Outer Court

Tech - Posts to comp.infosystems.www.*

From: "Philipp Lenssen" <phil@mrinfo.de>
Newsgroups: comp.infosystems.www.authoring.stylesheets
Subject: Re: CSS Despair
Date: Wed, 26 Sep 2001 10:16:16 +0200
Message-ID: <9os2to$fg$2@swifty.westend.com>
"Geoffrey Pointer" <geoffrey@bigpond.net.au> wrote in message
> Philipp Lenssen said:
> > You say it yourself: you had to follow guidelines to just get it right
> > *one* machine. HTML is for a whole lot more than that. It doesn't just
> > different machines, it covers different media, too.
> My point was that C as a programming language has a standard and that
> standard is very well met by CodeWarrior as an implementation. That
> is also very well implemented on many other systems by other companies.

I got that point. What I tried to say was that HTML has to cover more than
that. Can you run your C programs easily on a mobile phone? Do your programs
work on practically any screen size? Do they support text-to-speech?
Complete user configuration when it comes to layout issues? Run on any
system? Can they be automatically translated to different languages by free
services? Are they instantly available via internet, no installation needed?
And so on.

A web application (as this is the only thing you could compare to a C
program!), no matter what language is used server-side (Java, VBScript, C,
Perl, PHP, Python, or whatever you use), *has* to finally give the whole
cabundle to "the enemy", the unknown client. Many things that are build into
HTML by design take hard work in other, predefined (and seemingly "perfect"
and "working" environments). Every client side scripting will introduce
hundreds of complications you just can't test. So be defensive and deliver
only minimal scripting to the client (which may be none at all in most
I think you can write relatively fail-safe HTML. Of course, with CSS it's
nearly impossible to be on the safe side; you have to rely on the browsers
implementation to not cover CSS just partly, but to either ignore *or*
understand it completely (and no, HTML4 Transitional "layouting" is no
solution either; it will just introduce different problems).

> C
> ports very well if you write ANSI strict code. The problem occurs with
> porting interface, which I'm well aware is the problem that must be solved
> with web development.


> CSS, as far as I understand it, is aimed at providing a cross platform
> solution to these problems, but it seems to me that NO company has
> implemented CSS well.

Give IExplorer, Opera, and Netscape 6 some credit -- they deserve it for

> So what's up with the major developers of user agents (Microsoft,
> Netscape et al)? Maybe that's where I'm being idealistic, maybe they're
> like everyone else?

Maybe, just maybe CSS2 is really very hard to implement. I don't think the
Mozilla project lacked motivation, time or people to implement it. (OK,
there's never enough time.) You can also see they tried really hard to make
it standardized. I don't know how much they actually did right in the end,
since of course everyone of us has to write CSS that works with most of the
browsers, and not just one, which basically forces one to disregard nicer
features (like min- and max-width, or using more complex selectors). So what
we actually need is not *one* browser to get it right, but the *majority* of
browsers. (And by getting right I include browsers ignoring CSS, though of
course they should be a minority in medium screen, or else it would be not
really usable.)

> > Hmm, could you be a little more specific?
> I thought if you read the message as a whole that it was very specific. I
> pointed out examples where two major browsers don't agree on
> of the standards. (In fact I've found more since then and many more
> previously besides) I referred to the long term futility of hacks.

What I like about the CSS hacks is that they are mostly reduced to a hand
full of (easily readable) files, no matter how big the website. If you ever
written table layout you know how much of a pain it is to change a hundred
of websites. (Some people simply switch to templates, but I don't like the
idea of still finally delivering garbage to the client, no matter how much
work was reduced server-side.)

> > ... anyway most of us are getting paid to cope with web developing. Why
> > should things ever be that easy?
> This sounds a bit self defeatist to me. If web browsers implemented
> standards correctly and completely, you would still have to write your
> and debug it. It would still be hard work.

(This was my try at sarcasm, and I guess not a successful one.)

Philipp Lenssen
M+R Infosysteme