Table of Contents

Record of Known Bugs

This page records known problems in the CSS 2.1 Conformance Test Suite. For problems with the harness, see Test Suite Build System.

Please do not add new issues to this page, to record a bug, use the Test Suite Management System. Search for the test case, enter your comment and set the status to “Needs Work”.

The issues listed on this page will be migrated to the new management system shortly.

Tests that are Wrong or have Parse Errors

38 Incorrect RC6 testcases

  1. [MSFT] [RC6] font-family-rule-002 Reasons why: http://lists.w3.org/Archives/Public/public-css-testsuite/2011Feb/0116.html It is not ok to have one letter “X”; there should be 2 small filled black squares vertically lined up. line-height should be specified in the testcase, preferably set to a value of 1.25 in the testcase.
  2. [MSFT] [RC6] font-family-rule-003 Reasons why: http://lists.w3.org/Archives/Public/public-css-testsuite/2011Feb/0117.html font-family: Ahem!; is not a valid declaration: CSS validation report. Therefore, an “X” must be rendered; an “X” being rendered is not optional or a 'may'. Pass conditions should be updated accordingly. Additionally, the testcase misses the invalid flag.
  3. [MSFT] [RC6] background-position-applies-to-004 This background-position-applies-to-004 testcase is supposed to test background-position applied to a table-row. Now, if the tested table-row has only 1 single cell, then how can such testcase be different from testing background-position being applied to a table-cell? More reasons: http://lists.w3.org/Archives/Public/public-css-testsuite/2011Sep/0040.html

Tests relying on non-normative behavior

  • font-matching-rule-008 - fixed for RC4
  • Various [min-|max-|]height-applies-to- tests listed here : height, max-height, min-height supposed to be undefined for table row groups; max-height and min-height should not apply to table-row elements and table-cell elements - Some fixed in RC4 remaining fixes in RC5
  • content-counter-010 - Fixed for RC4
  • content-counters-010 - Fixed for RC4
  • [MSFT] [RC6] font-005 , [MSFT] [RC6] font-019 and [MSFT] [RC6] font-021: all 3 testcases assume that synthetic small-caps “x” must be taller/larger than normal lowercase “x” when the spec makes no statement of this sort. Also, using x/X is not best; better is e/E because glyphs have different shapes. More info
  • [nightly-unstable] c21-pseud-link-002 The tested link has <a href=“c21-pseud-link-002.xht”>
  • [MSFT] [RC6] first-letter-selector-015 Konqueror 4.6.4 fails this test only because a missing image icon is shown and the text “Missing” is shown (due to default image placeholder dimensions insufficient, too narrow, for the alt text) but not “Missing image”. Just because of that, I think the testcase should be reviewed.

Miscellaneous incorrect tests

0 RC4 testcase with incorrect pass condition / dependence on font metrics

92 RC6 testcases not actually testing what they are supposed to test or that are too easy to pass

  1. [MSFT] [RC6] containing-block-017 Reasons why: http://lists.w3.org/Archives/Public/public-css-testsuite/2011Jan/0063.html What do do? title and text assert should be adjusted to pinpoint section 10.1, bullet 4, sub-bullet 1 and not sub-bullet 2 (since ancestor is an inline box, not a block).
  2. [MSFT] http://lists.w3.org/Archives/Public/public-css-testsuite/2011Jun/0007.html: over 60 [RC6] [max|min]-[height|width] testcases can not fail.
  3. [MSFT] http://lists.w3.org/Archives/Public/public-css-testsuite/2011Jun/0007.html: about 30 [RC6] [max|min]-[height|width] testcases have an one-single-pixel-between-pass-and-fail-condition which is not suitable for a test suite and human testers.

11 invalid RC6 testcases because they rely on rounding issues

Tests that need Usability improvements

Ugly, not self-documenting:

background-alpha-001
background-alpha-002
background-alpha-003
background-alpha-004
background-alpha-005

[MSFT] [RC6] content-158 What to do? set display: inline-block to containing block and then a padding of 1em should be added to make viewing easier, faster.

