> There are two possible conflicts: something has requested a *lighter*
> frame than the author requested (or none) or something has requested a
> heavier frame than the author requested (or a frame when the author
> indicated none). Moving towards the default or away from the default
> will favor the author or not, depending on the specific conflict, the
> default, and the author's specification.
Any modification of a frame from the default could be used to emphasize
an element. I assume that the author wants something different than the
default, which in my experience is pretty boring to look at. Actually, I
would not refer to the "default", but rather to the "standard style" for
the table in question, which might be from basic html style, the user's
favorite style sheet, an imported style sheet, or just a style
declaration given in the body of the document.
How about the following rule for resolving framing conflicts in tables,
assuming that the standard is declared before the table begins?
1) The UA should scan all elements of each row, and assign each element
of a given row the top and bottom frame widths that are most different
from the standard of all the frame declarations for the elements of
that row, so that all elements in any row have the same top and bottom
frame widths.
2) The UA should treat columns in the same manner as rows, but assign left
and right frame widths (instead of top and bottom).
3) If there are still framing conflicts (width, style, or color) between
adjacent elements of the table, the UA should examine framing
declarations for the adjacent elements, and again use the declaration
that makes the shared frame most different from the standard.
The rules for determining "most different from standard" should be
reasonably straightforward, and there shouldn't be too may cases where a
shared frame is requested to be on both sides of standard, at least not
by good authors.
One example of the use of these rules is a request for a change in frame
properties around a specific <td> element for emphasis. Following the
rules above, the UA would change the frame width on the top and bottom of
all entries in the same row, and on the sides of all entries in the same
column, so that all elements line up properly. It would then draw a
special frame around the desired element if requested, while following
the standard declaration for all other elements (except of course for
elements with common edges), as illustrated below (for a simple increase
in frame width).
OOOOOOOOOOOOOO O OOOOOOOOOOOOO
11111111111111 1 1111111111111
22222222222222 2 2222222222222
33333333333333 3 3333333333333
44444444444444 4 4444444444444
55555555555555 5 5555555555555
This also emphasizes all other entries in the same row or column, but the
reader's eye would naturally be drawn to the desired entry as the author
intended, as shown below. If the author want to emphasize the difference
between 2 rows, he could make a table like this by specifying a different
top and bottom frame in the <tr> element.
000000000000000000000000000000
111111111111111111111111111111
222222222222222222222222222222
333333333333333333333333333333
444444444444444444444444444444
What do other people think of these rules? They would add power to the
table facility, and I can't think of any situations in which they would
cause bad effects in the hands of a reasonably careful author. There
aren't many situations where an author would want an extra-thick frame
around one element and an extra-thin frame around another in the same row
or column.
David
Work: seibert@hep.physics.mcgill.ca Home: 6420 36th Ave.
Physics Department, McGill University Montreal, PQ, H1T 2Z5
3600 Univ. St., Mtl., PQ, H3A 2T8, Canada Canada
(514) 398-6496; FAX: (514) 398-3733 (514) 255-5965