Am 12-Jun-02 schrieb David Gerber:
...snipp[]
> On a lot of places. Try to find out what happens if you specify
> a table WIDTH="200%" for example.
AFAIR the w3c-specs it should be handled this way:
width & height override the 'natural' dimensions
of an object.
Using the percentage form for such declarations
invokes the following:
- the specified width depends on the environs
and therefore can only calulated by the UA
(user agent, here: browser).
E.g. if a table is placed plain in the document,
and the page size is 800x600 pixels the
maximal width is clear, but if the table is
placed in another table cell (width 350)
the table width is limited by this dimension.
- width's greater than 100% are not specified
by the w3c, so there might be several possibilities
to handle it:
a. strictly use the specified width
(not recommended, see example above)
b. the UA has to calculate if the object
fits in the available environs with
the defined widths, if so it's alright,
if not the object has to be resized.
c. first step like b., but if the object
doesn't fit, leave it out, only a placeholder
is shown.
d.-z. there might be some more dull ideas
to break the layout.
IMHO only b. is the right way to handle it,
because any other way causes an inadequat
alteration of the layout.
For the height of an object this also is true,
except for the defined height of 1000%
(I can't remember right now for what
objects this height is defined, but for tables
I'm sure).
Here the w3c clearly states that this
height allows the table to be stretched
over the visual vertical available space,
so the user has to scroll.
>> > Strictly following the w3c HTML specs doesn't work because
>> > they leave room for different interpretations.
If you read the specs and collect information
about 'layout' works and is handled
your expected room for interpretations
shrinks dramatically, IMHO.
> See above. There are others, like the <IMG><IMG> vs
> <IMG> <IMG>. Where it's unsure if the first one must be
> vertically layouted. I think it shouldn't but MSIE does, so
> websites do expect it.
I haven't consulted the w3c specs for this issue,
because I'm not programming a HTML-UA ;-),
but I'm sure there's a description on
how the overall layout is ment,
vertically or horizontally in first place.
I guess it's ment as you described,
because HTML offers the tags for paragraphs
(p-tags, which in most cases are ignored by the
majority of the users) and also for defined
line breaks (br-tags).
So one might expect that the objects are 'glued'
together horizontally, line breaks are calculated
if not definded explicit.
In your first first example the space between
the two images is zero, because HTML has no
predefined 'object-spacer', AFAIR, in your second example
the space must be exactly one blank, whether this
construction can be broken by a line break
I can not decide (one must read the specs).
You have to get the same layout by using
<img> <img>, execept the fact that this construction
can be broken by a line break.
>> And what is wrong with room for interpretation?
>
> It gives different result for the layout of the page depending
> on what browser you use.
Whether this is a critical point or not I cannot decide,
because the page layout already differs if the user
defines his own replacement fonts, his own colors,
not to mention the color depth of the display.
>> Take this room (layout) and let the client user do their
>> interpretation.
>
> Then the user complains because the site looks weird or is
> unuseable.
IMHO therefore one should stick as close as possible
to the w3c-html-definitions and have a look at
how print layout works (only the basics are ingenious
the details cannot be adapted for online use, IMHO).
Regards
NR
--
Remark of the day :
For a warm boot,
put your computer in dryer for about 20 minutes.
_____________________________________________________________________
Voyager Mailing List - http://v3.vapor.com/
Voyager FAQ....: http://faq.vapor.com/voyager/
Listserver Help: mailto:[EMAIL PROTECTED]?Subject=HELP
Unsubscribe....: mailto:[EMAIL PROTECTED]?Subject=UNSUBSCRIBE
