160053 - Bugs in CSS2 features

Status: RESOLVED DUPLICATE of bug 153699

(Core :: CSS Parsing and Computation, defect)

Description

[David]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.1b) Gecko/20020721
BuildID:    2002072104

I appologize if CSS2 is known to have problems - I searched but cannot find
anything substantial about it. I understand that multiple bug reports should be
created, but there are so many problems that this would take a very, very long
time. I will attempt to list all of the bugs. These CSS2 bugs have been found
using <http://www.richinstyle.com/test/>, as was linked from
<http://www.mozilla.org/quality/help/index-to-tests.html>. For long problems
that I don't understand, I just quote the page. Note that i'm calling it "buggy"
simply because that is the terminologoy used on the test page. :-)

URI Resolution:
    Absolute URI support: Buggy

HTTP link header:
    URI MUST come first since: Link = "Link" ":" #("<" URI ">" *( ";" 
      link-param )
    This has ' as quotes - invalid - the relationship is relationship 
      = sgml-name | ( <"> sgml-name *( SP sgml- name) <"> )

Media case:
    media="Screen" -buggy
    media="AlL" -buggy

@import with media:
    @import url(importtruncate.css) screen_attribute; - Media are not 
      truncated on @import -buggy

STYLE:
    Section 6.2 of the HTML 4 spec states the first occurrence of the 
      character sequence "</" (end-tag open delimiter) is treated as 
      terminating the end of the element's content.. This document has 
      a STYLE block that includes that, so this shouldn't be red. 
        (It's red)
    This should not be red, since src is not a valid attribute for 
      <style> (to specify an external style sheet)
        (It's red)

**
[ FROM HERE ON I'LL JUST LIST FAILED TESTS, RATHER THAN SAYING "BUGGY" OR "IT'S
RED" FOR EVERY ONE ]
**

Style attribute:
    This tests that the style attribute has specificity 100, rathern 
      than infinite specificity.
    style="{color: red}"

ID validity:
    IDs may not contain character references; therefore this should not be red.

Class validity:
    -hyphen

Case sensitivity:
    Class and ID fail in three different tests.

Link pseudo-classes:
    This should not be red (:link:link{color: red})

Dynamic pseudo-classes 2:
    Anonymous content - not underlined when hovered over

Selector combinations:
    This has .class1:link{color: #f0f(purple)}

:first-letter
    First test fails (vas space after the first letter); Also:
    This first letter should have left and right margins (only (other 
      margins indicate a margin bug, not a selectors bug))
    This first letter should be bordered in red.
    However, this element should be red, because the selector was 
      grammatically correct, and so therefore there is no reason to 
      ignore the ruleset.

:first-line
    Looks REALLY messed up (colors different and stuff):
      This should be vertically aligned differently - 1st line raised 
      by 3 cm.

Lengths test:
    font-size: 50 - missing unit.

Inapplicability test:
    This text should not have a line break here: See? This sentence, 
      which is longer than most of them in this page, should not be 
      restricted wrt to width

Url() test:
    background: yellow url("wood.jpg")

3-digit rgb:
    Missing hash
    Too few digits and missing hash

6-digit rgb:
    Missing hash

RGB %:
    Missing %

RGB decimal:
    Integer values only

Inline element background-repeat test:
      This should have a background repeated horizontally. Since the background
goes behind font size and since the line box is much taller than that, it should
be repeated horizontally, line by line, with width equal to the width of the
text on each line and height equal to the height of each line. In addition,
there should be whitespace between each line

--------

There are a gazillion more of these test, and if so requested I will finish
them, but it's taken me a while to get this far. Note that IE fails on vastly
more tests than mozilla. :)

Reproducible: Always
Steps to Reproduce:
1.Goto http://www.richinstyle.com/test/
2.Run every last test (whew!)
3.Record the results

Actual Results:  Some tests failed. :-(

Expected Results:  Mozilla should have passed on every single CSS2 test. ;-)

Comment 1

[Boris Zbarsky [:bzbarsky]]

> I understand that multiple bug reports should be created

Yep.  ;)

> These CSS2 bugs have been found using <http://www.richinstyle.com/test/

These tests are notoriously buggy and broken.  In particular:

>    Absolute URI support: Buggy

The absolute URI in question gives a 404.  Broken test.

>   Media case:

Media should be case-insensitive

> STYLE:

These behaviors are in place purposefully, actually.  Further, the testcases are
triggering our "quirks" mode because they are not valid HTML (no doctype). 
Hence all sorts of backwards-compatible stuff happens. (see also bug 153753,
which is filed on this exact testcase).

> Style attribute:

It should have infinite specificity.  See the CSS2 errata.

> Class validity:

This is true for CSS1/2, but not for CSS3.

> Case sensitivity:

This is invalid.  We're doing it on purpose in quirks mode and the invalid HTML
on this page triggers that.

> Link pseudo-classes:

I believe this test is also invalid but do not have the reference off the top of
my head.

> Dynamic pseudo-classes 2:

This is a quirks-only issue done for compatibility with poorly authored content.

> Selector combinations:
>  This has .class1:link{color: #f0f(purple)}

No it does not.  Look at the source.  Nowhere is that present in the stylesheet.

>    font-size: 50 - missing unit.

Quirks parsing of css, yay.  Testing in standards mode would really help.  ;)

> 3-digit rgb:

Same thing.

> 6-digit rgb:

Same.

Further, most of the remaining issues are already filed (searching on bugs with
"richinstyle" in the URL shows you the list).  For example:

> @import with media:

bug 153755

> ID validity:

Bug 153777

>     First test fails (vas space after the first letter); Also:

Bug 153781

> RGB %:

bug 154755

> RGB decimal:

bug 154755 (patch fixes it)

Finally, bug 153699 really covers exactly what you are doing, so I'm marking
duplicate of that.  Please file one bug per test on the tests you described in
the original report that I did not comment as being invalid or already reported
(and have them link to the testcase; some of them I just could not find on
richinstyle).

And by all means send mail to the site maintainer and ask him to fix his tests.
 He has not updated them since 2000, in spite of the numerous errors and
repeated mails from me and others...

*** This bug has been marked as a duplicate of 153699 ***

Comment 2

[David]

*sigh* I'm afraid I wasted my time. :P

Am I supposed to mark this as closed?

Thanks.