or Alexis Menard in the real world.

Leave a comment

Week 26 Intel and Chromium : NaCl componentization and various bug fixes.

This update covers week 26 on Intel’s work in Chromium and Blink. Quiet past week for Intel as some of our developers were in vacations or busy inside Intel.

Blink :

  • Dongseong Hwang worked on clarifying the size of a document view (a.k.a layout size) that causes FrameView to emit the resize event.
  • Jun A Jiang fixed a crash in debug when restoring texture binding at Unit 0 in DrawingBuffer::prepareMailbox().
  • Raphael Kubo Da Costa removed some legacy pretty printers for gdb.

Chromium :

  • Pan Deng exposed receive header end time for URLRequestRedirectJob.
  • Dongseong Hwang fixed a bug in the compositor.
  • Shouqun Liu fixed a bug in Android with hidden video element.
  • Yael Aharon continued her effort to modularize NaCl (She separated NaCl switches to their own files and removed dependency of NaClHostMessageFilter code on Profiles).

Leave a comment

Week 25 Intel and Chromium : Modularization of NaCl started, memory and speed improvements.

This update covers week 25 on Intel’s work in Chromium and Blink.

Blink :

  • Mikhail Pozdnyakov landed a refactor in WTF::HashSet.
  • Mrunal Kapade landed a fix in SVG.
  • Alexis Menard decreased CSSProperty’s metadata by half in some scenarios. It makes a significant change in real world cases.
  • Dongseong Hwang fixed CanvasGradient::addColorStop to match with the spec.
  • Alexander Shalamov improved speed of KeyframeAnimation::fetchIntervalEndpointsForProperty when looking up keyframes (using a binary search rather than linear).

Chromium :

  • Yael Aharon started work to modularized NaCl (more with r207951, r207914r207806 and r206825).
  • Dongseong Hwang moved ShellWindowDelegateView from view to content namespace.
  • Caio de Oliveira Filho fixed a test case table in HistoryURLProviderTest for C++11.
  • Raphael Kubo da Costa removed dead code (r206765 and r206711).

1 Comment

Week 24 Intel and Chromium : Timeout support for XHR, unprefixing of @viewport rule and memory improvements.

This update covers week 24 on Intel’s work in Chromium and Blink.

Blink :

  • Dominik Rottsches landed timeout support for XHR.
  • Alexis Menard landed a refactor to decrease memory footprint of CSSProperty’s metadata in some scenarios.
  • Mrunal Kapade fixed a bug in SVG animations.
  • Dongseong Hwang continued cleanups in the graphic code.
  • Kenneth Rohde Christiansen removed support for device-width|height for @viewport rule following a spec change. He also unprefixed the @viewport rule of CSS Device Adaptation spec after putting the feature behing a runtime flag.
  • Jun A Jiang improved some drawing operations and test infrastructure.

Chromium :

Leave a comment

Week 23 Intel and Chromium : Passing CSS Media Queries W3C compliance tests and various bug fixes.

This update covers week 23 on Intel’s work in Chromium and Blink.

Blink :

  • Yael Aharon fixed an issue with popup menu not being aligned correctly when the view is transformed.
  • Alexis Menard removed some dead code in CSSValuePool, he also fixed very old legacy by replacing usage of integers in Blink when they should be CSSPropertyIDs or CSSValueIDs.
  • Kenneth Rohde Christiansen together with Rune (of Opera fame) spent some time and gave lot of love to the CSS Media Queries code. Kenneth removed unneeded caching and he also removed “forward compatibly syntax” support in alignment with HTML5 spec and other vendors. Kenneth and Rune are doing a great job fixing this code as now Blink is passing all W3C Media Queries compliance tests (when the last patch from Rune will land).

Chromium :

  • Yan Gu landed an optimisation in extensions regarding document.all.
  • Dongseong Hwang removed duplicated clear of ViewsDelegate::views_delegate in ~ShellViewsDelegateAura().
  • Sudarsana Nagineni fixed a memory leak after using the GTK file chooser.
  • Pan Deng exposed net::LoadTimingInfo through content_shell for testing purposes.

