September 2010 - webkit-dev - lists.webkit.org

Color Management
by Igor Trindade Oliveira 09 Sep '10

09 Sep '10

Hi, Continuing color management discussion, it would be great to get advice for some of the questions I have before starting coding up. There are already some libraries to parser and convert icc profiles. One of them is little cms[1] and it's being used by many applications, specially because it is simple and works with specifications version 2 and 4. However, after looking at how WebKit deals with externals dependencies, I had the impression that they are tried to be avoided as much as possible, and due to that, some tools end up being written from scratch(like SVG, MathML, image decoders, and other parsers). Now we have two possible paths: a) use an external dependency(littlecms for example); b) write from scratch all the ICC Profile specification; What do you guys think what the best approach? [1] http://www.littlecms.com/

5 6

attached()
by Eric Seidel 09 Sep '10

09 Sep '10

I'm about to remove the last non-lazy attaching element. https://bugs.webkit.org/show_bug.cgi?id=45365 I think we can now kill the whole concept of being "attached". attached() does not necessarily mean you have a renderer. It just means that something called attach() or lazyAttach() on you. Instead, we could change all elements to mark themselves (and their parents) as needing style recalc during insertedIntoDocument(), effectively "attaching/lazyAttaching". (We could also fix the current lazyAttach() code to check to see if a parent is already marked with a child needing style resolve, and stop walking to prevent N^2 behavior.) I'm not sure if it would make sense to keep the "attached" flag around to make sure the attach logic is never run twice. -eric

4 9

Context menu testing
by Philippe Normand 09 Sep '10

09 Sep '10

Hi, I've been writing a patch to improve the context-menu of the media elements and a layout test would be nice for this. I can send a contextClick() event on the media element but then, how do i know at which position to move the mouse to click on a specific item? The mac EventSender::contextClick() returns an array of the menu items titles but the GTK EventSender doesn't do so. And still, I'd need the position/size of the menu items. Any hints? Philippe

1 0

Treat warnings as errors by QtWebKit
by Osztrogonac Csaba 08 Sep '10

08 Sep '10

Dear WebKit Community, in the past weeks we have worked on fixing warnings of QtWebKit. Now it's time to treat warnings as errors to warrant better code quality. First we enabled -Werror on x86/Linux platforms, and in the near future we are going to enable it for other Qt platforms. (ARM/Linux, x86/Windows, ...) From now any warning will break "Qt Linux Release", "Qt Linux Release minimal" buildbots and the Qt EWS bot. Please take notice of their redness. FYI: master bug of this topic - https://bugs.webkit.org/show_bug.cgi?id=43191 With best regards: Csaba Osztrogonác (ossy on #webkit, ossy {at} webkit {dot} org) University of Szeged http://webkit.sed.hu

2 1

Fwd: [Bug 43551] Resource cache is validated for the second load, while not for the third and later
by Yuzo Fujishima 08 Sep '10

08 Sep '10

Hi, webkit-dev, Please allow me to forward the following, because I'm not sure all the interested parties have subscribed to the bug. Please comment on the bug if you like. Yuzo ---------- Forwarded message ---------- From: <bugzilla-daemon(a)webkit.org> Date: Wed, Sep 8, 2010 at 1:50 PM Subject: [Bug 43551] Resource cache is validated for the second load, while not for the third and later To: yuzo(a)google.com https://bugs.webkit.org/show_bug.cgi?id=43551 --- Comment #17 from Yuzo Fujishima <yuzo(a)google.com> 2010-09-07 21:50:45 PST --- Hi, Brady, Any comments on this? I think cache validation should happen every time, as Alexey pointed out. (I'd exclude the validations at screen/print switching time, though. But this is a different story.) I could understand limiting validation to only the first load for a page, trading correctness for the performance gain, I guess. I don't understand the current behavior, though: validation for the first and the second load, no validation there after. What is the rationale behind the current behavior? -- Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug. You reported the bug. You are watching someone on the CC list of the bug.

1 0

[svg] rect with invalid stroke value
by Yuzo Fujishima 08 Sep '10

08 Sep '10

Hi, SVG gurus, I have a question regarding LayoutTests/svg/custom/invalid-fill-hex.svg . It contains: <rect width="100" height="100" fill="#ff" stroke="#0"/> From the comment there I understand that the rect should be filled with black. But how about the stroke? I guess black stroke should be rendered, because #0 is also invalid. Is it true? The current test data (expected output) doesn't contain the stroke. I'm asking this https://bugs.webkit.org/show_bug.cgi?id=24742 fixes hex color parsing and as a result, the stroke above starts to be rendered with black. I'd like to find out whether the above test (or test data) is wrong or I am making a mistake. I think we should either: Rewrite <rect width="100" height="100" fill="#ff" stroke="#0"/> to <rect width="100" height="100" fill="#ff"> or rebaseline the test data. Which do you prefer (assuming the test needs to be changed?) Yuzo

5 10

Re: [webkit-dev] Throwing SECURITY_ERR on cross-origin window.location property accesses
by Rob Barreca 07 Sep '10

07 Sep '10

On Mon, Aug 16, 2010 at 2:16 PM, Sam Weinig <sam.weinig at gmail.com> wrote: > I am not sure I agree. Does our behavior actually cause any real bugs in > the places you have tracked down? The log spam really doesn't seem like an > issue, we can remove it if we want to, but have found it useful in the > past. There is definitely a bug and a real-world use case for this for WebKit. At Sprout (http://sproutinc.com) we have a generic platform to design mobile HTML5 ads which are served inside an IFRAME. We allow designers to link elements in top window and new window. If they choose "top window", in Android 2.2 when we do "window.top.location.href = url" instead of going to the URL or just halting code execution, the browser refreshes the current top-level page. We need a way to test for the security exception or at least detect some other property of window.top and then do window.open(url) instead when that security error is trapped or we detect different-origin. -- Rob Barreca Director of Development Sprout, Inc. Mobile: 808.224.1905 Confidential and Proprietary Property of Sprout; Do not distribute. The information contained in this email is confidential. This information is intended for use only by the individual to whom it is addressed. If you are not the intended recipient, you are hereby notified that any use, dissemination, distribution or copying of this communication and its contents is strictly prohibited. If you have received this email in error, please immediately notify the sender by return email and delete this email and attachments, and destroy all copies.

3 4

Number of pending commits
by Holger Freyther 07 Sep '10

07 Sep '10

Hi, I was just seeing that we have about 84 bugs which are in the pending-commit state, on the Gtk+ bugs I clicked the patches need to be landed manually. It would be nice to get a hand landing these patches, or asking the commit queue to take care of them. thanks holger

4 3

Access to blog admin page
by Ryosuke Niwa 05 Sep '10

05 Sep '10

Hi, I wrote http://rniwa.wordpress.com/2010/08/31/pushing-inline-styles-in-webkit/ and a lot of my colleagues liked it so I'd like to post it on WebKit blog. I registered myself as rniwa on WordPress but I don't have admin rights. Could appropriate authority give me a permission to post? Ryosuke Niwa Software Engineer Google Inc.

3 2

WebKit2 SharedSecondaryProcess model
by Balazs Kelemen 04 Sep '10

04 Sep '10

Hi all! I am wondering about do you plan for the mac and win to support the shared web process model? Actually, I have played around with that for Qt. I have a working, proof of concept implementation. The visited links and the back-forward list features are broken, but apart from that the browsers are working with the shared web process. I had to rework common parts of the code for achieving that, and I would like to contribute it in small parts. I think if we want to support that model in the future than we should start to implement it as soon as possible to assure that our design fits the needs of that. Cheers! Balazs Kelemen

6 8