January 2006 - webkit-dev - lists.webkit.org

SVN over HTTP
by David.Carson＠nokia.com 01 Feb '06

01 Feb '06

Hi, Those of us behind a firewall, I thought that I would have a go at configuring subversion to operate over a proxy. This would save us from having to go to public WLAN areas to download the source. I configured the ~/.subversion/Servers file and tried the test site (svn cat http://svn.collab.net/repos/svn/trunk/README) This worked fine. Then I tried: svn checkout http://anoncvs.opensource.apple.com/svn/webkit/trunk WebKit and got the following error: --- subversion/libsvn_ra_dav/util.c:780: (apr_err=175002) svn: PROPFIND request failed on '/svn' subversion/libsvn_ra_dav/util.c:780: (apr_err=175002) svn: PROPFIND of '/svn': 301 Moved Permanently (http://anoncvs.opensource.apple.com) --- Have I configured something wrong? David

5 4

KWQ Graphics code on Windows
by Justin Haygood 31 Jan '06

31 Jan '06

We have several libraries we can choose from: - GDI - GDI+ - Windows Presentation Foundation (XP/2003/Vista only) - Cairo (portable) - Mozilla's choice, LGPL'ed Which should we use? They all have their pros/cons GDI is a C library, Windows only, hardware accelerated on Vista, semi-hardware accelerated pre-Vista GDI+ is object oriented, portable (sorta), hardware accelerated on Vista, software accelerated pre-Vista WPF is object oriented, not native C++, hardware accelerated on all supported platforms Cairo is a C library, portable, hardware accelerated depending on implementation

6 15

Test cases needed when checking in bug fixes!
by Darin Adler 31 Jan '06

31 Jan '06

Hi guys. I've noticed recently that a bunch of bugs fixes are going in without regression tests. We must not do that. We need these regression tests so that WebKit keeps getting better instead of playing "Whac-a-Mole" <http://en.wikipedia.org/wiki/Whac-a-Mole#Colloquial_usage>. People investigating bugs: - Please make a reduced test case into a standalone test. This can help us figure out what the bug is, and saves the person fixing the bug precious time. People fixing bugs: - Please write a test. Anders even suggests writing the test before fixing the bug, which I think is a great idea. And if you can't write a test, please make a note in the change log explaining why. Reviewers reviewing patches: - Please do review-minus if there's no test, even if the fix is otherwise correct. Committers checking in patches: - If the reviewer forgot to ask for a test, mark the bug review- minus. I am one of the recent offenders here! At the end of the message I list eleven recent check-ins that: 1) look to me like bug fixes 2) don't mention any new tests that were added (either manual tests or tests in LayoutTests) 3) don't give any excuse for the lack of tests For these we should write some tests now, or update the change log to show the tests we did land and forgot to mention or to give excuses. -- Darin 2006-01-24 Darin Adler <darin(a)apple.com> - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=6696 REGRESSION: JavaScript window.opener set incorrectly 2006-01-24 Darin Adler <darin(a)apple.com> - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=6753 REGRESSION (r12282-12292): Crash loading TinyMCE test app 2006-01-24 Darin Adler <darin(a)apple.com> - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=6732 REGRESSION: Repro crash at Google Maps 2006-01-24 Maciej Stachowiak <mjs(a)apple.com> - fixed crash if a node is removed from the DOM before being reparented due to residual style http://bugzilla.opendarwin.org/show_bug.cgi?id=6778 2006-01-24 Tim Omernick <timo(a)apple.com> <rdar://problem/4339024> <http://bugzilla.opendarwin.org/show_bug.cgi?id=6785> "Fix for possible leak of applet, object, or emb ed (and plugin view)" 2006-01-25 Adele Peterson <adele(a)apple.com> http://bugzilla.opendarwin.org/show_bug.cgi?id=6816 Crash in dispatchMouseEvent - Missing nil check 2006-01-25 Alexander Kellett <lypanov(a)kde.org> Percentage values for width, height are not rendered correctly. http://bugzilla.opendarwin.org/show_bug.cgi?id=5331 2006-01-26 Eric Seidel <eseidel(a)apple.com> SVG shows up blank in WebKit+SVG http://bugzilla.opendarwin.org/show_bug.cgi?id=6620 2006-01-26 Adele Peterson <adele(a)apple.com> - fix for <rdar://problem/4422624> REGRESSION: crash loading webmail.mac.com 2006-01-26 Alexander Kellett <lypanov(a)kde.org> <text> elements with a scale on a parent <g> breaks nodeAtPoint Fix http://bugzilla.opendarwin.org/show_bug.cgi?id=6843 2006-01-26 Beth Dakin <bdakin(a)apple.com> Fix for <rdar://problem/4256504> khtml::RenderBlock::skipWhitespace crashes upon interaction with radio buttons.

