November 2006 - webkit-dev - lists.webkit.org

Keyevents for modifier keys
by Fabian Jakobs 27 Nov '06

27 Nov '06

Hi, while porting the JavaScript Framework qooxdoo (http://qooxdoo.org) to WebKit, we've found some strange behaviour in the WebKit keyboard handler: None of the modifier keys (Shift, Alt, Meta, Control) do fire key events, while all other browsers do fire them. This makes things like changing the action of drag and drop from move to copy using the ALT key (like in the finder) hard to implement. Best Fabian

1 1

cast operators for LPCWSTR / UString
by Justin Haygood 26 Nov '06

26 Nov '06

Would anyone object to a Windows only cast operator for LPCWSTR ßà UString? Benefits include: 1. You can use a LPCWSTR where a UString is expected 2. You can use a UString where a LPCWSTR is expected 3. The community can slowly develop a superfast method of converting between the two 4. Something similar can be done for WebCore::String, so that Windows developers get the same benefit OS X developers get (cast operators to and from the native Unicode string format) Thanks, Justin Haygood

1 0

(JavaScriptCore impl) UString -> unsigned short array
by Justin Haygood 26 Nov '06

26 Nov '06

I'm currently working on a project to embed JSCore (chosen since I was fairly familiar with it, it was C++, and its blazing fast on Windows) into an app to replace Perl as the company I work for's language of choice for quick and dirty scripting. Since we have a massive library of Win32 code to use with it, I was wondering: Is there a quick way to convert a KJS::UString to an array of Unicode shorts (aka LPCWSTR in Win32 speak). I came up with the much hackish solution below that works, but seems to me to be really inefficient: // Convert from UString to LPCWSTR. // FIXME: Find a better solution if possible // TODO: Convert into global C style function if above FIXME becomes invalidated const UChar* data = ustring.data(); int length = ustring.size(); WCHAR* winstringp = new WCHAR[length+1]; ZeroMemory(winstring, (sizeof(WCHAR)*length+1)); for( int i = 0; i < length; i++) winstrinpg[i] = data[i].unicode(); winstringp[length] = '\0'; LPCWSTR winstring = winstringp; If someone can think of a better way, that would be totally awesome. Justin Haygood | Software Engineer | EyeWonder, Inc. Phone: 678-891-2031 Cell: 404-271-2273 (jhaygood86(a)tmomail.net) jhaygood(a)eyewonder.com EyeWonder Richer Media. Richer Results. www.eyewonder.com <http://mail.eyewonder.com/exchweb/bin/redir.asp?URL=http://www.eyewonder.co m/>

2 3

Important missing bits in SVG support
by Andreas Neumann 21 Nov '06

21 Nov '06

Hi all, I am an active SVG developer and am following Webkit's progress regarding SVG for a while. I think it is soon at a state where it can be used for many SVG projects - not yet complete and bug free, but already usable. I am also happy to see that one can convert SVGs to PDF using Webkit - great! Thanks for your great work on the SVG support! Some of my work can be seen at http://www.carto.net/papers/svg/samples/ - note that all examples that run in Opera/Batik/Firefox are standard compliant, while some of the older or more complex examples might only work in ASV (I am working to fix these as well). I would like to share a list of missing features with you, that I would need to get most of my examples working in Webkit: * dynamic creation of use elements does not yet work: see http://www.carto.net/papers/svg/manipulating_svg_with_dom_ecmascript/create… (try the dynamic creation for <use/> elements) * support for pointer-events - this is very important for many of my GUI elements as well as mapping projects * support for SVGLocatable - all of these methods and properties are very important for coordinate system calculations and for getting bounding boxes: .getBBox(), .getCTM(), .getScreenCTM() (very important), .getTransformToElement() * support for SVGPoint and the .matrixTransform() method I think if these missing bits above are implemented, many of my examples will work and I'll be able to write bug reports on other things. Currently, most of my examples aren't functional in Webkit due to the missing SVGLocatable and SVGPoint methods/properties. I would greatly appreciate if you could look into supporting these methods in the near future. Thanks a lot, Andreas -- ---------------------------------------------- Andreas Neumann Institute of Cartography ETH Zurich Wolfgang-Paulistrasse 15 CH-8093 Zurich, Switzerland Phone: ++41-44-633 3031, Fax: ++41-44-633 1153 e-mail: neumann(a)karto.baug.ethz.ch www: http://www.carto.net/neumann/ SVG.Open: http://www.svgopen.org/ Carto.net: http://www.carto.net/

