Re: [webkit-dev] mouseenter and mouseleave events
On Apr 22, 2013, at 3:32 PM, Allan Sandfeld Jensen k...@carewolf.com wrote: On Wednesday 17 April 2013, Maciej Stachowiak wrote: Seems like a reasonable feature to add. Would you mind also reviewing the patch in question then or help me find a reviewer with the right knowledge? Most of the patch is boiler plate work. The central part is how and when the events are dispatched and confined to additional responsibility of updateHoverActiveState. I took a quick look at it and concluded that it would require some study of existing code to be confident of its correctness. I encourage reviewers with knowledge of DOM or events to take a look, as I will likely not have a chance for at least a few days. Regards, Maciej ___ webkit-dev mailing list webkit-dev@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] mouseenter and mouseleave events
On Wednesday 17 April 2013, Maciej Stachowiak wrote: Seems like a reasonable feature to add. Would you mind also reviewing the patch in question then or help me find a reviewer with the right knowledge? Most of the patch is boiler plate work. The central part is how and when the events are dispatched and confined to additional responsibility of updateHoverActiveState. Regards `Allan ___ webkit-dev mailing list webkit-dev@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] mouseenter and mouseleave events
On Tuesday 16 April 2013, you wrote: Hi On Tue, Apr 16, 2013 at 5:08 AM, Allan Sandfeld Jensen k...@carewolf.comwrote: I have recently uploaded a new patch to https://bugs.webkit.org/show_bug.cgi?id=18930 to implement mouseenter and mouseleave events. This sounds nice, but shouldn't you finish the support for image-rendering first? You added the code but it is not enabled anywhere. There is not one test covering the feature. That is because it is not meant to be enabled any time soon. The CSS4 definitions are only there to guide our implementation so it behaves in a forward compatible way. The problem is we enabled a small part of image- rendering back when it was a proposal for CSS3 (and because it is part of SVG CSS). The standardization effort is in CSS4 and will probably take a few years before it is ready to be enabled on any ports. Testing that animations are now in high quality when using the SVG CSS image- rendering: optimizeQuality, is unfortunately not possible with our existing layout tests. I could add a manual test though, if you think that is better. I also have a patch to add the rest of the ifdefs for the feature so the related code is more easy to localize, but I don't like adding the feature to the build systems since it shouldn't really be enabled. Not that I am going to defend the CSS4 parts being in the tree that hard. Now that I revisit the idea, I am less sure it was a good idea. It just seemed better than correcting an unstandardised feature without a guideline. Best regards. `Allan ___ webkit-dev mailing list webkit-dev@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] mouseenter and mouseleave events
On Wed, Apr 17, 2013 at 12:46 AM, Allan Sandfeld Jensen k...@carewolf.comwrote: That is because it is not meant to be enabled any time soon. The CSS4 definitions are only there to guide our implementation so it behaves in a forward compatible way. The problem is we enabled a small part of image- rendering back when it was a proposal for CSS3 (and because it is part of SVG CSS). The standardization effort is in CSS4 and will probably take a few years before it is ready to be enabled on any ports. If there is no plan to enable this anywhere for years. This should not be in the source today. Testing that animations are now in high quality when using the SVG CSS image- rendering: optimizeQuality, is unfortunately not possible with our existing layout tests. I could add a manual test though, if you think that is better. I also have a patch to add the rest of the ifdefs for the feature so the related code is more easy to localize, but I don't like adding the feature to the build systems since it shouldn't really be enabled. There should be at a minimum test for the parsing of the values. Regarding testing the outcome itself...if the current infrastructure is inadequate, you should look what can be done to fix that. Creating test coverage is part of adding a feature, you should not dump code somewhere and hope someone else will pick it up and ship it. I would prefer you finish the feature before moving on the next one. Benjamin ___ webkit-dev mailing list webkit-dev@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] mouseenter and mouseleave events
