W3C home > Mailing lists > Public > www-style@w3.org > December 2010

Re: [CSS21] 4.3.2 Lengths (reference pixel?)

From: Peter Linss <peter.linss@hp.com>
Date: Tue, 14 Dec 2010 23:46:46 -0800
Cc: "www-style@w3.org" <www-style@w3.org>
Message-Id: <EE13C14E-3504-4F8B-AC7B-B6D20304B6CE@hp.com>
To: Felix Miata <mrmazda@earthlink.net>

On Dec 14, 2010, at 10:45 PM, Felix Miata wrote:

> On 2010/12/14 18:29 (GMT-0800) Peter Linss composed:
>
>> Felix Miata wrote:
>
>>> You've not used many Linux distros over the years.
>
>> Not for desktop, daily use systems, no. The same as 99+%* of all  
>> other
>> web users.
>
>> Do you guys want us to break the web for 99%* of the users to satisfy
>> the<  1%*?
>
> Sure, tyranny of the minority by the majority is always justified,  
> right?

Please, "tyranny"?

>
> I don't understand what's so awful about introducing logical mm/cm/ 
> in/pt/pc
> etc. as lmm, lcm, lin, lpt, lpc to pacify the supposed majority  
> without
> breaking what works for the minority, and make unequivocal that CSS is
> dominated by units that do not directly correlate to apparently  
> identical
> units well understood in the whole rest of the world.
>
>> * yes, I'm making that number up to make a point, don't start a  
>> debate
>> over it, it won't change anything.
>
> http://www.w3counter.com/globalstats.php shows Linux presence on the  
> web is
> considerably more than inconsequential at 1.4%+. Elsewhere I've seen  
> that
> number more than doubled.
>
> Also, the 99%- of others are not all broken. And as I've written  
> already
> upthread, some web pages are built using absolute units for the  
> precise
> purpose of enabling configuration correction, or demonstrating  
> configuration
> error. They rely on CSS units that appear to be world standard units
> continuing to mean what they used to mean in CSS, and still do in  
> all other
> environments. Hijacking standard units to mean something different  
> promotes
> chaos. Language in the specs explaining special contextual meaning  
> will not
> prevent it.
>
> Additionally, tying so-called absolute units to a certain number of  
> pixels
> serves to promote the bogus notion that widespread use of px for  
> sizing is
> usually appropriate and good, when the accurate statement is that  
> use of px
> for sizing is usually bad, as an unnecessary and unfortunate  
> restraint on
> adaptability, accessibility and usability; convenience for designers,
> inconvenience for web users.

Ok. I think I get what the fundamental disconnect is here. We're NOT  
changing what an INCH is (or a cm), we're changing what a CSS PIXEL  
UNIT is. The change is that the ratio between CSS px units and in  
units is now fixed, not the ratio between device pixels and in units.

There was never any guarantee that a CSS px unit would ever and always  
be one device pixel. In fact it was always specified that it would not  
be on high resolution devices. Guess what, most screens are high  
resolution devices nowadays, and the trend is towards higher  
resolution. We allow conforming browsers to optimize the ratio of  
device pixels to CSS px units for screen media. This may not work out  
so that a CSS in measures an inch on the screen, deal with it. It most  
likely didn't work out that way for most users before either.

A CSS in unit should still work out to an actual inch in print media.  
It should still work out to an inch in screen media when practical  
_and the device is at 100% zoom_. There are, and always will be,  
screen media situations when any attempt to make an in unit actually  
measure a physical inch would be impractical, if not impossible. Think  
projectors. Do you really want the projector to measure the distance  
to the screen and adjust the display so that 12pt type is really 12pts  
tall? Of course not.

There will also be print media situations where this rule will be  
broken, think billboards.

The bottom line here is that this issue has already been discussed  
both publicly and within the working group. Very smart people who have  
been dealing with exactly this issue in CSS since the late 90's were  
involved. All of the points brought up in this thread have been  
brought up before, and they have all been considered. And a decision  
has been made. We have no intention of revisiting this topic unless  
something fundamental changes or new information is brought to the  
table. Restating the same arguments over and over does nothing to  
help. Rhetoric does not help. It just wastes valuable cycles that  
could be better spent elsewhere.

Are there examples of pages that broke because of this decision? Sure,  
we knew about the use cases. There are also examples of pages that  
were broken the other way. We made a decision, I believe we made the  
right decision. WIll this ever be revisited? Possibly, when the  
situation has changed and allows a better solution. For now, we have  
bigger fish to fry.

Moving on.
Received on Wednesday, 15 December 2010 07:47:24 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:35 GMT