October 2006 - webkit-dev - lists.webkit.org

calling javascript from Netscape style plugin
by Shalini Kamala 31 Oct '06

31 Oct '06

2 2

State of WebKit on Qt4 (aka Unity)
by Stefan Eilers 31 Oct '06

31 Oct '06

Hi! We are very interested to do some work for KHtml on Qt4. As it was launched as Unity and the patches are integrated into WebKit, we think this is a promising situation for the future. To get a first overview, I would like to ask what is currently missing? Where are the blank spots? Thanks in advance! Mfg., Dr. Stefan Eilers -- Dr.-Ing. Stefan Eilers Projekt Manager basysKom GmbH Robert-Bosch-Str. 7 | 64293 Darmstadt | Germany Tel: +49 6151 3969-962 | Fax: -736 | Mobile: +49 170 4213459 | Jabber: eilers(a)jabber.org stefan.eilers(a)basyskom.de | www.basyskom.de

2 1

Compiling with cmake on windows..
by Stefan Eilers 31 Oct '06

31 Oct '06

Hi! I try to compile WebKit with Qt/KDE support on Windows. There is one dependency issue, which I cannot solve. It asks for the directory that contains the file "LibXml2Config.cmake". I don't have this file on my system and it is not included in the libxml2 distributions.. Can somebody please tell me where to find it!? Thanks a lot! Mfg., Dr. Stefan Eilers -- Dr.-Ing. Stefan Eilers Projekt Manager basysKom GmbH Robert-Bosch-Str. 7 | 64293 Darmstadt | Germany Tel: +49 6151 3969-962 | Fax: -736 | Mobile: +49 170 4213459 | Jabber: eilers(a)jabber.org stefan.eilers(a)basyskom.de | www.basyskom.de

3 5

Re: Safari 2.0.4 bug with .qtl files?
by Mark Edwards 29 Oct '06

29 Oct '06

On Oct 29, 2006, at 1:48 PM, Scott Reynen wrote: > On Oct 29, 2006, at 3:03 PM, Mark Edwards wrote: > >> Well, its a bug that needs to be fixed in Safari. .qtl files are >> supposed to work in this way, its documented by Apple. I have >> legacy CD-ROM's that currently fail with Safari. >> >> What needs to be done to get this bug filed properly? > > You might want to contact the people who develop Safari. Contrary > to popular misconception, they aren't found on this list. They are > found here: > > http://webkit.opendarwin.org/contact.html > http://lists.apple.com/mailman/listinfo/webcore-dev Thanks Scott. I had forgotten I already filed a bug there: http://bugs.webkit.org/show_bug.cgi?id=10777 I also filed a bug with bugreport.apple.com. I am CC'ing webkit-dev(a)opendarwin.org. All of the relevant info on this issue is documented at the above link. Thanks. -- Mark Edwards

1 0

Javascript / Obj-c object property mismatch
by Rudi Sherry 26 Oct '06

26 Oct '06

I'm working with a WebView where I use DOM object manipulation in objective-c combined with javascript. I have the dom document and I add an object with some attributes like this: DOMNodeList *list = [ domDoc getElementsByTagName: @"BODY" ]; DOMNode *body = [ list item: 0 ]; myObject = (DOMHTMLObjectElement *) [ [ domDoc createElement: @"OBJECT" ] retain ]; [ myObject setAttribute: @"ID" : @"MyObjectName" ]; [ myObject setAttribute: @"NAME" : @"MyObjectName" ]; <snip> [ myObject setAttribute: @"command" : @"xxx" ]; [ body appendChild: myObject ]; In the javascript, I can get the object with this and check its value (this is in a javascript function that gets called): var myObject = document.body.children['MyObjectName']; alert(myObject.command); myObject.command = "abcdef"; alert(myObject.command); I have a WebUIDelegate with the runJavaScriptAlertPanelWithMessage that prints out the alert value and compares it against myObject: - (void)webView:(WebView *)sender runJavaScriptAlertPanelWithMessage: (NSString *)message { NSLog (@"alert: %@", message); NSLog( @"command attribute: %@", [ mSwfObject getAttribute: @"command" ] ); } In the above example, it prints out: 2006-10-26 08:44:29.771 TestInsideWebKit[9235] alert: xxx 2006-10-26 08:44:29.771 TestInsideWebKit[9235] command attribute: xxx 2006-10-26 08:44:29.771 TestInsideWebKit[9235] alert: abcdef 2006-10-26 08:44:29.771 TestInsideWebKit[9235] command attribute: xxx Why isn't the command attribute changed in the objective-c object when it's changed in the javascript function? It must be the same object since it started as 'xxx'. Any help would be appreciated, Thanks, Rudi

2 2

Removing table cells
by Patrick Hanna 24 Oct '06

24 Oct '06

I'm running into a problem where javascript calls removeChild on a table row and RenderTableCell::destroy() is called. The problem is that CellStruct.cell is not set to zero so RenderTableSection::paint tries to paint a deleted cell. Does anyone know how and where table cells are removed from the render tree? Thanks, Pat

1 0

NetScapeMoviePlugin
by Shalini Kamala 23 Oct '06

23 Oct '06

2 8

Nightly builds
by Adam Bryzak 21 Oct '06

21 Oct '06

Hi, Is it just me or are the nightly builds updated really frequently, like multiple times per day? It's getting hard to keep up! Adam Bryzak

3 3

Builds on which platforms?
by Andre-John Mas 19 Oct '06

19 Oct '06

Hi, Although my primary interest for WebKit is MacOS X, is there any chance that we could add to the web page a list of platforms that WebKit works on and kinda works on? Also for the platforms that WebKit kinda works on mention what the platform specific issues are. Andre

