At 3/7/2007 11:13 PM, Thierry Koblentz wrote:
What's your definition of tabular data? Actually, what if there was only one
row for our example?
Would you consider marking up the following with a table?
President..................................John Smith
What for you makes a list of "name/value" pairs tabular data?
Pairs, triplets... the number of columns doesn't matter. For me a
table is a dataset naturally structured in rows and columns -- in
which everything in each column belongs to one class of data and
everything in each row is one associated group. In your example, the
two columns appear to be Position and Name. It's the fact that
everything in the first column belongs to the class of Positions and
everything in the second column belongs to the class of Names that's
my clue that this is a regular dataset, and therefore a likely
candidate for table markup in HTML.
The fact that it can also be marked up as a definition list is great
-- that gives me two tools to choose from.
"what if there was only one row for our example?"
OK, I'll bite. Would I bother marking up a single row as a
table? For that matter, would I bother marking up a single row as a
definition list? Or any kind of list?
I guess it would depend on the circumstances. From where does the
data originate? What's its purpose and function? Does this single
row belong to a family of similar datasets elsewhere on the site that
have more than one row? For easy, consistent styling I'd likely
choose a single markup for all those datasets that have a similar
look and feel and function.
One of the most important criteria for choosing table or list markup
is going to be flexibility of styling. Tables can be restyled a bit
but they have more restrictive limits than lists which are
wonderfully malleable. I might choose DL simply to give myself and
future designers more latitude in styling. Today's leader dots might
yield to tomorrow's vertical stack of bold position over normal name.
To continue probing the edges of the problem, what if you strip the
positions from your dataset and leave just the names? Is that single
column tabular data? I say yes because everything in that one column
belongs to a single data class. Would I mark up such a dataset as a
table? Not usually, but I might if it belonged to a family of
datasets that were marked up as tables. If it stood by itself I
could choose among table, UL, and OL, and would likely choose list
markup to provide the greatest freedom of styling.
Thierry, I understand that you're looking for simple criteria for
deciding when to use table markup and when not to. Personally I
don't think hingeing it on the number of columns is the way to go. I
think the reality of HTML is that, when a dataset is sufficiently
simple in structure and number of columns, there's more than one way
to mark it up that can be equally effective structurally. A
single-column table can be an unordered list. A two-column table in
which the first column contains sequential numbers can be an ordered
list. A multiple-column table in which the first column(s) contains
row heads can be a definition list.
In the absence of any strong semantic cues, I'll choose a structure
for its stylability. That might sound like semantic heresy, but I
think it's inevitable when even so sparse a language as HTML gives us choices.
Regards,
Paul
__________________________
Paul Novitski
Juniper Webcraft Ltd.
http://juniperwebcraft.com
*******************************************************************
List Guidelines: http://webstandardsgroup.org/mail/guidelines.cfm
Unsubscribe: http://webstandardsgroup.org/join/unsubscribe.cfm
Help: [EMAIL PROTECTED]
*******************************************************************