Version: 4.5.1 (using KDE 4.5.1) OS: Linux Microsoft provides testcases for various browser compliance things (see http://samples.msdn.microsoft.com/ietestcenter/#html5). Some of the XHTML5 tests fails: #2 Body element with link attribute (text does not get color of visited link after clicking) #4 Colgroup element with valign attribute set to bottom #5 Colgroup element with valign attribute set to top #6 Style inside comments #7 Image element with border in % (50% testcase fails) #8 Object element with border in % #9 Obect element with border in pixels Reproducible: Always OS: Linux (i686) release 2.6.34.4-0.1-desktop Compiler: gcc
Created attachment 54035 [details] testcase i wrote testcases
> Some of the XHTML5 > tests fails: > > #2 Body element with link attribute (text does not get color of visited link > after clicking) http://samples.msdn.microsoft.com/ietestcenter/html5/xhtml/body_link.xhtml There is no initial, default (user agent style sheet) CSS rule for :visited or a:visited in Konqueror, that is after examining html4.css . So, this may not be a Konqueror bug. The testcase assumes and presumes that the user agent style sheet has and uses an initial, default (user agent style sheet) CSS rule for :visited or a:visited. The result of Microsoft test here does not prove a flaw, a failure or an incorrect implementation as far as rendering visited links in a color. > #4 Colgroup element with valign attribute set to bottom > #5 Colgroup element with valign attribute set to top http://samples.msdn.microsoft.com/ietestcenter/html5/xhtml/colgroup_valign_bottom.xhtml http://samples.msdn.microsoft.com/ietestcenter/html5/xhtml/colgroup_valign_top.xhtml valign is a non-CSS presentational hint and a deprecated attribute. "The UA *_may_* choose to honor presentational attributes in an HTML source document." 6.4.4 Precedence of non-CSS presentational hints http://www.w3.org/TR/CSS21/cascade.html#preshint The thing is valign would normally be mapped to CSS vertical-align property. But, vertical-align does not apply to colgroup elements; it should apply only to table cell elements and inline-level elements. http://www.w3.org/TR/CSS21/visudet.html#propdef-vertical-align So, XHTML5 or not, I'd be surprised if this test pass in HTML 4.01 transitional anyway. Even Microsoft CSS 2.1 test suite contribution confirms this in http://test.csswg.org/suites/css2.1/20110323/html4/vertical-align-applies-to-005.htm So, there is a contradiction here and there is a leniency provided in section 6.4.4 of CSS 2.1. FWIW, Firefox 5.0 fails those 2 Colgroup element with valign attribute tests... and probably would fail those 2 same tests written in HTML 4.01 transitional. > #6 Style inside comments > #7 Image element with border in % (50% testcase fails) http://samples.msdn.microsoft.com/ietestcenter/html5/xhtml/img_border_percent.xhtml border expressed in percentage in CSS code is invalid: http://test.csswg.org/suites/css2.1/latest/html4/border-width-013.htm border attribute specified with a percentage value is extremely rare, is very likely to be inconsistently rendered in mainstream browsers, border attribute on image is deprecated. For all reasons given before (leniency, precedence of non-CSS hints, etc), this barely can be considered a flaw, even a minor flaw, in the border implementation. > #8 Object element with border in % http://samples.msdn.microsoft.com/ietestcenter/html5/xhtml/object_border_perc.xhtml Same reasons. > #9 Obect element with border in pixels This one has merits: border in pixels on HTML object element should be rendered ... but I can not silenced the fact that the testcase uses this code: <div valing="center"> .. which is invalid attribute specification. regards, Gérard
Regarding border expressed in percentage (%) for <img> and <object> elements, it is not only deprecated but it also assumes pixel unit only in HTML 4 spec! " border = pixels [CN] Deprecated. This attribute specifies the width of an IMG or OBJECT border, in pixels. The default value for this attribute depends on the user agent. " http://www.w3.org/TR/html4/struct/objects.html#adef-border-IMG So, I'm quite surprised that it would be considered a valid test for XHTML5. ----------- The verb modifier "may" has a precise meaning in the spec (HTML4: http://www.w3.org/TR/html4/conform.html and CSS 2.1: http://www.w3.org/TR/CSS21/conform.html#defs ). " 5. MAY This word, or the adjective "OPTIONAL", mean that an item is truly optional. (...) " http://www.ietf.org/rfc/rfc2119.txt