RE: [Mantisbt-dev] DIVs Tables and CSS

[Kenzaburo I. <ke...@30...>]

Mmm discussion.

>> * Tables are much more widely understood than DIVs+CSS.
> 1) Yes, but that's because most people haven't tried/understood how to do
> it
> properly. Everybody copies from other people's table based design. This is
> a
> bit like arguing "all flies eat shit -> there must be something right about
> it"

It's funny how bad design propagates but I don't it's our place to change
this.  Users have a certain expectation on the app.  Witness the *continued*
use of = and == for assigment and equality.  Yet when something reasonable
like := comes along it's the subject of idiotic flamewars.

De facto standards place a certain amount of preconceptions into users.  They
will expect to see tables when they look at the code.  When they see DIVs and
no tables they will be confused, disturbed, frightened, frustrated, and
eventaully angry.  Given the certain fact that people are only using Mantis as
a tool in their greater goal I know this will happen.

>> * Tables give you more immediate information about the layout
>> of a document than DIVs+CSS.
> 2) Yes, but that's not the original intention of using tables. They should
> not be used for layout. This technique stems from DTP People that haven't
> understood that *ML separates content from layout and that layout is
> dynamic,
> not fixed. They always want to have everything placed exactly 12.63 mm from
> the left border.

The goal I have in using tables is to place items in basic quadrants and
alignments.  I don't have *any* idea of how to do this with DIVs right now. 
I've tried and failed miserably.

I try very very hard to avoid exact placement of any items.

I'm sure we all realize that things rarely used only as designed.  IRC and
email weren't designed to transfer multi-megabyte files.  Yet this is a key
feature.  The cat is out of the bag and there's no way you, I, or anyone short
of maybe MS can fix this.

>> * Both require you to look at the HTML and the CSS if you
>> want to do formating.
> 3) That's no point for/against using DIVs+CSS if I understand you
> correctly?

Right, it just means that with the current solution or a new solution you
still have to look at both files. Some people seem to think that CSS means you
don't need to touch the HTML, maybe, but you still have to look.

>> * I would argue that tables are easier to quickly bend to
>> your needs, especially if you are not as familair with CSS.
> 4) Hm, but that's because of 1) isn't it?

Yes.

>> * DIVs+CSS is not perfectly supported across even the latest
>> generation of browsers.
> 5) The things I tried worked on IE5.5, IE6 & Mozilla 1.0. Can you give me
> an
> example, where the layout is different?

I tried centering a 50% div block in the middle of a page.  Worked in IE6 *or*
in Moz/Opera but not both.  I did a google and found some pages that indicated
this was a significant problem.  I've since achieved success but this helps
prove a point: it's not simple to do a simple action using just CSS.  I
consider myself pretty expereinced but I had problems with it.  I spent about
3 hours twiddling with main page to use only DIVs.  It looks fine now but I
had to use 3 DIVs and spend 3 hours tweaking items until they worked.  I think
that's an indicator of how difficult this can become.

And actually I didn't achieve success since when I add a border the two 50%
width items are NOT the same width because they had different padding sizes. 
I consider this a nightmare for non-experts.

>> * Too much work to convert ;P
> If you leave out the "too" I'd agree. I did the main_page at home. If you
> build up your css systematically, it's not too tough. It even saves some
> space, because you don't need the <tr> anymore ;-)

This seems to be the most valid point to me.  But I don't think the tradeoffs
justify it.  I also think it's too much work.

> The idea is not to drop all tables. Pages like view_bug or report_bug can
> only be done with tables. The idea of using more DIVs+CSS is to be able to
> layout/place the DIVs independently from each other and to get away from
> the box in a box in a box approach.

You mention a table reduced goal.  Here's what I am open to: keep the tables
for alignment and sometimes percentage widths but wrap the displayed text in
span or div blocks.  Would that be sufficient?

I think everyone needs to think of our target audience in terms of fiddling
with code.  We can target people who are experts and either know or are
willingly to learn objects, templates, proper CSS and HTML design, and
anything else I've left out.  This leaves us with a dramatically reduced
number of people.

Or we can target your average developer, someone working on their Java, C,
C++, web page, or whatever.  They are non-experts and are using Mantis because
it suits their needs.  They want to tweak something and so dig into the code
and start to flail around.  They find something that looks familiar and latch
onto it and start tracing and understanding code form there.

Now, I'm making assumptions that a) there are many non-experts who fiddle with
the code b) they are fairly successful with the curent code and c) they would
be less successful with the proposed expert-level changes.

I'm very sure on a and c.  I'm completely unsure about b.  If b is close to
zero then there's no point in my objections.

Oh, I'm conducting a poll: http://mantisbt.sourceforge.net/vap.php3 - right
now templates is winning.  (might have to reload a few times)

Thanks,
-Ken