Re: table-height-algorithm-026 invalid

Le Ven 3 décembre 2010 12:29, L. David Baron a écrit :
> On Friday 2010-12-03 19:07 +0000, Arron Eicholz wrote:
>> On Friday, December 03, 2010 10:32 AM L. David Baron wrote:
>> > On Friday 2010-12-03 17:32 +0000, Arron Eicholz wrote:
>> > > On Saturday, October 16, 2010 3:04 PM L. David Baron wrote:
>> > > > http://test.csswg.org/suites/css2.1/20101001/html4/table-height-algo
rithm-
>> > > > 026.htm
>> > > > http://test.csswg.org/suites/css2.1/20101001/xhtml1/table-height-
algorithm-026.xht
>> > > > claim to be testing the assertion:
>> > > >   The baseline of a cell is determined by the bottom content
>> edge of
>> > > >   a cell when in-flow line boxes and table rows are not
present.
>> > > > However, both cells actually have in-flow line boxes.
>> Furthermore,
>> > > > the two cells should only align if the default padding and
>> border of
>> > > > a button exactly match the half-leading that results from the
>> normal
>> > > > line-height, which is not guaranteed by the spec.
>> > > Fixed based on Gérard Talbot's feedback
>> > I think the current version is still slightly off.  In particular,
>> the new version
>> > appears to work by using margin-left:-10px to cause two runs of _
>> characters
>> > to overlap.

If the testcase

http://test.csswg.org/source/contributors/microsoft/submitted/Chapter_17/table-height-algorithm-026.xht

uses instead margin-left: -4px (line 31), then the overlapping of
underscore characters would be reduced to the strictest minimum (1px).


Proposed modification
---------------------

line 29:            div
line 30:            {
line 31:                margin-left: -4px;
line 32:            }

I am not sure what's causing that small horizontal gap in Firefox 3.6.15
to begin with.

[
Addendum: Another idea is to just change the test and state that there
should be 2 black lines and that they should be vertically aligned
perfectly, side by side. That way, the testcase would elegantly avoid,
overcome the issue of the size of horizontal gap (no need to measure
it), without a negative margin and avoiding the antialias issue.

Expected results would look like:
______  ______

"Test passes if there are 2 black lines vertically aligned at the same
position"
]

 However, with subpixel antialiasing, the part where
>> they overlap
>> > can be darker because it's been double-drawn, which makes the line
>> not
>> > quite continuous.
>> Fixed changed the color to black.
> That's not sufficient to make the line continuous.
> (And I shouldn't have specifically mentioned subpixel antialiasing;
it's an issue with any type of antialiasing.)
> -David
> --
> L. David Baron                                 http://dbaron.org/
Mozilla Corporation                       http://www.mozilla.com/

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

CSS 2.1 test suite (RC5; January 11th 2011):
http://test.csswg.org/suites/css2.1/20110111/html4/toc.html

CSS 2.1 test suite contributors:
http://test.csswg.org/source/contributors/

Received on Sunday, 20 March 2011 21:24:00 UTC