387250 - CSS class formatting is sometimes applied to XHTML table elements having no class attribute

Status: RESOLVED WORKSFORME

(Core :: Layout: Tables, defect)

Description

[angrywolf]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.4) Gecko/20070515 FreeBSD/i386 Firefox/2.0.0.4
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.4) Gecko/20070515 FreeBSD/i386 Firefox/2.0.0.4

I have defined a few CSS classes for the elements <table> and <tr>, and made a <table> that applies those classes. I also added two other tables without using CSS classes: one with the same border attribute, and one with a different one. The latter looked as I expected, but the first one seemed to be using my class settings, though it shouldn't.

Reproducible: Always

Steps to Reproduce:
Just see the attached test.html


Expected Results:  
It should have rendered both classless tables without applying CSS classes. Just like in the Konqueror Internet browser that comes with KDE.


The problem can be reproducable on Windows 98 SE too, with some version of Mozilla (dunno exactly, before Seamonkey).

Comment 1

[angrywolf]

Attachment: A test case showing both the right and wrong rendering of classless tables

Comment 2

[philippe (part-time)]

The table with the attribute [border="2"] seems to 'inherit' the border-collapse property, but only for table-rows inside <tbody>. Or put it in other words, the border-painting is suppressed on the cells inside <tbody> That is not quite correct.

And if the styled table has a attribute [border="1"], then the same effect happens to the first unstyled table in the testcase.

That is a bit similar to bug 142998, maybe

I see this on the latest trunk builds also.

Comment 3

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

This is because of the broken MapAttributesIntoRule in nsHTMLTableElement.cpp.  Everything involving the |tableStyle| variable can't actually be done according to the rules that function is supposed to follow.

Comment 4

[Phoenix]

Is this still actual? I see no difference in testcase rendering between FF and IE8