Confusing/hard to judge

  • background-attachment-004 (also has external references to image files)
  • inlines-004-inlines-006: unclear what “aligned” means. Fixed for RC4
  • text-transform-bicameral-009, text-transform-bicameral-010: should attempt to pick a font that has the glyphs - Uses the browser's default, which should be falling back to whatever it can find.
  • [MSFT] [RC6] background-130 - still not fixed in RC6: “on top of” should be replaced with “above”; “stripe” seems preferable to “short box”
  • outline-003 - fixed for RC4
  • text-align-applies-to-001 - fixed for RC4
  • [Hixie] [RC6] inline-block-002 since “Try resizing the window so that inline-blocks wrap in one or two columns.” can never be accomplished. It should be saying instead/rather something like “Try resizing the window so that inline-blocks wrap internally into 2 lines of text.”
  • [Hixie] [RC6] inline-block-004 since “Try resizing the window so that inline-blocks wrap in one or two columns.” can never be accomplished. It should be saying instead/rather something like “Try resizing the window so that inline-blocks wrap internally into 2 lines of text.”
  • [Hixie] [RC6] inline-block-005 since “Try resizing the window so that inline-blocks wrap in one or two columns.” can never be accomplished. It should be saying instead/rather something like “Try resizing the window so that inline-blocks wrap internally into 2 lines of text.”
  • [Hixie] [RC6] first-line-pseudo-017 says “This line should be green.” but it is its background color. Proposed correction: “This line should have a green background.”
  • [Hixie] [RC6] first-line-pseudo-018 says “This line should be green.” but it is its text. Proposed correction: “This text should be green.”
  • [Hixie] [RC6] hover-selector-002 says “When you hover over this text (…)” : the this word normally is self-pointing, self-referencing… but not in this test. Such formulation is confusing.

9 RC6 testcases with too technical pass conditions

  1. [Moz] [RC6] table-percent-width-001 My neighbour would ask “What's a 100px by 100px green square?”
  2. [Hixie] [RC6] background-alpha-001 My neighbour would ask “What's a 'multi-bit alpha channel'?” Pass conditions are not so clear and text is too technical.
  3. [Hixie] [RC6] background-alpha-002 My neighbour would ask “What's a 'multi-bit alpha channel'?” Pass conditions are not so clear and text is too technical.
  4. [Hixie] [RC6] background-alpha-003 My neighbour would ask “What's a square div? How tall/big is/should be the 'size of two tiles'?” Pass conditions are not clear or easy for normal people and text is too technical.
  5. [MSFT] [RC6] cursor-applies-to-005 My neighbour would ask “What's a default cursor? How does it look like?” I propose to add to the expected results: “The default cursor often looks like an arrow.”
  6. [MSFT] [RC6] cursor-applies-to-006 My neighbour would ask “What's a default cursor? How does it look like?” I propose to add to the expected results: “The default cursor often looks like an arrow.”
  7. [Hixie] [RC6] first-line-pseudo-011 The pass conditions text should be only: “The next three boxes should look identical.” The sentence ”(If the second and third look different, then margin collapsing through empty elements is broken. Go to these tests for more details.)” should be removed.
  8. [Hixie] [RC6] inline-replaced-width-014 The pass conditions text should be only: “There should be 2 identical filled green rectangles and no red.”
  9. [Hixie] [RC6] inline-replaced-width-015 The pass conditions text should be only: “There should be 2 identical filled bright green rectangles and no red.”
  1. [Hixie] c527-font-000 : “There should be a small gap between each line.” Reason: how small is “a small gap”? How can a normal tester (your mother, your bus driver, your neighbour) realistically establish that this gap here is small while that gap over there is not small? How exactly can this test fail?
  2. [Hixie] c527-font-002 : “9px gap between lines” Reason: how can a normal tester (your mother, your bus driver, your neighbour) realistically establish that this gap here is 9px while that gap over there is not 9px? The testcase does not provide a ruler, some kind of green overlapping red mechanism, etc. Where is the start line and the end line?
  3. [Hixie] c527-font-003 : “small and cursive, with double line spacing” Reason: it is not clear how such test can be evaluated as pass or fail, even by a machine.
  4. [Hixie] c527-font-004
  5. [Hixie] c527-font-005
  6. [Hixie] c527-font-006
  7. [Hixie] c527-font-007
  8. [Hixie] c527-font-008
  9. [Hixie] c527-font-009
  10. [Hixie] c527-font-10 : Reasons: All these testcases rely on the tester being able to measure a distance of a few pixels between lines, or the font-size, without a comparing ruler, without a reliable working measurement system or some kind of overlapping color mechanism.

More explanations on why these c527-font-* testcases need to be reworked or redesigned to be easier for normal tester-persons to evaluate: http://lists.w3.org/Archives/Public/public-css-testsuite/2010Jan/0043.html , http://lists.w3.org/Archives/Public/public-css-testsuite/2010Aug/0000.html , http://lists.w3.org/Archives/Public/public-css-testsuite/2010Oct/0061.html

Should be ref tests

c5526c-display-000

3 RC6 testcases creating unneedlessly and unexpectedly an horizontal scrollbar

  1. [Hixie] [RC6] vertical-align-121 Removing <p class=“ahemprereq”> whole paragraph is likely going to remove the unneeded and unexpected horizontal scrollbar at the same time.
  2. [Hixie] [RC6] white-space-collapsing-001 Removing <div class=“prereq”> whole paragraph is likely going to remove the unneeded and unexpected horizontal scrollbar at the same time.
  3. [Hixie] [RC6] white-space-collapsing-002 Removing <div class=“prereq”> whole paragraph is likely going to remove the unneeded and unexpected horizontal scrollbar at the same time.