Leave a comment

Week 22 Intel and Chromium : Fixes in CSS Media Queries and even more cleanups.

This update covers week 22 on Intel’s work in Chromium and Blink.

Blink :

  • Mrunal Kapade fixed some styling issues when styling controls.
  • Kenneth Rohde Christiansen continued his crusade to improve Media Queries code. He fixed inconsistencies between the spec and the CSSOM in appendMedium and deleteMedium methods.
  • Alexis Menard cleaned the css/ directory with WebKit prefixed class and file names (no unprefixing, just renames). r151410 and r151426.
  • Pan Deng worked again on fixing ResourceLoadTiming resolution. He also fixed an issue in the devtools related to the canvas profile button.

Chromium :

  • Halton Huo landed a fix in the H.264 decoder.
  • Pan Deng landed the 4th step of his ResourceLoadTiming resolution work.
  • Dongseong Hwang cleaned unused code in the graphics stack.
  • Mrunal Kapade also cleaned some code related to files handling.
  • Caio Oliveira landed a refactor around the DownloadManager.

Leave a comment

Week 21 Intel and Chromium : Fixes, Refactorings and Cleanups.

This update covers week 21 on Intel’s work in Chromium and Blink. Most of the team was attending the Tizen Conference so this week was a bit more quiet.

Blink :

  • Pan Deng fixed ResourceLoadTiming resolution loss issue.
  • Alexis Menard backported from WebKit a bunch of CSS internal refactorings (r151189r151127r151124r151103r151073r151009r150925). He also fixed an ASSERT due to an hazardous cast in debug related to CSS Background and WebKit Masks.
  • Dongseong Hwang made resize-events test not flaky.
  • Rijubrata Bhaumik worked on the CSS Dialog, he also removed the compile time flag for DataList.
  • Jun A Jiang cleaned some graphics related code and implemented few extensions for WebGL (r150722).
  • Hongbo Min fixed the position of dragstart event.
  • Yael Aharon removed DISABLE_ROUNDED_CORNER_CLIPPING build flag.

Chromium :

  • Rijubrata Bhaumik disabled DataList on Android.
  • Caio Oliveira worked on the Download Manager : cleanup,  and various refactors (r201539, r201768, r202229).
  • Hongbo Min renamed ‘harddisk’ to ‘fixed’ for storage unit type in System Info API.

Leave a comment

Week 19-20 Intel and Chromium : Cleanups, Media Queries, and bug fixing.

This update covers week 19-20 (last two weeks) on Intel’s work in Chromium and Blink.

Blink :

  • Yael Aharon removed ENABLE_PAGE_POPUP build flag and replaced it by a runtime flag. Dongseong Hwang removed unused code path in GraphicsLayer.
  • Kenneth Rohde Christiansen deprecated the unofficial -webkit-transition media feature because the same can be achieved with CSS conditionals, it’s prefixed and CSS Transitions are unprefixed. He also enabled resolution media queries with fixed implementation as well as the (color-index) media query feature for non-indexed screens. He also improved consistency with other browsers.
  • Alexis Menard fixed a regression where Event#stopPropagation() does not halt bubbling for webkitTransitionEnd. He also removed getPropertyShorthand and isPropertyImplicit from the public JS APIs, these methods were only used by the inspector.
  • Shouqun Liu fixed a texture copy issue in video painting on Canvas for Chrome on Android.
  • Mikhail Pozdnyakov started a crusade to improve HashTraits : he landed an optimization for HashTraits<RefPtr<P> >::emptyValue(), he removed RetainPtrObjectHash and RetainPtrObjectHashTraits and did cleanups and fixes.
  • Various bug fixes by Yael Aharon on label tag or select.

Chromium :

  • Chunyang Dai enabled the Intel’s VTune profiling suite support in chrome.
  • Yael Aharon fixed an ASSERT when navigating to chrome://nacl.
  • Caio Oliveira got rid of DownloadItemImpl::UpdateProgress().
  • Sudarsana Nagineni fixed several memory leaks in gtk_clipboard_dump utility.


