September 2006 - webkit-dev - lists.webkit.org

Re: [webkit-dev] XHTML, XML and WebKit editing
by Rob Burns 15 Sep '06

15 Sep '06

Todd and Maciej, Thank you both for your help. I was not loading this as an application/xhtml+xml but instead as html. This has taken care of many issues that i though i was going to have to work around and makes me quite happy. The content is no longer being rearranged, and its looks like well-formed and valid xhtml. I'm curious why the DOM would rearrange the HTML content in that way. Any ideas? BTW Maciej, why would getElementsByName not work? I haven't tried, but the other HTML specific methods seem to be working. From what I can tell the XHTML content still seems to be instantiated with the DOMHTML element subclasses of the DOMCore classes. I think that's how it should be. if the DOM is to treat HTML elements as subclasses then it makes sense to treat the XHTML elements as HTML subclasses too (at least for XHTML 1.0 and probably 1.1). Beyond that it may require new subclasses descended off of the core DOM classes or perhaps another approach. For example, CoreDOM nodes could associate instance variables (say dictionaries) that encapsulate properties about that element-class. So for example, a 'img' element could be instantiated as a straight CoreDOM element but with instance variables indicating that it can have 'id', class', title', etc., attributes and must have a 'src' and 'alt' attribute. Other instance variables would indicate its prohibited, required, permitted, minimal and maximum content model. By including these instance variables in the CoreElement (and CoreAttribute) objects, the DOM would be duly abstracted and easily extended to other situations. Ideally these instance variables could be set directly from the document schema and sub-schema. Perhaps this it the direction things are heading already, but for now, I'm happy to see the HTML subclasses work. I'm using a few category methods attached to a couple of the DOMHTML subclasses. Thanks again, best wishes, Rob On Aug 24, 2006, at 3:56 PM, Todd Ditchendorf wrote: > Are you loading the XHTML programatically from ObjC? If so, are you > using - [WebFrame loadHTMLString:baseURL:] ? > > If that's the case, I'm not totally positive, but I think using > this method instad will fix your issue: > > - [WebFrame loadData:MIMEType:textEncodingName:baseURL:] > > and passing in @"application/xml" or @"text/xml" as the mime type. > > Todd > > > On Aug 24, 2006, at 12:33 PM, Rob Burns wrote: > >> I sent the following message to the sdk list, but perhaps this >> list is more appropriate. Briefly, when loading XHTML into an >> editable WebView, the contents are getting rearranged: moving >> elements from the HEAD to the BODY. I thought someone here might >> know where the code is located that does this editing of rhd >> source upon load. In addition to rearranging elements, the WebView >> is also removing "/" from self-closing elements. I >> >> Originally, was loading the file as HTML, but i've since >> corrected that but still encounter this bug. >> >> (http://bugzilla.opendarwin.org/show_bug.cgi?id=10507) >> >> Thanks for any assistance. >> >> Rob >> >> >> >>> I'm working on an application that uses WebKit's editing >>> capabilities. Though the headers show ample support for XHTML >>> concepts and Safari handles XHTML very well, the use of XHTML in >>> an editing situation seems to be difficult. >>> >>> 1) First of all, WebView immediately converts all elements in an >>> XHTML document to upper case, which is not consistent wit the >>> recommendation. >>> >>> 2) It also converts all self-closing elements to HTML style >>> implicitly closed elements. >>> >>> 3) WebView removes the DocType declaration and the XML >>> declaration from the document. >>> >>> I've found work-arounds for these unfortunate bugs, but now I'm >>> encounter more difficult problems. If an XHTML document includes >>> elements from another namespace in the head, WEbView graciously >>> moves them to the body element instead. The previous issues I >>> planed to just fix upon save, but now I can't work with non-XHTM >>> elements in the head whiole the document is editable by WebView >>> because WebView jmoves those elements to where they do not >>> belong. I filed a bug report on this, (http:// >>> bugzilla.opendarwin.org/show_bug.cgi?id=10507) >>> >>> Has anyone else encountered these problems and found ways to work >>> around them. I'd like to get in there and fix them myself, but >>> I'm not that familiar with C++ and I'm still getting my head >>> around the vast array of classes. Any thoughts? >>> >>> best, >>> Rob Burns >> >> _______________________________________________ >> webkit-dev mailing list >> webkit-dev(a)opendarwin.org >> http://www.opendarwin.org/mailman/listinfo/webkit-dev >

