August 2009 - webkit-dev - lists.webkit.org

Can someone give current status for Webkit/GTK build under mingw
by Nevo 05 Aug '09

05 Aug '09

hi : Im not sure if Im asking the right place .... Can someone give the current status for Webkit/GTK build under MinGW ? Now I've tried to configure and compile Webkit/gtk 1.1.5 under my Mingw cross toolchain (gcc 4.2 g++ 4.2 ). Directly "configure/make" was not working as I expected, and after some tunings in generated GNUmakefile and modifications in the code , I was able to compile the webkit libraries. But interestingly , even with "--enable-shared" , I couldnot produce .dll files at all, only .a and .la files are placed under .libs dir. So if someone can point out which version is supporting MinGW without the need of hack, that would be helpful . Thanks Nevo

2 1

Is there a way to call JavaScript easily
by douchuan 05 Aug '09

05 Aug '09

I'm doing a research that extends Java to a dynamic language using JavaScript. I have readed some docs about SpiderMonkey, Rhino, JavaScriptCore, but lots of information just about how to call Java from JS easily, but there isn't any information how to call JS from Java. Is there possibility about my study? thx

1 0

willCacheResponse
by Darin Fisher 05 Aug '09

05 Aug '09

There is a Mac-only willCacheResponse notification that gets recorded by several of the layout tests when layoutTestController.dumpResourceLoadCallbacks(true) is called. However, this notification which originates at the ResourceHandle level is currently #if PLATFORM(MAC). I noticed that the tests which depend on this notification are currently skipped on platform/win. For Chromium, we have simply rebaselined with the willCacheResponse events excluded. Having a separate baseline is unfortunate and not running the tests on platform/win seems even worse. I wonder if it wouldn't be better to exclude willCacheResponse by default, and then only enable it selectively for tests that are about testing it. That way, we can un-skip the other tests that only fail on non-Mac as a side-effect of this notification not being supported. We could also try to support willCacheResponse on all platforms, but that seems like a lot of work for little gain. Afterall, it is a notification that WebKit doesn't itself care about. It is just something provided to the embedder of the WebKit Framework on Mac. I didn't see any bugs on file about this issue. -Darin

2 1

freeing static variables
by Zoltan Herczeg 04 Aug '09

04 Aug '09

Hi all, Valgrind reports a lot of memory leaks when QtLauncher quits, because many static local variables are not freed. I did a little research and realized there is no consistent way to define static variables in webkit: WebCore/css/CSSSelector.cpp: CSSSelector::extractPseudoType() using a DEFINE_STATIC_LOCAL() macro WebCore/bindings/js/JSDOMWindowBase.cpp: JSDOMWindowBase::commonJSGlobalData() static JSGlobalData* globalData; WebCore/platform/qt/CursorQt.cpp: Cursors* Cursors::s_self = 0; (no static keyword) I belive it would be a good thing to define a template for static variables, which can (optionally) free static variables (enabled in debug mode, but sometimes it would be good to enable it in release mode as well) And I am wondering whether it would be worth to free (some) static variables if they are used in a given time period. This may help to reduce the memory consumption in the long run. This would be an optional feature, an extra flag (or timeout value) for the template, which can be enabled or disabled at compile time. If disabled, the value of this flag is not used at all. Regards Zoltan

