RE: CSS1 and tables

Let me preface this post by saying I'm not speaking as an official
Microsoft spokesperson; I usually don't stick my foot too far in my
mouth, so I can speak for Microsoft sometimes, but right now I'm sharing
a personal viewpoint, not an official Microsoft stance.

> David Perrell [SMTP:davidp@earthlink.net] wrote:
>I knew that worked with FONT, but thought it was illegal HTML. If
that's the
>legacy you chose not to support, I'd commend your forward-thinking
strategy.
>Inheritance gets crazy without hierarchical containment.

I would commend my own forward-thinking strategy if that were the case.
However, we took this approach in Spry Mosaic when I worked for Spry -
in large part due to my naive insistence on doing so - and took a
tremendous support hit from people complaining our Web browser sucked
because it wouldn't work with their pages, which must, of course, be
perfectly legitimate because they looked fine in another browser.  (This
is not a rant about our competition, BTW.)  And certainly, if we care
about actually GROWING our market share, we need to be perceived as
better than our competition, not worse.  By "better", I mean we have to
do most things as well as our competitors, with enough attractive
additional stuff that our product looks more enticing.  Well, from
personal experience as a person who browses the Web looking for standard
fare (looking for information on particular topics, buying concert
tickets, investigating particular products and reading people's
diaries), I know that up until a few short months ago, we would look far
more broken if we removed that backwards compatibility (call it sideways
compatibility if you want) than if we did not support stylesheets at
all.

Harsh words, especially when you consider they're coming from someone
who has invested most of the last year and a half of his professional
life in stylesheet support.

We just can't do this in HTML rendering and expect people to be happy
with our product.  Hey, let's say we're a perfect world, and Microsoft
and Netscape (who, let's agree, "own" >90% of the browser market right
now, and we can presume the other <10% would follow suit) are about to
ship product - and we call each other on the phone, and make a blood
pact to forget legacies and both follow the same very strict DTD (stop
laughing, darn you!).

Would we be able to convince everyone to upgrade?

>I suspect you are right about large numbers of sites with tables -- and
much
>else -- enclosed in a <FONT> element. But that legacy behavior was
never
>standardized.

Except inasmuch as that behaviour is displayed by >90% of the Web
browsers in use, and is therefore a de facto standard by a landslide.

>By not supporting inheritance of an ancestor's font properties
>into a table, you appear to be rejecting both the CSS1 inheritance
model and
>the HTML specification in favor of support for non-standard legacy
behavior.

I disagree.  I have described how I personally do not consider this
"breaking the inheritance model," but rather an occasionally
inconvenient default stylesheet rule.  It's possible in the future we'll
be able to move that out of code into an actual default HTML stylesheet,
and then you the user will be able to make that choice.  This ABSOLUTELY
DOES NOT break the inheritance MODEL - it's only a defaulting of
properties on table cells.  I know for a fact that indeed our code DOES
inherit style properties from table to table cell, and then overwrites
those properties with initial values.  (Hey, look at font-family - it
DOES inherit properly through table cells, because we (Microsoft) added
it to HTML (yeah, I know, boo hiss - can't blame me for that one, I
wasn't part of the IE team then), and therefore we got to define whether
it inherited through table cells or not.  Obviously, we thought it made
sense to inherit it.  Netscape, to my knowledge, disagrees.

>'Legacy behavior' sounds nice, but the legacy of the last few years of
HTML
>development is a gaggle of inconsistent and unrelated elements. Please
don't
>corrupt the CSS inheritance model with 'exceptional' support for
vestigial
>perversions. Bite the bullet and cut the crap.

I'll keep it in mind for IE 5.0 to see if we can truly move this into a
default stylesheet, and then you can make that decision yourself as a
user.  I can, of course, pretty much guarantee that the initial default
stylesheet will support legacy rendering, however - not guarantee as a
personal thing, because quite frankly I agree with you, it's lame; but
on a professional level, I've been forced to understand who our customer
base is, and what they want - and when it comes to purity, the price
they're willing to pay is not always that high.

The one bright glimmer on the horizon is XML; there's not as much room
for interpretation of error handling, in my opinion.

	-Chris
Chris Wilson
cwilso@microsoft.com

Received on Tuesday, 7 October 1997 11:57:54 UTC