WebKit Bugzilla
New
Browse
Search+
Log In
×
Sign in with GitHub
or
Remember my login
Create Account
·
Forgot Password
Forgotten password account recovery
RESOLVED FIXED
40880
SVG properties fill and stroke do not accept system colors
https://bugs.webkit.org/show_bug.cgi?id=40880
Summary
SVG properties fill and stroke do not accept system colors
Rob Buis
Reported
2010-06-19 10:35:35 PDT
The problem is clearly demonstrated by the given url
Attachments
Patch
(7.72 KB, patch)
2010-06-19 10:41 PDT
,
Rob Buis
no flags
Details
Formatted Diff
Diff
Test all css system colors
(31.43 KB, patch)
2010-06-24 00:26 PDT
,
Rob Buis
no flags
Details
Formatted Diff
Diff
Incorporate Dirks suggestions into test
(23.14 KB, patch)
2010-06-24 13:26 PDT
,
Rob Buis
krit
: review+
Details
Formatted Diff
Diff
Show Obsolete
(2)
View All
Add attachment
proposed patch, testcase, etc.
Rob Buis
Comment 1
2010-06-19 10:41:21 PDT
Created
attachment 59188
[details]
Patch
Rob Buis
Comment 2
2010-06-24 00:26:26 PDT
Created
attachment 59620
[details]
Test all css system colors
Dirk Schulze
Comment 3
2010-06-24 00:53:50 PDT
I would put all system colors into an array and create the elements on the fly by with it's property in a loop, that would make the test much more readable. Something like: svgElement = document.getElementsByTagName("svg")[0]; systemColors = new Array("Window", "WindowFrame", ...) for(int i=0;i<systemColors.length;++i){ ... Every PASS message has the same result. Maybe you can add the current system Color name and the real RGB numbers. This makes it easier to analyse the test on regression that may occure later and should be easy to implement with the mentioned loop of above. What do you think?
Rob Buis
Comment 4
2010-06-24 01:03:20 PDT
Hello Dirk, (In reply to
comment #3
)
> I would put all system colors into an array and create the elements on the fly by with it's property in a loop, that would make the test much more readable. Something like: > > svgElement = document.getElementsByTagName("svg")[0]; > systemColors = new Array("Window", "WindowFrame", ...) > for(int i=0;i<systemColors.length;++i){ > ... > > > Every PASS message has the same result. Maybe you can add the current system Color name and the real RGB numbers. This makes it easier to analyse the test on regression that may occure later and should be easy to implement with the mentioned loop of above. > > What do you think?
You are right of course! I already felt when reediting the text that it was a bit "exhaustive" and verbose. I like the suggestions and will create an improved patch, after which we can hopefully wrap this one up :) Cheers, Rob. PS: even though I discovered "Menu" system color needed to be supported compared to the old patch(done with this patch), it doesnt solve "text on background with similar color" problem. For the moment I assume this is just because unfortunate system colors were supported, or maybe some not at all with a default to black?
Dirk Schulze
Comment 5
2010-06-24 01:17:39 PDT
(In reply to
comment #4
)
> You are right of course! I already felt when reediting the text that it was a bit "exhaustive" and verbose. I like the suggestions and will create an improved patch, after which we can hopefully wrap this one up :) > Cheers, > > Rob. > > PS: even though I discovered "Menu" system color needed to be supported compared to the old patch(done with this patch), it doesnt solve "text on background with similar color" problem. For the moment I assume this is just because unfortunate system colors were supported, or maybe some not at all with a default to black?
Hi Rob, Menu and others have a default color. Menu black, others can have different colors. This is allowed according to the Spec: "For systems that do not have a corresponding value, the specified value should be mapped to the nearest system value, or to a default color." At the long term even the default vlaues schould move to the platform code. But this is not our problem at the moment and should be solved in RenderTheme instead of SVG. PS: I was wrong, system colors are part of CSS2 and therefor have to be supported by SVG :-)
Rob Buis
Comment 6
2010-06-24 13:26:18 PDT
Created
attachment 59689
[details]
Incorporate Dirks suggestions into test
Dirk Schulze
Comment 7
2010-06-24 23:16:04 PDT
Comment on
attachment 59689
[details]
Incorporate Dirks suggestions into test Great Patch! r=me
WebKit Review Bot
Comment 8
2010-06-25 00:46:07 PDT
http://trac.webkit.org/changeset/61837
might have broken Qt Linux Release
Rob Buis
Comment 9
2010-06-28 00:40:20 PDT
Fixed by
r61837
.
Note
You need to
log in
before you can comment on or make changes to this bug.
Top of Page
Format For Printing
XML
Clone This Bug