Re: HTML 4.0, TABLE/TD width units incongruity

> Basically MS fixed Netscape's bug. <g> I hope the
> Mozilla.org/gecko/NGT folks fix it likewise. Logically, the
> percentage should be of the object of which the sized item is an
> element. That is, a table is a percentage of a page because it is an
> element of the page; since a cell is an element of a table, its
> width as a percentage should be of the table.

The orthogonal approach is the way to go: whenever you have a WIDTH
parameter and % value, it should always be taken directly from the
containing element, and specifically from the "inner width", e.g. the
width that is available for the content, excluding any possible
paddings.

In case of TD, the containing element is not TABLE, but TR (minor
point, but is meaningful, as there can be padding between TR and the
TABLE proper border).

And with orthogonality I mean, that you should be able specify WIDTH
to almost any element. I once wanted to design input forms that would be
independent of the page width. Without a WIDTH attribute and %-value
this is impossible (I needed free text entry box).

If there are attributes that make sense with many elements, allow them
to work with them similar way with same specification. Do not make
zillion special cases (like WITH with TABLE and TD), which just bloat
the implementation without gaining anything for the user.

I don't think it is sensible to cripple the standard, just because one
browser implements something incorrectly.

Received on Thursday, 14 January 1999 03:33:59 UTC