W3C home > Mailing lists > Public > public-css-testsuite@w3.org > May 2010

More Microsoft testcases with the fractional pixel problem: 1cm, 1mm and 7.5ex in border testcases

From: Gérard Talbot <css21testsuite@gtalbot.org>
Date: Tue, 18 May 2010 10:53:12 -0700
Message-ID: <c64e4b6c54cc7c85e9b59cb50ef071cd.squirrel@cp3.shieldhost.com>
To: "public-css-testsuite@w3.org" <public-css-testsuite@w3.org>
Cc: "Arron Eicholz" <Arron.Eicholz@microsoft.com>
Hello all,

Arron wrote:
> If there are cases that have this fractional pixel problem let me know.

There are at least 3 additional testcases (border ones: all involving
1cm or 1mm or 7.5ex) which show a tiny red border (a machine running the
test suite would detect the 1px red border) or which show a small
difference of height between 2 black squares and which are related to
the fractional pixel being truncated by Firefox 3.6.3, Konqueror 4.4.3
and Safari 4.0.5:

1px red border:
http://test.csswg.org/source/contributors/microsoft/submitted/Chapter_8/border-bottom-width-036.htm
(1cm)

1px red border:
http://test.csswg.org/source/contributors/microsoft/submitted/Chapter_8/border-bottom-width-047.htm
(1mm)

Now this one is much more troublesome:
http://test.csswg.org/source/contributors/microsoft/submitted/Chapter_8/border-bottom-width-083.htm

If the tester carefully examine the 2 black squares in Firefox 3.6.3 or
Konqueror 4.4.3, (s)he will notice a tiny difference of height. So,
visually, the testcase "fails" for those browsers.

The building logic of the testcase has 2 problems:
        <style type="text/css">
            div
            {
                display: inline-block;
                font: 12pt ahem;
                width: 1in;
            }
            #reference
            {
                background-color: black;
                height: 1in;
                margin-left: 5px;
            }
            #test
            {
                border-bottom-style: solid;
                border-bottom-width: 7.5ex;
                height: 0;
            }
        </style>

1- The testcase uses 4 length units: pt, in, px and ex! Shouldn't such
testcase only use the same unit for all elements? Why resort to 4
different length units in the same testcase?

2- The testcase does not use different colors (say, green and red) and
the overlapping technique. So, a failure of 1px may not be easily
noticed if the tester is asked/invited to compare respective heights..
but a machine would

When examining values in DOM inspector, the difference of [offset]height
is 1 pixel. When examining values in Konqueror's domtreeviewer, #test's
border-bottom-width is 97px while #reference's height is 96px. So that
testcase definitely has the fractional pixel truncation problem.

There may be other testcases still undiscovered/unreported yet.

regards, Gérard
-- 
Contributions to the CSS 2.1 test suite:
http://www.gtalbot.org/BrowserBugsSection/css21testsuite/

CSS 2.1 test suite (alpha 2; March 16th 2010):
http://www.w3.org/Style/CSS/Test/CSS2.1/20100316/html4/toc.html

CSS 2.1 test suite contributors:
http://test.csswg.org/source/contributors/
Received on Tuesday, 18 May 2010 17:54:19 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 20 September 2010 17:52:02 GMT