[thelist] front end design: liquid design

[Gunlaug Sørtun]

Felix Miata wrote:
> On 2008/10/30 20:23 (GMT+0100) Gunlaug Sørtun composed:
> 
>> Felix Miata wrote:
> 
>>> It boils down that the only relevant relatives in CSS are the em 
>>> and the ex. Fortunately they can work, if only more than a token 
>>> number of developers would use them for something besides 
>>> arbitrarily shrinking most text from whatever user preference 
>>> sizes happen to be.
> 
>> That's the problem with 'em'/'ex' - they're locked to font-size.
> 
> I've never seen a demonstration of anything like that sort of 
> assertion. I'm a firm believer that em is very good, and clearly the 
> best CSS tool for most non-image sizing, with nothing else coming 
> close to good for anything besides web-hosted magazine pages.

An 'em' is based on font-size - locked to it, and I'm a firm believer
that this locking reduces its usefulness for anything but font-sizing
and minor adjustments that are directly related to font-size. I use 'em'
a lot for adjustments - to prevent early breaking when documents are
subjected to font resizing, but not for whole-page layouts.

>> If used _directly_ as units for element-dimensions they easily 
>> force layouts wider than the available screen if one sizes text up,
>>  or cramp everything together in very narrow "pages" in those very 
>> rare cases were text is already large enough to allow for 
>> down-sizing.
> 
> Again, I've yet to see a demonstration of this that doesn't also 
> incorporate excessive attempt to control. CSS is not a page layout 
> program. Its purpose is style suggestion.

Right, so we "suggest" the best we can with what we have at hand. It
only has to work reasonably well in browsers that have a reasonably good
CSS support - and where that support is "on".

> There's a range of reasonableness in content size to viewport size 
> relationship beyond which no reasonable person should expect results 
> to remain reasonable. Most people who want or need huge text don't 
> try to use little windows, just as people who like mousetype 
> generally avoid large windows.

Most end-users I know of that need larger text, try to bump up text
after they have reduced resolution on relatively small screens for the
same reason. These end-users tend to curse 'em' sized layouts because
they force them to scroll horizontally to read whole lines, and prefer
the more stable and predictable 'px' sized or '%' sized layouts that
either stay at fixed width or adapt to available browser-window width.

Not too many "conditional" layouts that allow font-resizing or
page-zooming (except in IE7) while adapting reasonably well to available
browser width around, and I haven't received any comments about those
"conditionals" since they tend to work reasonably well for most
end-users regardless of their needs and/or preferences. End-users simply
don't notice, or care about, what method is used as long as it works.

>> The 'em' unit can be useful for 'max-width' in 
>> percentage-dimensioned - what I call "conditional" - layouts,
> 
> Very.

So that's what I sometimes use 'em' for.

>> but it is pretty useless for dimensioning anything but text beyond 
>> that.
> 
> No way. Got some proof?

Your own examples - those elements/layouts that are sized directly in
'em' - not the "conditional" ones. I think most of your examples turn
out bad, and only work somewhat when I use the "fit to width" option -
in Opera, or turns off styles completely - in same browser. Don't work
any better in other browsers, btw - in my opinion, of course. Whether
that's good enough proof or not is of course open for debate, but you'll
have to prove me wrong if you want me to change my mind.

>> Browsers generally not so great scaling of objects other than text,
>>  makes the 'em' even less useful beyond basic text-scaling and 
>> "dimensional conditioning".
> 
> The significance of image scaling quality varies by context. On most 
> web pages, absolute image quality is secondary if not tertiary or 
> inconsequential to what the page is about. That image scaling quality
>  is less than ideal should not be a significant factor in styling 
> decisions for most web pages.

An alright argument when one is in total control. Useless as argument
when someone else is and one wants to keep the job.

> If users were permitted to see more of how bad some browsers scale 
> images, there would be more demand for browser improvement, and more 
> gravitation to browsers that do better. The current mediocre scaling 
> state tends to perpetuate itself by maintaining disincentive to 
> improve.

Yes, it is bad that we're underwriting status quo, but too many front
end coders/designers would be out of business if they didn't follow the
muddy stream. Don't think we have a strong enough "union" and strong
enough "common interests" to force anything through here.

>> Today the status is that "nothing works really well", which means 
>> it is not possible for anyone to create something anywhere near the
>>  ideal web publishing solution that can/will work for all 
>> end-users, with html/CSS alone - and the alternatives are not great
>>  either.
> 
> Em styling works well enough that magazine pages feigning to be web 
> pages should not be the general dev practice that users actually 
> suffer on today's web. Those who _really_ need to publish in the 
> traditional sense, pseudo-paper, usually ought not to be using 
> HTML/CSS, and probably PDF instead. Most who think that's what they 
> need probably don't actually need to, and should be content to 
> publish in synergy with the web's inherent nature using HTML/CSS, 
> including sizing predominantly using em.
> 
> Now I'm really disappointed that you never got down here during your 
> USA trip so that we could have figured out why we have this apparent 
> difference of opinion. :-(

Sorry, ran out of time and initiative. Maybe another time.

Regarding difference of opinion:

I think it boils down to that "we can do whatever we think is right or
best or whatever in web design, but not everything will pay off in
day-to-day work and/or leisure."

You're idealistic, and think you have found a method that works, while
I'm pragmatic and won't promote widespread use of any method that has as
little chance to work as other methods - in that they won't work unless
the conditions are right.

As a front end coder/designer I'll settle for a bunch of compromises
that will make it work somewhat for most - almost regardless of
conditions, despite the fact that I'm then underwriting a whole range of
weaknesses in existing web design practices, User Agents and end-user
behavior and preferences.

As an end-user I'll keep on setting my 'minimum font size' and use other
options as I see fit, regardless of whether designs fail or not.
It is up to each front end coder/designer to decide what ones
end-products should be able to take before they "fail" - and make ones
creations live up to set goals by providing them with the right set of
"suggestions".  It is not my business as an end-user to "protect" other
people's designs by preparing the "right" conditions for them.

regards
	Georg
-- 
http://www.gunlaug.no