3 2

Re: [webkit-dev] SVG display in embed elements chooses ASV instead ofWebkit
by Andreas Neumann 20 Nov '06

20 Nov '06

Hello, >> >> I personally would prefer that the upcoming webkit would use its own >> SVG engine instead the Adobe viewer. ASV, wasn't updated for ages >> (since 2001). > > If a user has explicitly installed a plugin we should probably assume > they want it specifically to do something that we don't support and > specail casing certain mimetypes to not trigger a plugin would seem > ugly.... ok - I think you convinced me. ASV will soon be history, anyway. I just de-installed it now, to better test Webkits SVG support. > Okay, list time. These are things we need to get going to be > comparable to ASV (at least to a reasonable degree) > * Text support -- this still isn't complete, we still lack a number > of styling attributes, and don't support vertical text > * Text path -- we don't support it at all > * Alternate glyphs in text -- eg. SVG characters, etc > * Animation -- we have most of the support in place now, but it > still needs to be completed > * Filters -- currently unstable, also the qquality of the output is > not all that great, and they can be tremendously inefficient In my personal opinion, supporting all the DOM methods is more important than say filter and animation support. But I would certainly welcome if the full SVG 1.1 specification would be supported ;-) Sometimes one can do workarounds by script for some missing features. As an example one can do animation by script using the DOM, if certain SMIL animation features aren't available. Not as elegant, but certainly doable. SVG text and SVG font support is also high on my priority list, otherwise one can't get reliable results in layouts. > > Those are the big ones from the top of my head, but in addition there > are numerous redraw issues (typically foreignObject), instabilities, > and places where excessive memory is used. yes, I noticed a few redraw issues, as well, esp. in conjunction with changing text elements by script. I can write bug reports later, if they hadn't been reported already. > > Also we should ideally support SVG as an image format :D yes, that would be welcome Thank you for your support - if you want me to test certain new SVG features, I'd be happy to do so. Just let me know on this list. Andreas -- ---------------------------------------------- Andreas Neumann Institute of Cartography ETH Zurich Wolfgang-Paulistrasse 15 CH-8093 Zurich, Switzerland Phone: ++41-44-633 3031, Fax: ++41-44-633 1153 e-mail: neumann(a)karto.baug.ethz.ch www: http://www.carto.net/neumann/ SVG.Open: http://www.svgopen.org/ Carto.net: http://www.carto.net/

1 0

SVG display in embed elements chooses ASV instead of Webkit
by Andreas Neumann 20 Nov '06

20 Nov '06

Hello, When I view HTML files with an embedded SVG file (using the <embed /> tag) webkit loads the SVG in the Adobe SVG viewer instead of the native SVG implementation. Is this known or intended behavior? I certainly would prefer the <object /> tag instead of the <embed/> tag, but IE7 does not like Adobe SVG viewer in the object tag :-( Here is an example: http://www.carto.net/papers/svg/samples/canvas.shtml I am testing with MacOSX 10.4.8 (PowerPC) with the latest Apple Webkit (nightly build). Thanks for any information on this issue. Andreas -- ---------------------------------------------- Andreas Neumann Institute of Cartography ETH Zurich Wolfgang-Paulistrasse 15 CH-8093 Zurich, Switzerland Phone: ++41-44-633 3031, Fax: ++41-44-633 1153 e-mail: neumann(a)karto.baug.ethz.ch www: http://www.carto.net/neumann/ SVG.Open: http://www.svgopen.org/ Carto.net: http://www.carto.net/

2 2

cairo.h not found when building on Windows (and some other errors)
by Maciej Kołodziej 16 Nov '06

