[Webkit-unassigned] [Bug 6018] WebKit+SVG should pass all of Hixie's test (and adopt them)

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Wed Dec 31 05:03:18 PST 2008


helder.magalhaes at gmail.com changed:

           What    |Removed                     |Added
                 CC|                            |helder.magalhaes at gmail.com

------- Comment #2 from helder.magalhaes at gmail.com  2008-12-31 05:03 PDT -------
(In reply to comment #1)
> Looking again we seem to fail quite a number of these.
Revisiting this issue with revision 39474, using a nightly [1] on Windows XP

> Failures:
> http://hixie.ch/tests/adhoc/svg/cascade/001.xml
Now OK.

> http://hixie.ch/tests/adhoc/svg/cascade/002.xml
Now OK.

> http://hixie.ch/tests/adhoc/svg/data-types/002.xhtml
I believe there is an error in the test itself. To force SVG content to resize
to the SVG canvas, one needs to set "preserveAspectRatio" property to "none"
(the lacuna value is "xMidYMid"). By making the following modification, the
test passes (same in Firefox 3.2a):

-  <svg height="40px" width="10em" viewBox="0 0 100 100"
+  <svg height="40px" width="10em" viewBox="0 0 100 100"
preserveAspectRatio="none" xmlns="http://www.w3.org/2000/svg">

> http://hixie.ch/tests/adhoc/svg/dynamic/005.xml (opera also fails, firefox
> passes)
Now OK.

> http://hixie.ch/tests/adhoc/svg/error/002.xml
Not sure about this one. The test states that:

  I'm assuming, here, that the UA doesn't support XLink itself
  in an SVG context, but that it merely supports XLink's attributes in
  an SVG context as defined by SVG, and that therefore the
  xlink:type="simple" xlink:show="embed" xlink:actuate="onLoad"
  attributes are all simply ignored.

Thing is I'm not sure if it should be ignored, at least not according to a
recent version of the specification [3]...
Another thing that confuses me in this test is that linking to an external
style sheet should be done using the "xml-stylesheet" element...
Finally, the text seems to contradict the test itself. If the external style
information was to be ignored, than one should read "This line should be red."
or the test needed to be modified in accordance.
In my personal opinion, I believe this is a "pass".

> http://hixie.ch/tests/adhoc/svg/error/003.xml
Now OK. I'd add that the default stroke length (1) is also pretty thin to be
seen, specially in TFT monitors using anti-aliasing. I'd propose increasing the
stroke width to 3 or 5 in order to make the "rectangle with a blue border" more

- <rect x="0" y="0" width="300" height="200" stroke="blue" fill="url(#test)"/>
+ <rect x="0" y="0" width="300" height="200" stroke="blue" stroke-width="3"

I also find the explanatory text a bit confusing...

  on the next line the given paint server is not valid, therefore it is
  unsupported and XXX raised as last call issue XXX

> http://hixie.ch/tests/adhoc/svg/error/006.xml
Not sure about this one. Nothing is displayed and, FWIW, Firefox 3.2a also
displays nothing...

> http://hixie.ch/tests/adhoc/svg/error/012.xml
I believe there's an error in this test. According to document error
processing, nothing is stated regarding partial document rendering: Webkit does
display a "highly perceivable indication of error", although it displays the
rendering up to the erroneous XML portion, which is a red rectangle and seems
to point towards failure. The broken element as red sounds reasonable though.
I'd say this is a "pass" and would suggest the following modification to the

- <rect x="10" y="10" width="200" height="200" fill="red"/>
+ <rect x="10" y="10" width="200" height="200" fill="green"/>

> http://hixie.ch/tests/adhoc/svg/error/013.xml
Partially failing. No "highly perceivable indication of error" is presented,
although the document in failure isn't rendered (as expected).

> http://hixie.ch/tests/adhoc/svg/error/017.xml
Still fails. A circular reference (from element with identifier "d" to "a"
which refers "b", "c" and "d" again) is an error [5].

> http://hixie.ch/tests/adhoc/svg/links/001.xml (we don't support link tabbing
> yet)
Still fails. No tab support exists, as stated. Opera 10a passes, but only if
the iFrame is focused first (there's no tab navigation between the main and the
iFrame document).