KDE 4.8 is really really nice! Congrats! Just KMix popup is confusing…


Today I upgraded to KDE 4.8, thanks to ArchLinux to provide the packages really fast. Globally I’m pretty happy with the move, more polish, more stable, it really shine! So far the only glitch I had was nepomuk resetting where to index (I on purpose disable where my dev environment is) so it started indexing like hell my development directories, freezing my machine at the very first startup. So I’d say congrats to the KDE for such a beautiful release. Nice work!

On the more down side I’m really confused about the new KMix popup when you click on the speaker icon in the systray. I felt confused about it (so were all my colleagues around) :

I have hard time to understand this popup. It seems that moving the slider to the right would decrease the volume (as it goes in direction of the mute button which is anyway against the regular left-to-right == increase for people using the LTR layout) but no *cough* it increases the volume.

Also clicking on the mute button doesn’t give feedback whether it is checked or not again leading to confusion. Any idea from some designer (Nuno, …?) on how to improve it? I would implement it myself.

UPDATE : It seems that the button is not properly updated as it should. That helps a bit to understand how it works :D.

Anyway thanks all for the work!


Why GMail looks bad with Oxygen?

I’m working on WebKit and more exactly on QtWebKit. While running rekonq to test the status of QtWebKit (eating my own dog food). GMail looked really really bad. All the check-boxes are ugly and clipped. Here is a screenshot :

I did investigate and tried to understand why it is like that. The conclusion is : It’s not a WebKit bug neither an Oxygen one, It’s a GMail bug. The column which contains the check-boxes with the drag handle has an hard-coded size of 24px. Here is the declaration of the CSS class applied to that column :

td.oZ-x3 {
text-align: left;
padding: 0 0 0 1px;
padding-top: 0px;
padding-right: 0px;
padding-bottom: 0px;
padding-left: 1px;
width: 24px;

Of course this is wrong because Oxygen check-boxes are 21px wide so of course it doesn’t let that much space for the drag handle and the margins. That explain why it’s ugly. Now you can say WebKit should be smart enough but we follow the spec I would say, firefox too and guess what with the Oxygen style support in Firefox it’s ugly the same way too :

So as a good user, I’ve reported the bug to the GMail Team and hope they can come up with a fix that would make many KDE people very happy.


My only blog post about KHTML/Webkit…

Rhaaa again this discussion show up.

I liked konqueror in KDE3 when it was just working both as web browser and file manager. In KDE4 i used dolphin because i find the UI very powerful but i never used Konqueror. The reason is (at least the last time i checked) that it doesn’t support properly gmail, doesn’t work that well with youtube (sometimes the flash plugin is not loaded), a LOT of AJAX website are borked, not rendered properly. Of course i don’t speak about official website from governments that are working/rendered properly in Arora/Firefox but not in Konqueror. This is a show stopper for Mr EVERYBODY and this is the kind of users we are targeting (not advanced users with crazy workflow using features that almost nobody uses). And yes Mr EVERYBODY sometimes use proprietary thingy, that is the way it is. I think giving away some features for a time and make 90% of people happy is a good deal.

Me as an OSS user i don’t like Firefox because it is lacking with KDE integration but i prefer to use it than tweaking KHTML with identities or to avoid having a poor surfing experience. The bad thing as well is that Kubuntu ship konqueror as default, Firefox is not installed unless you go to the package manager. What a poor experience for user when he start browsing for the first time.

I think pretty much everybody agree that we still want Konqueror but with a different engine that just works with very active maintainers. I really appreciate the work that has been done in KHTML in the past but now it’s time to focus elsewhere.

I also hear devs : “gna gna Qt Webkit lacking this and this” but i have to say :

This link can improve Qt/Webkit

I am sure that the webkit team in Oslo will be happy to review patches as well as APIs to access what is missing. (Btw Qt Webkit in 4.6 come with a full DOM API). I am also pretty sure that a KHTML developer will find quickly his way to webkit in order to help in that area. Working together seems to be a good option i have to say.

That’s all for me, i just wanted to remember the Qt open repository to make what is missing happen in order to deprecate KHTML and to stop wasting resources on a double work. Man we don’t have two times our spare time.