W3C home > Mailing lists > Public > www-style@w3.org > February 2000

Re: Shrinkwrap

From: fantasai <fantasai@escape.com>
Date: Mon, 28 Feb 2000 18:03:34 -0500
Message-ID: <38BAFEC5.86BF60@escape.com>
To: www-style@w3.org
Eric A. Meyer writes:

|    Because it's the most certain case.  Assume for a moment that
| setting all three to 'auto' would cause "shrinkwrap" effects.

IMO, setting just the width and one margin to 'auto' should cause
shrinkwrapping effects. Aside from that point..

| How does a user agent decide the shrinkwrap width in a changeable
| environment?  For example, paragraph text which is half a line
| long on your desktop system's monitor might be four lines long on
| my PDA, or even 90% of a line on my desktop system.  For that
| matter, changing the width of your browser window on one system
| could radically alter the situation.

Precisely why I don't want to set a definitive value for the width.

The user agent calculates the width of each line as if the auto margins
were 0. Then the width of the content box is reduced by the difference
between the content width and the longest line. This space is added to
the 'auto' margin, or split between the margins if both are set to
'auto'.

How is the 'intrinsic width' of a table calculated? Tables without an
explicit width are currently shrinkwrapped; and paragraphs should be
even easier to deal with--less boxes.
This does, of course, prevent incremental display.

(cut)

|   So anyway, one assumes that you had a reason for wanting
| "shrinkwrapping."  Could you describe it in more detail?  Maybe it
| will provide us the beginnings of a path to adding such behavior
| to the specification.

Taking example of a list, used for purposes of a table of contents.

1. Introduction
2. Document Structure
3. Text
4. Content Structure
5. Links
6. Formatting/Markup
7. Appendix

Supposing I want a border around the list, one that reflects the size of
the visible content. Placing a border on this list would paint a box
which spans the entire width of the screen. Placing a width on the
content box would meet up with the problems you've described above
(varying screen widths, etc.).

If you want to see this, you can use Navigator 4.x's incorrect CSS
interpretation to view Chapter 3 of the MathML 2.0 draft. (I ran into it
on my search.)


Another use is centering a block of left aligned text (such as the above
list). This is actually where I ran into my problem.

You can also take your H1 example--it's much more elegant like
this, no?

For anyone who wants the content width to extend the whole length of the
containing block, they set the margins to zero.

Such a redefinition might cause compatability problems with CSS1/2,
though...
Received on Monday, 28 February 2000 18:10:27 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:04 GMT