> http://hixie.ch/tests/adhoc/svg/links/002.xml
I believe there's an error in this test. Linking to an external style sheet
should be done using the "xml-stylesheet" element... By making the following
modification, the test passes (same in Firefox 3.2a):

- <svg xmlns="http://www.w3.org/2000/svg"
-  <style xlink:href="002.css"/>
+ <?xml-stylesheet type="text/css" href="002.css"?>
+ <svg xmlns="http://www.w3.org/2000/svg"
+  <text y="30" x="10" fill="red" font-size="20">This line should be

> http://hixie.ch/tests/adhoc/svg/links/003.xml
Not sure about this one. Text states that:

  Spec says that <a> can only link to animation elements,
  otherwise it is invalid; that known attributes with
  unsupported values are treated as if they hadn't been
  specified when rendering; that if the attribute is not
  specified, the effect is as if an empty value ("") was
  specified; and that an empty attribute value (xlink:href="")
  means that no link traversal occurs upon activation of the 'a'
  element. So. Nothing should happen if you click it

But, at least according to SVG 1.2, there's nothing wrong about it [6]. I'd
only mark this as "partially failed" as the target element's bounding box isn't
centered as expected [7] after activating the link.

Few more test failures or commented tests (ones not listed here and previously
are passing!):
 * http://hixie.ch/tests/adhoc/svg/links/pass.txt - probably this one shoudn't
appear in the list...
 * http://hixie.ch/tests/adhoc/svg/markers/001.xml - invalid XML in the text
case: apparently due to an "INCOMPLETE" status of the test. ;-)
 * http://hixie.ch/tests/adhoc/svg/mixed/001-demo.xml - apparently failed:
missing content ("TEST a" to "TEST d"), although selecting all and pasting into
a text editor shows that text is there (which may give a clue on what's
happening). :-)
 * http://hixie.ch/tests/adhoc/svg/mixed/002-demo.xml - same as
 * http://hixie.ch/tests/adhoc/svg/perf/007.xml - using the javascript array
notation is not SVG conformant and apparently fails in Firefox 3.2a (and
probably in ASV and Renesis also) [8]. By making the following modification,
the test gains interoperability (with Firefox 3.2a only, ASV and Renesis would
need a string as "setTimeout" argument...):

- var svg = document.getElementsByTagNameNS('http://www.w3.org/2000/svg',
- var t = document.getElementsByTagNameNS('http://www.w3.org/2000/svg',
+ var svg = document.getElementsByTagNameNS('http://www.w3.org/2000/svg',
+ var t = document.getElementsByTagNameNS('http://www.w3.org/2000/svg',

 * http://hixie.ch/tests/adhoc/svg/processing-model/001-demo-support.xml - I'm
not quite sure what the result should be...
 * http://hixie.ch/tests/adhoc/svg/processing-model/001-demo.xml - there seems
to be an issue with the test: file "001-demo-redirect.xml" doesn't exist so,
currently, a black fill seems to be a "pass"...
 * http://hixie.ch/tests/adhoc/svg/processing-model/002-demo-support.xml - I'm
not quite sure what the result should be...
 * http://hixie.ch/tests/adhoc/svg/processing-model/002-demo.xml - same as
 * http://hixie.ch/tests/adhoc/svg/processing-model/003.dtd - same as
 * http://hixie.ch/tests/adhoc/svg/rendering-model/004.xhtml - no zoom support
to perform test...
 * http://hixie.ch/tests/adhoc/svg/text/001.xml - partially failed: font is
used in rendering (after local install) but the alignment is not perfect -
glyph is positioned slightly above the expected location (Firefox 3.2a passes).
 * http://hixie.ch/tests/adhoc/svg/use/002.xml - there seems to be an issue
with the test: file "002-test.xml" is not there so, currently, a the red
rectangle seems to be a "pass"...

As stated, all items which are not here are considered OK, so this is quite a
good progress! :-)

I'd ask Ian Hixie to review the tests with potential issues and post some
follow up whenever possible. Thanks! ;-)

Happy new year,

 Helder Magalhães

[1] http://nightly.webkit.org/
[2] http://www.w3.org/TR/SVGTiny12/coords.html#PreserveAspectRatioAttribute
[3] http://www.w3.org/TR/SVGTiny12/linking.html
[4] http://www.w3.org/TR/SVGTiny12/implnote.html#ErrorProcessing
[5] http://www.w3.org/TR/SVGTiny12/linking.html#circular-iri
[6] http://www.w3.org/TR/SVGTiny12/linking.html#LinksIntoSVG
[7] http://www.w3.org/TR/SVGTiny12/linking.html#FragmentIdentifierTraversal

Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

More information about the webkit-unassigned mailing list