16 Nov '06

All, I'm trying to build WebKit on my Windows machine. I have cygwin installed, load the vcvars32.bat and run WebKitToos/Scripts/build-webkit. Everything goes fine for some time (I get over 300 MB in the build output folder) but then I get this in BuildLog.htm: ... HTMLAreaElement.cpp HTMLAppletElement.cpp HTMLAnchorElement.cpp FormDataList.cpp CanvasStyle.cpp CanvasRenderingContext2D.cpp c:\Projects\eudora\WebKit\WebCore\platform\graphics\AffineTransform.h(36) : fatal error C1083: Cannot open include file: 'cairo.h': No such file or directory CanvasPattern.cpp Compiling... CanvasGradient.cpp ... And then below: ... PageWin.cpp ..\..\bridge\win\PageWin.cpp(42) : error C2027: use of undefined type 'WebCore::FrameView' c:\projects\eudora\webkit\webcore\rendering\RenderObject.h(48) : see declaration of 'WebCore::FrameView' ..\..\bridge\win\PageWin.cpp(42) : error C2227: left of '->containingWindow' must point to class/struct/union/generic type ..\..\bridge\win\PageWin.cpp(48) : error C2039: 'windowRect' : is not a member of 'WebCore::Page' c:\Projects\eudora\WebKit\WebCore\page\Page.h(42) : see declaration of 'WebCore::Page' ..\..\bridge\win\PageWin.cpp(49) : error C2270: 'windowRect' : modifiers not allowed on nonmember functions ..\..\bridge\win\PageWin.cpp(50) : error C3861: 'mainFrame': identifier not found ..\..\bridge\win\PageWin.cpp(55) : error C2440: 'return' : cannot convert from 'RECT' to 'WebCore::FloatRect' No user-defined-conversion operator available that can perform this conversion, or the operator cannot be called ..\..\bridge\win\PageWin.cpp(58) : error C2039: 'setWindowRect' : is not a member of 'WebCore::Page' c:\Projects\eudora\WebKit\WebCore\page\Page.h(42) : see declaration of 'WebCore::Page' ..\..\bridge\win\PageWin.cpp(60) : error C3861: 'mainFrame': identifier not found FrameWin.cpp c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(39) : error C4430: missing type specifier - int assumed. Note: C++ does not support default-int c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(39) : error C2143: syntax error : missing ',' before '&' c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(41) : error C4430: missing type specifier - int assumed. Note: C++ does not support default-int c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(41) : error C2143: syntax error : missing ',' before '*' c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(58) : error C2146: syntax error : missing ';' before identifier 'objectContentType' c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(58) : error C2433: 'WebCore::FrameWin::ObjectContentType' : 'virtual' not permitted on data declarations c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(58) : error C4430: missing type specifier - int assumed. Note: C++ does not support default-int c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(58) : error C4430: missing type specifier - int assumed. Note: C++ does not support default-int c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(58) : warning C4183: 'objectContentType': missing return type; assumed to be a member function returning 'int' c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(59) : error C2143: syntax error : missing ';' before '*' c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(59) : error C2433: 'WebCore::FrameWin::Plugin' : 'virtual' not permitted on data declarations c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(59) : error C4430: missing type specifier - int assumed. Note: C++ does not support default-int c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(59) : error C4430: missing type specifier - int assumed. Note: C++ does not support default-int c:\projects\eudora\webkit\webcore\bridge\win\FrameWin.h(59) : warning C4183: 'createPlugin': missing return type; assumed to be a member function returning 'int' ..\..\bridge\win\FrameWin.cpp(62) : error C3861: 'clearRecordedFormValues': identifier not found ..\..\bridge\win\FrameWin.cpp(75) : error C2039: 'm_submittedFormURL' : is not a member of 'WebCore::FramePrivate' c:\Projects\eudora\WebKit\WebCore\page\FramePrivate.h(41) : see declaration of 'WebCore::FramePrivate' ..\..\bridge\win\FrameWin.cpp(77) : error C2039: 'm_submittedFormURL' : is not a member of 'WebCore::FramePrivate' c:\Projects\eudora\WebKit\WebCore\page\FramePrivate.h(41) : see declaration of 'WebCore::FramePrivate' ..\..\bridge\win\FrameWin.cpp(80) : error C2102: '&' requires l-value ..\..\bridge\win\FrameWin.cpp(82) : error C3861: 'clearRecordedFormValues': identifier not found ..\..\bridge\win\FrameWin.cpp(96) : error C2027: use of undefined type 'WebCore::FrameView' c:\projects\eudora\webkit\webcore\rendering\RenderObject.h(48) : see declaration of 'WebCore::FrameView' ..\..\bridge\win\FrameWin.cpp(96) : error C2227: left of '->containingWindow' must point to class/struct/union/generic type ..\..\bridge\win\FrameWin.cpp(105) : error C2027: use of undefined type 'WebCore::FrameView' c:\projects\eudora\webkit\webcore\rendering\RenderObject.h(48) : see declaration of 'WebCore::FrameView' ..\..\bridge\win\FrameWin.cpp(105) : error C2227: left of '->containingWindow' must point to class/struct/union/generic type ..\..\bridge\win\FrameWin.cpp(128) : error C3861: 'prepareForUserAction': identifier not found ..\..\bridge\win\FrameWin.cpp(155) : error C2660: 'WebCore::FrameWinClient::createNewWindow' : function does not take 3 arguments XMLTokenizer.cpp ... Anyone knows how to fix those errors? These files are present in the source tree, so I guess there's somethng wrong with the build paths. There's also one more error which I think is connected with the first one: ... RenderThemeWin.cpp ..\..\rendering\RenderThemeWin.cpp(26) : fatal error C1083: Cannot open include file: 'cairo-win32.h': No such file or directory ... Thanks, Maciej.