Basically, "may" means if you support it, fine, ok; basically it means "you do not have to but it's okay if you support it". But if you do not support a "may" sentence, do not comply a "may" sentence, the user agent can still achieve a complete, perfect compliance with the spec nonetheless. ------------- As far as HTML5 specification goes so far: - border on <img> is allowed but discouraged in favor of CSS http://dev.w3.org/html5/html4-differences/#changed-attributes - border on <object> element no longer exists http://dev.w3.org/html5/html4-differences/#absent-attributes So even #9 http://samples.msdn.microsoft.com/ietestcenter/html5/xhtml/object_border_pixel.xhtml is questionable, debattable as a valid test for XHTML5. - There is no word none whatsoever with regards to rendering and implementing percentage value on border attribute for both of these elements. HTML5 differences from HTML4 W3C Working Draft 24 May 2011 http://dev.w3.org/html5/html4-differences/ Gérard
<img> element does not have a border attribute in HTML5: http://www.w3.org/TR/html5/embedded-content-1.html#the-img-element <object> element does not have a border attribute in HTML5: http://www.w3.org/TR/html5/the-iframe-element.html#the-object-element Gérard
No valign attribute for <colgroup> elements in HTML5: http://www.w3.org/TR/html5/tabular-data.html#the-colgroup-element My conclusion is that those 10 tests are, at least in part, questionable, debattable; they may serve more the purpose of proving that IE9 and IE10 Platform Preview (1 and later) are better than other browsers than to prove a real and useful compliance with an emerging and in-developement W3C web standard like HTML5. Overall, public test suites e.g. HTML5 test suite ( http://w3c-test.org/html/tests/harness/harness.htm http://w3c-test.org/html/tests/harness/approvedtests.txt ) are much more reliable, preferable, trustworthy, useful, relevant, helpful than a small set of very carefully selected testcases. regards, Gérard
I just realized today that the tests you were referring to (which have been submitted by Microsoft) have also been approved by W3C! http://w3c-test.org/html/tests/approved/xhtml5/body_link.xhtml http://w3c-test.org/html/tests/approved/xhtml5/body_text_00ffff.xhtml http://w3c-test.org/html/tests/approved/xhtml5/html_style_in_comment.xhtml http://w3c-test.org/html/tests/approved/xhtml5/img_border_percent.xhtml http://w3c-test.org/html/tests/approved/xhtml5/object_border_perc.xhtml http://w3c-test.org/html/tests/approved/xhtml5/object_border_pixel.xhtml I still think this is wrong... in particular the border percentage ones. The 2 colgroup testcases have not been approved; at least they are not in the http://w3c-test.org/html/tests/harness/approvedtests.txt regards, Gérard
> > #2 Body element with link attribute (text does not get color of visited link > > after clicking) > > http://samples.msdn.microsoft.com/ietestcenter/html5/xhtml/body_link.xhtml > > There is no initial, default (user agent style sheet) CSS rule for :visited or > a:visited in Konqueror, that is after examining html4.css . So, this may not be > a Konqueror bug. There is a Konqueror bug wrt :visited links: Bug 24820 : visited links don't change colour until after they are hovered over with the mouse. Gérard
(In reply to comment #0) > Microsoft provides testcases for various browser compliance things (see > http://samples.msdn.microsoft.com/ietestcenter/#html5). Some of the XHTML5 > tests fails: > > #4 Colgroup element with valign attribute set to bottom > #5 Colgroup element with valign attribute set to top Rolf Eike Beer, I hope you are still reading this bug report. You see, 3 weeks before you filed this bug, those 2 colgroup with valign testcases were rejected (for the reasons I invoked) by W3C HTML group (public-html-testsuite mailing list): " colgroup_valign_bottom.xhtml colgroup_valign_top.xhtml Rejected. There are no requirements in the specification about colgroup@valign. " http://lists.w3.org/Archives/Public/public-html-testsuite/2010Aug/0011.html August 24th 2010 but Microsoft never removed those testcases from their IE testcenter website! Gérard Talbot
> Rolf Eike Beer, > > I hope you are still reading this bug report. I do. > You see, 3 weeks before you filed this bug, those 2 colgroup with valign > testcases were rejected (for the reasons I invoked) by W3C HTML group > (public-html-testsuite mailing list): Fine with me if we do not need to care here.
