Bug 8320 – MultiLengths Ignored on COLGROUP and COL Width Attribute

• Component: Tables
• Status: NEW

[D. Brodale]

Use of multilengths (or relative lengths) for the value
of width attributes set for a table colgroup or col seem
to be ignored by Safari. Percentage lengths work as
expected.

Although reported against OS X 10.4.x, the same errant
behavior can be seen with OS X 10.3.9, as well.

Recent versions of Opera, Camino, Firefox, and Internet
Explorer (both Windows and OS X builds) respect the use
of multilengths, which are defined in the W3C's HTML 4.0.1
Specification.

See:

Basic HTML Data Types (MultiLength):
http://www.w3.org/TR/html4/types.html#type-multi-length

Calculating the Width of Columns (Proportional):
http://www.w3.org/TR/html4/struct/tables.html#h-11.2.4.4

[D. Brodale]

Created attachment 7638 [details]
Test Case: Multilengths Assigned to COL Widths


Simple example of multilength use as value to width
attributes on COLs. Expected ratio of column widths
is 2:1:3 (100px:50px:150px). Safari renders as 1:1:1
(100px:100px:100px) instead.

[mvijay]

Hi,

I have been analysing the bug and here are some of my observations.

As described in the w3.org, while calculating or parsing the width of columns using proportional specification (ex, width="1*"), the CSSParser::parseValue() is returning "False" and the value is not handled. Whereas in other case, if the width of columns is percentage specification(ex, 
width="10%") then CSSParser::parseValue() internally parses the col width and applies 
the values accordingly.

This behaviour is same in Firefox as well.

So, is it Safari browser intentionally has not given implementation for the proportional specification while calculating column width?

[Pravin D]

Are we going to support relative

[Brent Fulgham]

Safari and Chrome agree on behavior here, but Firefox uses varied column widths. I'm not sure which behavior is correct.

[Radar WebKit Bug Importer]

<rdar://problem/96545449>