W3C home > Mailing lists > Public > www-svg@w3.org > June 2010

Re: Test updates for SVG

From: Alex Danilo <alex@abbra.com>
Date: Thu, 03 Jun 2010 12:09:16 +1000
Message-Id: <GZ1F3L.9Q3B0OVK3UVM@abbra.com>
To: Patrick Dengler <patd@microsoft.com>
Cc: "www-svg@w3.org" <www-svg@w3.org>
Hi Patrick,

--Original Message--:
>In general, as I go through these tests, it seems that well over 50% of the items that have not yet been accepted are broken or simply not there.  This should be of huge concern for delivering a quality spec.  I've listed the issues I ran into while working on the additional set I said I would get on my flights back. 
>#-1 ) This is a key issue.  Why is it that struct-use-01-t.html shows non-italic text in the comparison image, and yet italic text in every browser except Opera, when it plainly says italic? 
><use y="260" xlink:href="#usedText" fill="#0F0" font-weight="bold" font-size="25" font-style="italic"/>

Which version of the test are you looking at (1.1F2 I presume)?

The published 1.1 suite has no reference to SVG fonts, and all viewers I tried
display italic correctly.

Looks like the 1.1F2 test has inherited the Tiny 1.2 version which
references the SVG font.

The 1.2 Tiny test references the external SVG font which has no italic face
so the viewers fall back to the normal glyphs in the SVG font. We need to
check CSS for the font selection here (I can't remember off the top of
my head what the rules are). A viewer could easily synthesize an oblique
version of the SVG font if needed for the 1.2 Tiny case.

So yes, you're right it should be italic - however that is dependent on
the CSS rules for font selection in that case.

>My understanding was that these images were being snapped from a vender neutral process.  Did I get that wrong?  

As far as I know they are - most likely Batik which is an Apache project.

>If not, I would like to have all of the .png's for tests that IE passes from IE.  Is that cool? I am sure I can get these quickly.

I think it's a bit presumptuous to do that until we ratify what the correct
CSS behaviour is.

>#0) Also, why do a lot of tests have a 
><font-face-uri xlink:href="../resources/SVGFreeSans.svg#ascii"/>
>When the have nothing to do with SVGFont testing.  I think we may really need to hold off on these tests and get them in shape.

1.1 defines SVG Fonts as part of the spec, so the tests use features from
across the board. They are not unit tests, and so you'll find many features
being tested in most tests.

Not ideal from an incremental development point of view, but anyway the 1.1
suite as published doesn't seem to suffer that - 1.2 Tiny test "improvements"
have been integrated into the 1.1 second edition suite by the looks of things.

The group needs to debate this. I won't waste my time debating the need/
lack of need for SVG fonts in this forum. IMO it takes less time to implement
them than reply to countless emails on the subject.

>I highly recommend that you allow me to consult my test team, and for others to consult their test teams to see if we can come up with a consolidated, above the belt, concerted effort to get these to be correct and balanced

Agreed. But take that with the grain of salt that the test suite is not
a conformance suite. It indicates implementability and there are multiple
implementations of all features in the current test suite.

It would be great to get a summary from your test team about what sucks
from their point of view and review it as a group. Best if it's one big
list of issues than piece-meal from an efficiency point of view.

>#3) styling-pres-01-t.svg seems wrong. Someone needs to review this.

It tests an error case for the fill property. It's been reviewed many times.

I'll try to look at the other test notes you made when bandwidth permits.


>#4) not sure what styling-pres-css-09-f is supposed to be doing; there is no text
>#5) struct-use-12-f needs an image
>#6) struct-use-08-b has no text, description or pass criteria. 
>#7) struct-use-06-b fails hilariously different in every browser
>#8) struct-svg-02-f.svg does not seem to be complete, nor work.
>#9) struct-image-12-b does not have any descriptive text, any has questions within the text of the test; and there is no related image.  
>#10) struct-image-11-b.html : operator script, expected results and actual test do not match
>#11) struct-dom-13 : doesn't pass in any browser, as far as I can tell, and has no PNG image.
>#12) shapes-rect-03-t : states    "The test has passed if there is no red visible" but the comparing image has red visible.
Received on Thursday, 3 June 2010 02:10:09 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 8 March 2017 09:47:21 UTC