2 1

gdk build/ js support issues?
by schmindor 10 Nov '06

10 Nov '06

Not sure what I am seeing here, but installed the latest sources for the gdk build, built, and tried a few test websites. Noticed: document.write not working (checked, does work in the windows and safari builds); then tried using other javascript that would give me a visible indication since I know alerts aren't implemented for GDK yet...doesn't work either (attached)... Is anyone else using the GDK version? Ideas as to why javascript support is having issues for just that platform?

2 4

Proposal for Support Library repository
by Mike Emmel 09 Nov '06

09 Nov '06

Most of the webkit ports outside of the Mac are dependent on a number of support libraries and build tools. In some cases these are modified and or later versions then those generally shipped with the system. In general no one has the right versions of all libraries installed. The proposal is to create a new svn root called say webkitlibs that would act as a place to keep source versions of the various libraries and uncommon tools used with webkit. This would allow the use of a simple script in many cases that could check the local versions and grab only needed remote versions of various libs. Later prebuilt support libs could be added. In some cases say KDE or gdk the repository could simply house scripts that grabbed the right versions from the original server. Trying to track down all the libraries needed to build webkit for each platform is quite problematic today. This would also make it easy to create readme and wiki instructions later as webkit usage expands. The reason to go with a second repository is it will in general no one person needs all possible downloads and adding them to the WebKit repository would quickly bloat the download size. This could also if wanted become a place to house other open source projects that build on webkit.

1 0

(no subject)
by Mike Fischer 08 Nov '06

08 Nov '06

Hi! I am seeing a strange problem with transparent PNG files in the release version of Safari/WebKit on Max OS X 10.4.8 PPC: I created a 1 pixel PNG image with color white and an alpha value of approx. 60%. When I display this in an IMG tag (or as a CSS background image) over other content the other content does not appear lighter as expected but more gray (not really darker). When I use a CANVAS element to draw a white fill color with 60% alpha the background is lightened as expected. The current nightly build does not show this problem. By searching for bugs related to PNG on <http://bugs.webkit.org> I couldn't find anything relevant. Also Firefox 1.5.0.7+, Opera 9.02 and OmniWeb 5.5.1 show the expected color effect, i.e. lightening the background. OTOH using a 1 pixel PNG with color black and an alpha of 50% works exactly as expected, i.e. darkening the background for a shadow effect. My questions: a) Was there a bug in WebKit causing this that got fixed for the current nightlies? Or is this some unintended side effect and I need to file a new bug to ensure that this stays fixed in the future? b) Is there a workaround for the released version of WebKit (419.3)? (Other than using a CANVAS element ;-) A test page demonstrating the effect is here (the top transparent area shows the problem): <http://home.fi-works.de/test/test-png-alpha.html> Note that the testing area is built using JavaScript DOM methods in this example. It essentially builds three layers: At the bottom some text is created. Then (DIV id:"bgc1" z.index:1) some color areas are created and in the top layer (DIV id:"bgc2" z-index:2) the transparent testing areas are created. Thanks! Mike -- Mike Fischer Softwareentwicklung, EDV-Beratung Schulung, Vertrieb Web: <http://homepage.mac.com/mike_fischer/index.html>

4 6