On Wednesday 17 April 2013, Benjamin Poulain wrote: On Wed, Apr 17, 2013 at 12:46 AM, Allan Sandfeld Jensen k...@carewolf.comwrote: That is because it is not meant to be enabled any time soon. The CSS4 definitions are only there to guide our implementation so it behaves in a forward compatible way. The problem is we enabled a small part of image- rendering back when it was a proposal for CSS3 (and because it is part of SVG CSS). The standardization effort is in CSS4 and will probably take a few years before it is ready to be enabled on any ports. If there is no plan to enable this anywhere for years. This should not be in the source today. I can understand that. I will see how to best clean it up, or possibly roll it back. I would prefer you finish the feature before moving on the next one. When you are not working for Apple, it takes months to get a review of a patch. Just the former mouseenter/mouseleave event patch had been sitting in the review queue since September. If I worked on one thing at a time I would have months at a time were I would be sitting idle. Sorry I can't do that. `Allan ___ webkit-dev mailing list webkit-dev@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-dev
[webkit-dev] mouseenter and mouseleave events
Hi webkit-dev I have recently uploaded a new patch to https://bugs.webkit.org/show_bug.cgi?id=18930 to implement mouseenter and mouseleave events. These events are part of DOM3 http://www.w3.org/TR/DOM-Level-3-Events/#events-mouseevents and supported by MSIE, Mozilla and pre-blink Opera. They are also listed as issues jQuery needs to work around in WebKit https://bugs.webkit.org/show_bug.cgi?id=110007 Previously they have not been implemented because they were not considered important, and could cause performance regression by issuing more events on every mouse-move that may not even be used by many websites. After I refactored hit-testing in the fall and especially moved the update of hover/active state out of hit-testing, it is now possible to dispatch these events where the hover state is updated, which makes the logic of dispatching mouseenter and mouseleave events correctly simple. It is also possible to only do one check for possible capturing listeners before issuing any mouseenter or mouseleave events for a mousemove. This is what the new patch does. It does incur an additional overhead by checking for the existence of capturing listeners, but the check is of lower magnitude than existing iterations over the tree, and should keep performance comparable, and is in all circumstances faster than issuing new events that fall for deaf ears. Are there more objections for the support of mouseenter and mouseleave events in WebKit? Best Regards `Allan ___ webkit-dev mailing list webkit-dev@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] mouseenter and mouseleave events
Hi On Tue, Apr 16, 2013 at 5:08 AM, Allan Sandfeld Jensen k...@carewolf.comwrote: I have recently uploaded a new patch to https://bugs.webkit.org/show_bug.cgi?id=18930 to implement mouseenter and mouseleave events. This sounds nice, but shouldn't you finish the support for image-rendering first? You added the code but it is not enabled anywhere. There is not one test covering the feature. Benjamin ___ webkit-dev mailing list webkit-dev@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-dev
Re: [webkit-dev] mouseenter and mouseleave events
On Apr 16, 2013, at 5:08 AM, Allan Sandfeld Jensen k...@carewolf.com wrote: Hi webkit-dev I have recently uploaded a new patch to https://bugs.webkit.org/show_bug.cgi?id=18930 to implement mouseenter and mouseleave events. These events are part of DOM3 http://www.w3.org/TR/DOM-Level-3-Events/#events-mouseevents and supported by MSIE, Mozilla and pre-blink Opera. They are also listed as issues jQuery needs to work around in WebKit https://bugs.webkit.org/show_bug.cgi?id=110007 Previously they have not been implemented because they were not considered important, and could cause performance regression by issuing more events on every mouse-move that may not even be used by many websites. After I refactored hit-testing in the fall and especially moved the update of hover/active state out of hit-testing, it is now possible to dispatch these events where the hover state is updated, which makes the logic of dispatching mouseenter and mouseleave events correctly simple. It is also possible to only do one check for possible capturing listeners before issuing any mouseenter or mouseleave events for a mousemove. This is what the new patch does. It does incur an additional overhead by checking for the existence of capturing listeners, but the check is of lower magnitude than existing iterations over the tree, and should keep performance comparable, and is in all circumstances faster than issuing new events that fall for deaf ears. Are there more objections for the support of mouseenter and mouseleave events in WebKit? Seems like a reasonable feature to add. - Maciej ___ webkit-dev mailing list webkit-dev@lists.webkit.org https://lists.webkit.org/mailman/listinfo/webkit-dev