Rolf Eike Beer, I went over the Public HTML testsuite mailing list and more testcases from that testsuite and there are problems with testcases. One recent email (dated august 23rd 2011) caught my attention: http://lists.w3.org/Archives/Public/public-html-testsuite/2011Aug/0015.html in which one of the testcase approver says " I've asked a fellow Mozilla developer to review the HTML forms tests Microsoft recently submitted. [1] Their comments (which I agree with) are below. (...) While I appreciate the work Microsoft has spent on these tests, I would not approve any of them. " --------- Here's my email sent today to such mailing list: img_border_percent , object_border_perc and object_border_pixel testcases http://lists.w3.org/Archives/Public/public-html-testsuite/2011Sep/0017.html --------- I have asked 3 times in IE blog since july 1st 2011 to have the 2 colgroup with valign testcases removed from IE test center. They have not complied so far. --------- Rolf Eike Beer, one thing is certain in my mind. Konqueror with KHTML rendering engine is lagging behind other browsers in HTML5 compliance and support, especially newer features, advanced stuff. Konqueror and its KHTML rendering engine lacks developers to fix problems, bugs, crashes, hang (serious, severe) bugs. regards, Gérard
Rolf Eike Beer, 1- What would really help here is if someone could run the HTML5 test harness (924 tests) and then report the results for Konqueror with KHTML rendering engine. 2- Also, the same should be done for other new W3C spec being developed and with available, accessible test suites or with available, accessible sets of testcases (canvas, media queries, CSS3 modules, DOM3 interface attributes, methods, ECMAScript 5.1 test suite, etc.) This is best, ideal: we then can all know/see how well Konqueror is doing and where exactly it fails. 3- Another thing which would help is to visit this KDE forum thread and to vote accordingly: Poll/survey: Your opinion about konqueror future What would you like to happen to Konqueror ? http://forum.kde.org/viewtopic.php?f=18&t=96876 Right now there is *_no developer_* dedicated (or even partially dedicated) to tackle and fix Konqueror's KHTML rendering engine bugs (spec violations), crashes, hang bugs, various types of problems and issues. All of the commits to KHTML rendering engine in the last 12 months clearly indicate this. ---------------- I got answers with regards to http://w3c-test.org/html/tests/approved/xhtml5/img_border_percent.xhtml http://w3c-test.org/html/tests/approved/xhtml5/object_border_perc.xhtml http://w3c-test.org/html/tests/approved/xhtml5/object_border_pixel.xhtml in http://lists.w3.org/Archives/Public/public-html-testsuite/2011Sep/0019.html [1] http://www.whatwg.org/html/#attributes-for-embedded-content-and-images " When an img element, object element, or input element with a type attribute in the Image Button state is contained within a hyperlink and has a border attribute whose value, when parsed using the _rules for parsing non-negative integers_ [2], is found to be a number greater than zero, the user agent is expected to use the parsed value for eight presentational hints: four setting the parsed value as a pixel length for the element's 'border-top-width', 'border-right-width', 'border-bottom-width', and 'border-left-width' properties, and four setting the element's 'border-top-style', 'border-right-style', 'border-bottom-style', and 'border-left-style' properties to the value 'solid'. " [2] http://www.whatwg.org/html/#rules-for-parsing-non-negative-integers So, it seems (I requested a confirmation of such reasoning) that <img border="50foo" ...> or <img border="50pt" ...> or <img border="50%" ...> or <object border="50bar" ...> or <object border="50rem" ...> or <object border="50%" ...> should all be rendered the same: with a 50px border on each side. I posted http://lists.w3.org/Archives/Public/public-html-testsuite/2011Oct/0000.html regards, Gérard
Dear user, KHTML (and KJS) was a long time more or less unmaintained and got removed in KF6. Please migrate to use a QWebEngine based HTML component. We will do no further fixes or improvements to the KF5 branches of these components beside important security fixes. For security issues, please see: https://kde.org/info/security/ Sorry that we did not fix this issue during the life-time of KHTML. Greetings Christoph Cullmann