W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > October to December 2001

Re: [CSS-TECHS] Test suites, font units of measurement

From: Jo Miller <jo@bendingline.com>
Date: Wed, 7 Nov 2001 08:36:28 -0500
Message-Id: <p05100306b80ee1cb5236@[]>
To: WAI GL <w3c-wai-gl@w3.org>
At 10:46 -0500 11/5/01, Charles McCathieNevile wrote:
>I had a look at the test page for fontsizes using iCab 2.5.3 on Mac OS X.1
>and it seemed to do what I expected, but it is not clear what it is supposed
>to look like. Do you know that there is already a fairly comprehensive test
>suite for CSS - should we use that instead?

Thanks, Chaals. Yes, by all means let's make use, whenever possible, 
of the test suite that exists. [1] The purpose of the preliminary 
test page I was working on [2] was narrow: I wanted to test which 
units of font measurement allow easy resizing of text by the user in 
various browsers (emphasis on the "easy" -- using a clearly labeled 
"size" button in your browser's toolbar qualifies as easy; whipping 
up a little javascript to custommize your browser's behavior does 
not). I was also interested in seeing the effects of different 
DOCTYPE statements on the way certain browsers (mainly IE for Mac) 
handled font sizes.

Here's some previous correspondence with Joel on that subject, for background.

>Date: Mon, 29 Oct 2001 15:44:22 -0500
>To: Joel Sanda <joels@ecollege.com>, "'Wendy A Chisholm'" <wendy@w3.org>
>From: Jo Miller <jo@bendingline.com>
>Subject: RE: Contributing to WCAG CSS Techniques?
>Hi Joel,
>I'm interested to hear your observations on the differences in the 
>way IE6 and IE5.x handle inheritance. I noticed a few oddities, and 
>I'm sure you're way ahead of me. [...]
>As for testing, I'll gladly take the Mac end (I've got a Windows 
>machine too but will happily leave it to you). Oh, I also have 
>AOL/Windows, but not AOL/Mac. When I need something tested on Linux 
>I just bug Ian. <grin/>
>BTW, I'm so fed up with the continued existence of Netscape 4/Mac 
>that I finally wrote a damn javascript to detect it and serve up an 
>alternative style sheet. So far it seems to be working.
>Real-world implementation is what I'm usually preoccupied with. If 
>we want people to trust and use our techniques document, we should 
>keep it as consistent as possible with the realities of current 
>browser support and browser bugs (overly theoretical CSS documents 
>are not very useful for designers who are trying to get their pages 
>to work cross-platform).
>The bear I've been wrestling with lately is what units of 
>measurement to recommend that people use for font sizes. It's not as 
>simple as saying "use relative units" -- especially with the way 
>that Netscape tends to compound the percentages. When we simply tell 
>people to use "em" or percentages for text sizes and leave it at 
>that, we often create more problems than we solve -- em and 
>percentages in pages of any complexity require a hell of a lot of 
>cross-platform testing and tweaking to get right, and most people 
>don't have the time, resources, or experience to see it through. If 
>we're pushing relative units, we need to give them more practical 
>implementation guidance (or point them towards a place where they 
>can get it).
>Are there any feasible alternatives? Lately I've had good test 
>results with the supposedly fixed measurements "small," "x-small," 
>etc. Their appearance is consistent and they can be resized easily 
>by the user on both IE/Windows and Netscape/Windows. Opera's easy 
>zoom feature works on the entire page, and IE/Mac allows easy 
>resizing of any text regardless of units, which brings me to a 
>Joel, do you have any idea why IE/Mac allows the user to resize text 
>easily using the "size" button in the menu-bar, regardless of units, 
>whereas IE/Windows still does not? If the IE/Windows "Size" button 
>worked for pages where font-sizes had been specified in pixels, it 
>would be nothing short of revolutionary. Designers could go ahead 
>and do what they want (namely, use pixels) without creating barriers 
>to accessibility.
>I would also be happy to tweak our own style sheet so that examples 
>render better on the Mac platform.  Shouldn't take major 
>adjustments. [...]


>In general, the following units of measurement allow for easy 
>resizing on IE 5-6 for Windows (the most popular browser), IE 5 for 
>Mac, Netscape 4.x-6 on Windows, and AOL 6 (?) for Windows:
>	-em
>	-percentages
>	-x-small, small, medium, large, x-large
>In Netscape for Windows, points can be resized, but not pixels. In 
>IE for Windows, neither points nor pixels can be resized. IE 5.x for 
>Mac and Opera for Mac/Windows allow the user to resize all text 
>(pixels, points, you name it). I'm still testing with AOL.

All of this should be considered preliminary, and I would appreciate 
input from people who know more than I do. Thanks!


[1] http://www.w3.org/Style/CSS/Test/
[2] http://jomiller.com/css/fontsizetest.html

Jo Miller

"It's not our job to appeal to the lowest common denominator, Doug.
  It's our job to raise it." -Jed Bartlet
Received on Wednesday, 7 November 2001 08:36:51 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 16 January 2018 15:33:39 UTC