0 RC6 testcase without pass conditions

  1. The following testcases: [Hixie] [RC6] margin-collapse-101 , [Hixie] [RC6] margin-collapse-107 , [Hixie] [RC6] margin-collapse-108 , [Hixie] [RC6] margin-collapse-109 , [Hixie] [RC6] margin-collapse-110 , [Hixie] [RC6] margin-collapse-111 , [Hixie] [RC6] margin-collapse-112 , [Hixie] [RC6] margin-collapse-113 , etc.. all have a sentence pointing to testcase [RC6] margin-collapse-106 but the link is broken.

0 RC6 testcase that rely on unnecessary assumptions

Tests that have HTML or CSS Validation errors

9 testcases that have HTML Validation errors

  1. [Moz] before-after-dynamic-attr-001 The html element can not have a class: class=“reftest-wait”; body element can not be empty.
  2. [Moz] before-after-dynamic-restyle-001 The html element can not have a class: class=“reftest-wait”; body element can not be empty.
  3. [Hixie] [RC6] inlines-014 line8: <met> should be <meta>
  4. [MSFT] [RC6] viewport-004 The frame element must not have a closing, end tag: its content model is empty. Validation report of viewport-004.htm

6 RC6 testcases that have CSS Validation errors

Tests that need Metadata improvements

3 testcases with wrong, incorrect meta assert text

0 RC6 testcase possibly missing http flag

1 RC6 testcase missing the paged flag

7 RC6 testcases missing the ahem flag

73 RC6 testcases missing the invalid flag

  1. [Hixie] [RC6] c526-font-sz-003 font-size: -0.5in; will be parsed and reported as an illegal value.
  2. [Hixie] [RC6] c548-ln-ht-002 line-height: -1em; will be parsed and reported as an illegal value.
  3. [MSFT] [RC6] max-height-001: because “Negative values for 'min-height' and 'max-height' are illegal.”
  4. [RC6] max-width-001: because “Negative values for 'min-width' and 'max-width' are illegal.”
  5. [MSFT] [RC6] min-height-001: because “Negative values for 'min-height' and 'max-height' are illegal.”
  6. [MSFT] [RC6] min-width-001: because “Negative values for 'min-width' and 'max-width' are illegal.”

1 RC6 testcase missing the interact flag

  1. [Hixie] [RC6] floats-110 since resizing the viewport is part of the test pass/fail condition

0 RC6 testcase missing the dom flag

0 RC6 testcase with some coding issues

3 RC4 testcases using, relying on CSS3 grammar or CSS3 properties

  1. page-break-after-010 relies on CSS3 at-keywords for paged media - CSS3 bits not part of tests, so ok
  2. page-break-before-003 relies on CSS3 at-keywords for paged media - CSS3 bits not part of tests, so ok
  3. page-break-before-005 relies on CSS3 at-keywords for paged media - CSS3 bits not part of tests, so ok

0 RC6 testcase with wrong meta flag

0 RC6 test with missing title

2 RC4 testcases which could render a missing image icon as acceptable handling of url(missing-image.png)

CSS 2.1 spec says: “User agents may vary in how they handle invalid URIs or URIs that designate unavailable or inapplicable resources.”

coming from CSS 2.1, section 4.3.4 URLs and URIs

and

“The value is a URI that designates an external resource (such as an image). If the user agent cannot display the resource it must either leave it out as if it were not specified or display some indication that the resource cannot be displayed.”

coming from CSS 2.1, section 12.2 The 'content' property

So, a missing image icon is an acceptable, reasonable indication that the resource cannot be displayed.

  1. [Moz] before-after-images-001 ; “before-after-images-001 is invalid, as it assumes a particular

treatment of broken images in 'content'.” coming from http://lists.w3.org/Archives/Public/public-css-testsuite/2010Dec/0175.html

RC5 test cases with font rounding issues

  1. orphans-00*
  2. widows-00*
  3. Also any “paged” test cases

General Issues

 
test/css2.1/issues.txt · Last modified: 2011/11/18 11:08 by gtalbot
Recent changes RSS feed Valid XHTML 1.0 Valid CSS Driven by DokuWiki