2 1

WebKit starter page
by Joost de Valk 31 Jan '06

31 Jan '06

Hi all, xenon once made this layout: http://colloquy.info/webkit-page.png . Today i html-ified it with comments from bdash and some others on #webkit. I didn't have the .psd so had to do some dirty trick on the menu, the rest of it seems ok to me: http://dev.outbreak.nl/webkitsite/index.xhtml replacing the .xhtml by .html works as well ofcourse :) i'd like to hear your thoughts on this! Kind regards, Joost -- Joost de Valk #: AlthA on #webkit @ irc.freenode.net @: webkit(a)joostdevalk.nl W: http://www.joostdevalk.nl "The only people who find what they are looking for in life are the fault finders." - Foster's Law

4 4

Server side DOM manipulation with javascript
by Benoit Marchant 31 Jan '06

31 Jan '06

Hi I'm exploring the possibility of using javascript server side as a templating language, in the context of AJAX applications. To that end both a javascript engine and a DOM document, ideally available as "document" from javascript, are needed. The idea is to be able to reuse the same code server and client side. The DOM document class could be a native one eventually, like the XMLDocument class in Windows. In that context, I don't need rendering, which means that WebCore is most likely more than I would need in that regard. To be more specific, I'm targeting xhtml, and the server would most likely be running Windows. Apparently JavascriptCore now compile and run on windows. Any feedback on real usage there ? Has someone tried that before ? What would be involved in implementing such a solution ? Thanks in advance for any feedback or suggestion. Benoit

3 5

KWQ on Windows
by Patrick Forget 31 Jan '06

31 Jan '06

Hello, I have mentioned before that I have also started a port of kwq to the Windows platform. Somehow, I thought I could continue working on it but didn't find the time. I have been contacted twice by people wanting to finish the port to Windows. What is really missing is a public cvs/svn to help working on a common code. As far as I understand, there is several version (and several level of porting) of the code lying around in private. I didn't want to continue until I am sure no one else try to do same thing. A public cvs would really help to focus and share the work. I would have no trouble working on the part that I do find fun (QPainter being one of it). Can it be possible from Apple to set a win32 branch in there SVN? Or maybe we can continue on the current sourceforge project (by contacting the previous maintainer)? I am ready to put my file public and take time to document my changes (and maybe rediscover what I have done). Other developers might be more advanced in there work, it may be a good opportunity to show up. I will also note that Qt 4.0 (now 4.1) is now available on Windows with a GPL license. If the KWQ api behave exactly the same as Qt, this can be a easy way to port WebCore (it become GPL in this case). The main avantage can be to check if the higher level of WebCore work on Windows (and it may help the KDE Project to get back the changes done by Apple). *Darin Adler wrote:* >There's more work to do at the WebCore level. We're doing a lot to >make a better separation between the platform-independent vs. >platform-specific bits. For example, we're replacing the old "KWQ" >layer (which had API "kind of like" Qt, but was really trying to be a >platform abstraction among other things) with a new structure where >platform-independent code is more cleanly separated.' Can we get a quick overview on the changes planned (in the folder /platform I presume) ? It would help to see if porting the current work on kwq now is needed now or we should wait a bit. Patrick Forget

2 1

"Meta" keyword for bugzilla?
by David D. Kilzer 31 Jan '06

31 Jan '06

It would be nice to have a "Meta" keyword for bugzilla.opendarwin.org so that meta (tracking/master) bugs like these could be flagged as such: http://bugzilla.opendarwin.org/show_bug.cgi?id=6851 http://bugzilla.opendarwin.org/show_bug.cgi?id=3356 http://bugzilla.opendarwin.org/show_bug.cgi?id=6627 http://bugzilla.opendarwin.org/show_bug.cgi?id=6519 http://bugzilla.opendarwin.org/show_bug.cgi?id=6622 [speculating this will evolve into meta bug] See the description for "meta" on bugzilla.mozilla.org: https://bugzilla.mozilla.org/describekeywords.cgi Dave

3 2

DOMDocument outerHTML and NSStringEncoding
by Benoit Marchant 30 Jan '06

30 Jan '06

Hi I would like to save the outerHTML of a DOMDocument, but to do the right thing and get an NSData out of that string, I need to know what is the NSStringEncoding that should be used, which would be derived from the the charset declaration of the document, and ISOLatin1 if there are no charset declared I think. I didn't find an obvious answer looking at the documentation. Is there one ? And if not, is there a way to avoid to re create a mapping between charset declarations and NSStringEncodings ? Thanks, Benoit

2 1

Re: [webkit-dev] KWQ Graphics crossplatform
by Matteo Muratori 30 Jan '06

30 Jan '06

Just to inform webkit team, Amanith 0.3 has been released today. This release adds a full SVG 1.2-compliant enhanced alpha compositing layer, supporting all 24 compositing operations (base and extended Porter-Duff) in pure OpenGL. This makes Amanith the first opensource and crossplatform library that supports all of them in hardware (15 of 24 are available also without fragment programs). In addition a full geometry caching system has been introduced to speed up all drawing tasks. Regards, Matteo - www.amanith.org

1 0

Google Mail Buttons not respecting native look
by André-John Mas 29 Jan '06

29 Jan '06

Hi, I was about to file a bug regarding the way the "Archive", "Delete" and "Report Spam" buttons are rendered in the GMail pages by Safari. Then I decided I had better try a small test case with a page that contained: <form> <input form="submit"> <input form="button"> </form> and found the buttons rendered fine. Can anyone work out whether this is a 'bug' or something particular that Gmail is doing? I tried using the DOM inspector in Firefox (it is a bit more advanced that Safari's) to see if I could notice anything in particular, but I am not seeing it. Andre

2 3