.1%} ?

And I would also want to strike a call for adjacent borders to collaps
in the same way as margin does.

-- 
Rex

Received on Monday, 19 August 2002 17:15:14 UTC