5 6

Care and feeding of the commit-queue flag
by Adam Barth 04 Aug '09

04 Aug '09

If you're not interested in automating patch landings, you can ignore this message. As of this weekend, we have an experimental commit queue working. The queue still requires a bunch of manual attention from me, but I'm hoping to gradually reduce that over time without changing the interface. Here's how to use the queue: * If you're a committer, you can approve a reviewed patch for the commit queue by setting the commit-queue+ flag on the patch. The next time the commit queue scans Bugzilla, it will pick up the patch, build it, test it, and land it. The primary use case is if you just reviewed a patch by a non-committer, you can set the commit-queue+ flag at the same time as the r+ flag and the commit queue will take care of committing the patch for you. WARNING: Approving a patch for the commit queue (i.e., setting the commit-queue+ flag) is the moral equivalent of committing the patch. You should only set this flag if you would otherwise commit the patch. The commit queue does some sanity checks (like making sure the patch has an r+ from a reviewer), but you should exercise your good judgement. Frequently Asked Questions: Q) This commit queue thing is dumb. Can I opt out? A) Sure! Just ignore the commit-queue flag and use whatever workflow you like. Q) How does the commit queue know I'm a committer? A) It looks at <http://trac.webkit.org/browser/trunk/WebKitTools/Scripts/modules/committers…>. We're missing a bunch of committers. If you'd like to be listed, send me a patch. :) Q) How often does the commit queue scan Bugzilla for approved patches? A) When I'm running the queue, it scans every 10 minutes. I'm probably more likely to run the queue at night for a while to avoid disrupting the tree. Q) Can I see my patch make progress through the commit queue? A) Not yet. I need to figure out how to give feedback. Currently it spews a bunch of text to a terminal on my machine, but no one else can see that. Q) Help! The commit queue destroyed the tree! A) Please look for me on IRC and we'll figure out what went wrong. Q) This commit queue thing is great! Can I run the commit queue on my machine? A) Almost all the code is checked into SVN. The only missing piece is the master process that controls the queue. If you're really interested, I can check in this script, but my current plan is to iterate on it a few times. Q) What platforms does the commit queue build / test? A) Currently, it just builds and tests the Mac port on my laptop. As try servers come online, we can integrate them into the commit queue for better pre-commit coverage. Thanks for your patience while we try out this system. Hopefully it will be useful. Adam

1 0

Towards a commit-queue
by Adam Barth 04 Aug '09

04 Aug '09

I've been experimenting with a wrapper for bugzilla-tool that runs in a loop and tries to land patches [1]. Ideally, there would be enough information in Bugzilla so I could kick off this script and make tea while it does its thing. However, deciding whether a bug with an r+ patch is ready to be landed requires human intervention. I went through the queue tonight and manually marked the bugs as commit-ready by adding [commit+] to their titles. This worked well, except that it made the bug titles ugly. I think a better solution is to have flag analogous to r+ which is commit+. The flag means "I certify that this patch is ready to be landed according to WebKit process." Unlike r+, this flag would be settable by any committer because it is the moral equivalent of typing "svn commit". In rare cases, such as unreviewed build fixes, a patch could have commit+ without an r+. In future, future land when we have a working commit queue, the actual "svn commit" command might be run by a bot instead of a human. Adam [1] You can see my progress here: https://bugs.webkit.org/show_bug.cgi?id=27918

8 13

SLL_Pop list is not checked for NULL
by Luka Napotnik 03 Aug '09

03 Aug '09

In JavaScriptCore/wtf/FastMalloc.cpp:696 The list should be checked for NULL. Eg. static inline void *SLL_Pop(void **list) { if (!list || !(*list)) return NULL; void *result = *list; *list = SLL_Next(*list); return result; } Greets, Luka

4 4

Reporting exceptions from worker context to users
by Drew Wilson 03 Aug '09

03 Aug '09

Hi all, Currently, unhandled exceptions are sent from worker context over to the parent page where they are logged to the console. This works fine for dedicated workers, but not for shared workers which can have multiple active windows. The immediate solution that springs to mind is to broadcast the exception to every window associated with a shared worker, and have each window log the unhandled exception to its console. Is there another option that might be better (is there the concept of an inspector/console that is not associated with a specific window, for example)? -atw

4 7

click() method for anchor and img elements in WebKit API
by Piotr Dobrogost 03 Aug '09

03 Aug '09

Hi According to https://developer.mozilla.org/en/DOM/element.click Gecko does not support click() method on some elements (like anchor or img). Does WebKit support click() method on anchor and img elements? If not are there any plans to support it and is it possible to implement this functionality atop of current API? I'm using Qt port of WebKit and it would be enough for me if this port would have such a function in its API. -- Piotr Dobrogost *** curlpp.org - c++ wrapper for libcurl ***

1 0

JavaScript bindings changed to call scriptExecutionContext virtual function?
by Darin Adler 02 Aug '09

02 Aug '09

I noticed that many JavaScript binding implementations are now calling the virtual function scriptExecutionContext on DOM nodes. This should never be done! That's a virtual function, so it's not as fast as calling document(). The virtual Node::scriptExecutionContext() function should be made private so that we can catch at least some cases of this programming mistake at compile time. When this change was made, this presumably slowed down DOM bindings, since document() is an inline function that expands to a single memory read and scriptExecutionContext() is a virtual function call. When was this done? Can the person who did it fix it? There are various techniques that can be used to fix this. One is to only call scriptExecutionContext in generic code, and use functions named document() in less generic code that's working with a Node rather than an EventTarget. Another is to break scriptExecutionContext into virtual and non- virtual parts as we do with functions like Node::prefix and Node::virtualPrefix. -- Darin

4 7