8191 - Parsing of BGCOLOR is incorrect

Status: VERIFIED FIXED

(Core :: Layout: Tables, defect, P3)

Description

[Crysgem]

Apprunner Build ID: 1999061308

The color of the cells renders as a decided grey as parsed by Ancestor 4.6 and
Enforcer 5.0, but a putrid... green... as extruded by the beast.

Comment 1

[Crysgem]

Attachment: Comparison of color - MSIE 5.0<=-=>Apprunner

Comment 2

[John Morrison]

Definitely 'trivial' ;-)

<table width="100%" border=1>
<tr><td>What if bgcolor contains more than 6 hex characters
(e.g., 7 * 'd')</td></tr>
<tr bgcolor="#ddddddd"><td>#ddddddd -- ~grey ??</td></tr>
<tr bgcolor="#aaaaaaa"><td>#aaaaaaa -- ~grey ??</td></tr>
</table>

Comment 3

[Daniel Roberts]

Oops...  Looks like this already has a simple test case if you ask me.

Comment 4

[Hixie (not reading bugmail)]

[TESTCASE]
The HTML 4.0 specification states that color attributes should use the format
specified by sRGB. sRGB does not specify a format. HTML does, in the DTD, give
a format: just six digits, prefixed by a hash. I suggest that the CSS format be
used (that is the easiest to use, since we then just need to pipe the color
from HTML to the CSS engine unmodified). The CSS format is a superset of the
HTML format.
If we use the CSS format, then colors should be specified using either three
or six digits. Seven digits is invalid.
Thus, what we should be doing is ignoring the colors above altogether.

HTML4: http://www.w3.org/TR/REC-html40/types.html#h-6.5
HTML4 DTD: http://www.w3.org/TR/REC-html40/sgml/loosedtd.html#Color
SRGB: http://www.w3.org/Graphics/Color/sRGB
CSS2: http://www.w3.org/TR/REC-CSS2/syndata.html#value-def-color

CURRENT BEHAVIOUR:
For 1-9 digits, we act exactly like we do with CSS. Unfortunately, our CSS
acts wrongly (:-/). For 10+ digits, we just display black.

See test case:
   http://www.bath.ac.uk/%7Epy8ieh/internet/projects/mozilla/bgcolor.html

The CSS problem is covered by bug 8780.

[removing 4.xP marker as this bug is also present in legacy browsers]

Comment 5

[Peter Linss]

Swiping this bug.

I had this all working nicely once upon a time (1-9 digits were taken only in
quirk mode for compatibility).

Apparently someone broke this. Grrrr.

Comment 6

[Crysgem]

*As a dwarf stumbling among the ruins of giants*
I have no words for the Great Issues discussed herein/tofore... but the initial
bug is indeed crushed, as proved by the Apprunner of 1999070808. Hail to your
algorithms!