4 5

crash with WebKit
by Rob Burns 15 Sep '06

15 Sep '06

I'm getting a repeated crash with WebKit. It seems to be related to a delay in loading XML into the main WebFrame: especially on the initial launch. The small XHTML loads quicker on subsequent new windows. If the window resigns key while the data is loading the crash occurs. Usually this load happens very quickly and so there's no time for the window to resign key while loading, but with my delay its happening more frequently. So I guess what's happening here is the WebCore FontFallback is trying to respond to a notification about a window resigning key and then query the WebView for information that's just not there (since the data's not finished loading). I'm not familiar enough with the WebKit codebase to understand much more than that. The crash log: #0 0x093145cc in WebCore::FontFallbackList::fontDataAt at FontFallbackList.cpp:60 #1 0x0926f830 in WebCore::Font::lineSpacing at Font.cpp:391 #2 0x091cf154 in WebCore::RenderObject::lineHeight at RenderObject.cpp:2541 #3 0x091bb60c in WebCore::RenderFlow::lineHeight at RenderFlow.cpp:298 #4 0x091984e8 in WebCore::RenderBlock::lineHeight at RenderBlock.cpp: 3248 #5 0x09194a10 in WebCore::RenderBlock::layoutInlineChildren at bidi.cpp:1751 #6 0x091a5394 in WebCore::RenderBlock::layoutBlock at RenderBlock.cpp: 487 #7 0x091a4cc8 in WebCore::RenderBlock::layoutBlockChildren at RenderObject.h:465 #8 0x091a53cc in WebCore::RenderBlock::layoutBlock at RenderBlock.cpp: 489 #9 0x091b0ab8 in WebCore::RenderView::layout at RenderView.cpp:117 #10 0x09104428 in WebCore::FrameView::layout at FrameView.cpp:473 #11 0x090e9408 in WebCore::Frame::setIsActive at Frame.cpp:3343 #12 0x9293fad8 in _nsnote_callback #13 0x90803010 in __CFXNotificationPost #14 0x907fb0ec in _CFXNotificationPostNotification #15 0x92929ee0 in -[NSNotificationCenter postNotificationName:object:userInfo:] #16 0x9374c95c in -[NSWindow becomeKeyWindow] #17 0x9374c138 in -[NSWindow _changeKeyAndMainLimitedOK:] #18 0x9374b288 in -[NSWindow makeKeyAndOrderFront:] #19 0x93827708 in -[NSApplication _cycleWindowsReversed:] #20 0x93789988 in _NSKeyboardUIHandleSymbolicHotKey #21 0x937e49a8 in -[NSApplication _handleKeyEquivalent:] #22 0x936ee408 in -[NSApplication sendEvent:] #23 0x936e5d10 in -[NSApplication run] #24 0x937d687c in NSApplicationMain #25 0x005d9fac in main at main.m:13 The crash log doesn't touch any of my classes, so I thought I'd see if anyone is familiar with this crash. Best Wishes, Rob

1 0

TREE IS CLOSED for performance regresson
by Maciej Stachowiak 14 Sep '06

14 Sep '06

Hi everyone, We've had a ~10% performance regression sometime this week - the TREE IS CLOSED while we investigate. Regards, Maciej

1 0

S60 WebKit Browser Plug-ins
by Ashish Gokhale 13 Sep '06

13 Sep '06

Dear All, We need to develop a MIME-type plug-in for handling text/xml content. We have tried the NPBitmap example that comes along with S60 3rd Ed Kit but that doesn't seem to work. Our plug-in compiles correct, but not called by the plug-in framework. Any clues how to test plug-ins on S60 Emulator? Thank you -AG

1 0

default style sheets?
by schmindor 12 Sep '06

12 Sep '06

Is there a default style-sheet implemented for windows for items without specified styles, for instance, links are blue, clicked links are purple, text boxes turn blue when they get focus, etc? If so, when can I find it?

5 5

Safari Extensions
by riklar 09 Sep '06

09 Sep '06

Hello, When opening Webkit, I get a message, telling me to disable Safari Extensions on my HD, to prevent Webkit from crashing. I'll be glad if someone could indicate me where are these extensions located. TIA Rik

1 0

Linux cache implementation?
by schmindor 08 Sep '06

08 Sep '06

Can someone point me at the code/ classes for how pages are put into the cache in the Linux port? Has this effort been started?

2 1

Re: still using cmake < 2.4.3 ?
by Adam Treat 07 Sep '06

07 Sep '06

On Thursday 07 September 2006 4:42 pm, Alexander Neundorf wrote: > On Thursday 07 September 2006 22:35, Adam Treat wrote: > > Regression problem!!! > > > > Right now if you try and build unity AND/OR WebKit from Apple with either > > CMake 2.4.3 or cvs trunk, it doesn't work. Can a CMake dev have a look > > at this? > > > > Here is the error building WebKit svn: > > > > .../WebCore/css/cssparser.cpp:3098:25: error: tokenizer.cpp: No such file > > or directory Sorry, I think that was the wrong error. Here is the one that is first encountered. And yes, JSHTMLInputElementBaseTable.cpp is a generated file. WebKit/WebCore/bindings/js/JSHTMLInputElementBase.cpp:25:43: error: JSHTMLInputElementBaseTable.cpp: No such file or directory > Is this a generated file ? > > Bye > Alex

1 0

Re: still using cmake < 2.4.3 ?
by Adam Treat 07 Sep '06

07 Sep '06

Regression problem!!! Right now if you try and build unity AND/OR WebKit from Apple with either CMake 2.4.3 or cvs trunk, it doesn't work. Can a CMake dev have a look at this? Here is the error building WebKit svn: .../WebCore/css/cssparser.cpp:3098:25: error: tokenizer.cpp: No such file or directory Adam On Monday 04 September 2006 3:59 pm, Alexander Neundorf wrote: > Hi, > > if you can answer the above question with "yes", then it's really time for > you to update to cmake 2.4.3 (http://www.cmake.org/HTML/Download.html) > > As announced 3 weeks ago, in one week, september 12th cmake 2.4.3 will be > required to build KDE. > > So, get your shiny new cmake 2.4.3 now ! :-) > > Bye > Alex

1 0

Web Editing and FontManager
by Rob Burns 06 Sep '06

06 Sep '06

Hello all, I'm trying to understand how the FontManager interfaces with WebView. to alter the behavior and replicate in other ways. Does anyone know how FontManager doe s its magic with WebView? What methods are called, etc.? Looking at Blot.app for example, the "Bold" menu item targets FontManager addFontTrait method. After that how does the WebView end up inserting " <SPAN class="Apple-style-span" style="font- weight: bold;"></SPAN>" wrapped around the selectedDOMRange? I imagine much of this happens in the black-box of the FontManager, but what method finally get called in the WebKit APIs? Any pointers would be appreciated. Best wishes, Rob

2 1