1 0

loader refactoring plan
by Maciej Stachowiak 19 Oct '06

19 Oct '06

Hi everyone, As you may know, a number of us have been tearing apart the WebKit loader-related code to make it simpler and more portable. I decided I should post a few more details of the ongoing refactoring plan. For starters, let me give a brief overview of how loading worked in WebKit prior to all the refactoring work, for a few different scenarios. These diagrams are oversimplified but should give the basic idea. Apologies for the HTML formatting of this message but I wanted to make sure the ASCII art came out right. Main resource loaded, started from app via API: [ WebFrame ] (ObjC, WebKit) [ WebDataSource ] (ObjC, WebKit) [ WebMainResourceLoader ] (ObjC, WebKit) [ WebFrameBridge ] (ObjC, WebKit) [ WebCoreFrameBridge ] (ObjC, WebCore) [ FrameMac ] (C++, WebCore) [ Frame ] (C++, WebCore) Main resource loaded, started from engine via link click or form submission: [ Frame ] (C++, WebCore) [ FrameMac ] (C++, WebCore) [ WebCoreFrameBridge ] (ObjC, WebCore) [ WebFrameBridge ] (ObjC, WebKit) [ WebFrame ] (ObjC, WebKit) [ WebDataSource ] (ObjC, WebKit) [ WebMainResourceLoader ] (ObjC, WebKit) [ WebFrameBridge ] (ObjC, WebKit) [ WebCoreFrameBridge ] (ObjC, WebCore) [ FrameMac ] (C++, WebCore) [ Frame ] (C++, WebCore) Subresource load: [ DocLoader ] (C++, WebCore) [ Cache ] (C++, WebCore) [ Loader ] (C++, WebCore) [ ResourceLoader ] (C++, WebCore) [ WebCoreResourceLoaderImp ] [ WebCoreFrameBridge ] (ObjC, WebCore) [ WebFrameBridge ] (ObjC, WebKit) [ WebFrame ] (ObjC, WebKit) [ WebDataSource ] (ObjC, WebKit) [ WebSubresourceLoader ] (ObjC, WebKit) [ WebCoreResourceHandle ] (ObjC, WebCore) [ ResourceLoader ] (C++, WebCore) [ Loader ] (C++, WebCore) [ Cache ] (C++, WebCore) [ CachedResource ] (C++ WebCore) Notice how it all goes through a crazy number of layers, with crazy jumping back and forth between WebCore and WebKit. Also, much of the essential logic is in platform-specific Objective-C or Objective-C++ code. Here is the new target design: Main resource loaded, started from app via API: [ WebFrame ] (ObjC, WebKit) [ FrameLoader ] (C++, WebCore) ---> [ FrameLoaderClient ] (C++ abstract base class, implemented in webkit to deliver delegates) [ DocumentLoader ] (C++, WebCore) [ MainResourceLoader ] (C++, WebCore) [ ResourceLoader ] (C++, WebCore) [ FrameLoader ] (C++, WebCore) Main resource loaded, started from engine via link click or form submission: [ FrameLoader ] (C++, WebCore) ---> [ FrameLoaderClient ] (C++ abstract base class, implemented in webkit to deliver delegates) [ DocumentLoader ] (C++, WebCore) [ MainResourceLoader ] (C++, WebCore) [ ResourceLoader ] (C++, WebCore) [ FrameLoader ] (C++, WebCore) Subresource load: [ DocumentLoader ] (C++, WebCore) [ Cache ] (C++, WebCore) [ FrameLoader ] (C++, WebCore) ---> [ FrameLoaderClient ] (C++ abstract base class, implemented in webkit to deliver delegates) [ SubresourceLoader ] (C++, WebCore) [ ResourceLoader ] (C++, WebCore) [ Cache ] (C++, WebCore) [ CachedResource ] (C++ WebCore) As you can see, there will be much fewer layers involved in most loads, and the vast majority of the logic will be cross-platform C++ in WebCore. WebKit will be limited to channeling down API calls and channeling up delegate callbacks, since these are essential points of type and language translation. Here's where we stand right now: - All loading related code has been factored out of WebFrame and WebDataSource into WebFrameLoader and WebDocumentLoader. - Dispatch of delegates is via WebFrameLoaderClient ObjC protocol declared in WebCore and implemented in WebKit - All loader-related classes have been moved out of WebKit and into WebCore/loader/mac - All loader-related code has been pulled out of WebFrameBridge and moved down to WebCoreFrameBridge Here are some steps that remain: - Move loading-related code from WebCoreFrameBridge to WebFrameLoader and collapse layers as necessary - Move loading-related code from Frame and FrameMac to WebFrameLoader and collapse layers as necessary - Convert loader/mac code from ObjC classes to C++ classes - Make WebFrameLoaderClient a C++ abstract base class instead of an ObjC protocol - Shuffle things around so that Frame owns a FrameLoader rather than the bridge owning it. - Merge DocLoader with WebDocumentLoader - Change loader/mac code to use WebCore types - Build up ResourceLoader abstraction enough to support everything needed by the higher-level logic. We're trying to do this stuff sort of in parallel. I am working on cleaning up and building out the ResourceLoader API. Here is a first- pass set of headers that show the planned interface for ResourceLoader, ResourceLoaderClient, ResourceRequest, ResourceResponse, and related classes: Some details of this may change. I'm not sure CachedResourceResponse is a great name or fits so well in concept as it stands. I am going to start by changing ResourceLoader to the new API, then ResourceLoader and ResourceLoaderClient are next, to be followed by addition of ResourceResponse. The more special-purpose types will come last. I'll also try to document this interface, since it is one of the main things platform porters will have to implement. Regards, Maciej

1 0