After the announcement that the Apple developers have turned on their SMIL support in order to pass Acid3 test, I was excited enough to download the MacOS nightly and run through the SVG animation test suite. I was pretty disappointed.

Out of 58 tests involving animation, WebKit r31344 fails 54, gets 3 of the tests partially right and only gets 1 test to completely pass. To put this into ‘acid’ terms, this means that WebKit r31344 gets a score of 5/116 when it comes to SVG animation support (as compared to Opera 9.5′s 110/116). See my top-secret scoring mechanism here.

My concern at this point is that releasing such a nonconforming SMIL implementation into the wild will just frustrate authors and users. But I suppose there isn’t much SMIL out there at the moment. Here’s one data point, at least: My menus no longer animate properly in WebKit r31344, producing ‘fluttering’ icons… Is there a way I can turn off WebKit’s native SMIL when visiting my website? Does anyone know of any other sites out there using SMIL in their UI?

Anyway, I’m very much looking forward to future Webkit releases that will improve upon the SMIL implementation – and please release the build for Windows too. And I would be remiss if I didn’t congratulate the WebKit team for releasing a downloadable application that shows 100/100 on Acid3. Hopefully this post is just a little bit of cold water in the face to those fans thinking that WebKit is somehow “perfect” :)

§445 · March 26, 2008 · Opera, Safari, Software, Technology, Web · Tags: , , , , , · [Print]

18 Comments to “Webkit Nightly: Not Smiling”

  1. Robert Blaut says:

    Jeff, Webkit nightly has an alfa quality. You shouldn’t expect perfect stability or feature complete at this stage. There is a long way before Webkit engine stabilizes and become stable engine in future version of Safari.

  2. Robert Blaut says:

    BTW. Your post crashes the latest Webkit nightly on Mac ;)

  3. Fyrd says:

    Yeah, I was kind of hoping to see full SMIL support myself, especially since Acid3 seemed to require it. Kind of like how the IE8 beta supported only enough generated content to pass Acid2 (though they’ve mentioned they’ll provide full support for that before they go final).

    Anyway, the Webkit guys will have to work on something now they pass Acid3, so hopefully more SMIL support will come soon. :)

  4. @Robert: I realize that WebKit builds are alpha quality. However, this is a clear case where a broken feature was enabled just so they could pass the test. To me this is a dangerous move that shows a disregard for web authors. I would have been less upset if WebKit could at least have passed, say, 20% of the SMIL test suite. Thankfully it is not a spec that is widely used at the moment.

  5. David says:

    Safari is the first to reach 100/100. Opera falls to 99/100.

  6. David Hyatt says:

    @Jeff: We had SMIL enabled prior to shipping Safari 3.1 in the WebKit tree. We turned it off because (quite responsibly) we did not want to ship a broken implementation. Now that 3.1 has been released, we have turned it back on and plan to continue working on it. If we ship a new version of Safari that passes Acid 3 but has horrible SMIL support, then you’d have grounds for complaint. Until then, indignation over WebKit’s mainline development tree having partially implemented or experimental features turned on is unwarranted.

  7. @David: I do appreciate that the WebKit team is starting to work on SMIL, and I’ve said so – I’m very excited about the possibility. Kudos to you and your team. WebKit is now the second-best browser to support SVG natively.

    However, I also think I have a right to be disappointed in the quality of SMIL support, since I use that very feature on my website. And the Safari blog encourages users to download nightly builds. If I want to fix my website for WebKit nightly users that takes extra work. In other words, your enabling SMIL does more harm than good to my website at the moment (since I am using script to work around lack of SMIL support).

    You could have kept the SMIL implementation turned on in a private branch until some semi-decent level of support was reached (I chose an arbitrary 20% above). Or you could have continued to use the special compile flag. The spirit of the Acid test is feature compliance, not “barely compliant”.

  8. David Hyatt says:

    Nightly builds are not meant for end users to use. They are development builds. If nightly builds don’t work with your Web site, then that is not a big deal. You shouldn’t fix your Web site just to accommodate development builds.

  9. huxley says:

    @Jeff

    Individuals like yourself who download and test the latest nightly builds of WebKit help us to create the highest quality product.

    For that, we thank you.

    How can I help?

    The number one way you can help is to report issues you encounter while using WebKit.

    Report the problems, you can add your report to Bugzilla or comment on an existing ticket.

  10. Rob says:

    Changing your site to suit bugs in a nightly will wear you out. Changing to meet the expected behaviour of the next release based on nightlies just makes you an early adopter. Which could also wear you out but at least is more rational.

    @David: I don’t follow webkit closely enough to know your build process but I really don’t get why you’d turn off a feature to ship a release. I mean, you must be working on other features that won’t ship in this version, shouldn’t that concurrent development be handled by branching?

  11. masklinn says:

    > If I want to fix my website for WebKit nightly users that takes extra work.

    Which is exactly why you shouldn’t bother: add a little warning maybe, report bugs to the Webkit team, but do *not* crappify your code to try to patch the ever-changing bugs of a nightly. It’s not worth it, and only fairly hardcore geeks bother with nightlies.

  12. Ok, I’ve opened two of the simplest possible SMIL bugs I could think of: 18187 and 18188. It would be great if someone could point me to a simple SMIL example that works in WebKit.

    On the other hand, does anyone have a good source for either the WebKit of Apple logo in SVG? I’d like to use it on my blog (linking back to their website, of course). I haven’t received a response to my email query from Dave or Maciej yet…

  13. AG says:

    (Offtopic)

    Id like to know Wingogi results for your SVG test suite.

  14. Test says:

    >Wingogi results for your SVG test suite

    There are big problems with fonts in public windows WinGogi build, it’s impossible to make correct test.

  15. “Does anyone know of other sites using SMIL for UI?”

    You know I am.

    http://deerring.com

    Sincerely;

    James

  16. Marc Nothrop says:

    Jeff, you may be interested to know that the bug that allowed Acid3 tests 75 and 76 to pass (17077) has been closed, and work has moved to:

    18375: Make SVG animation work, “…where ‘work’ is defined as passing reasonable amount of SVG animation test suite so that remaining issues can be tracked as individual bugs.”

    Hopefully SVG/SMIL support will see rapid improvement in WebKit.

  17. Update: As Marc Northrop points out, both of my bugs have now been resolved with the submission of Bug 18375