Question

I’m developing an HTML5 app for mobile/touch device deployment and are utilising PhoneGap and only targeting the iOS platform for the moment (Webkit).

My issue occurs with the CSS3 transitions (and it happens to regular jQuery animations as well) in that Webkit tends to do some strange things. For instance, in the code available to view here (http://jsfiddle.net/lvl99/dSjcj/) when you step through the pages, going back in the sequence (i.e. from page 5 to 1) will render the animation fine, however stepping forward (i.e. from page 1 to 5) will produce inconsistent rendering, mostly on the side of ugly (remember to view it in Safari. I’ve been using Safari 5.1.7).

I was originally developing using the jQuery Mobile framework, but the key functionality that I was using was the page routing through hashtags and the transitions, but since I had these issues with the transitions, I tried developing a simpler solution to avoid any JS/CSS conflicts with jQM. Alas, it’s possible that the errors I’ve been coming up are actually Safari/Webkit related.

I originally used jQuery.animate() on the left property, to now using CSS3 transform technique similar to the jQM way of doing things, also to help with speed and test whether it would be more willing to render properly. Both haven’t worked to varying degrees of unworkingness.

Fortunately, Firefox renders everything fine. It has no problems, however Firefox isn’t the target platform and when the project is compiled within Xcode in the PhoneGap environment, it expresses the same problems that Safari has with it. When I was still using jQM with an earlier development version, Safari would render it fine (including Safari on the iPad Simulator), however UIWebView wouldn’t. This made me think that perhaps it was a Nitro JS engine issue (as in, perhaps UIWebView didn’t have the speed/power/capability to render the transition properly — I’ve attempted to transfer all animations to CSS3 to relegate the rendering operations to the GPU).

I’ve used various plugins like jQuery Transit, jQuery Animate Enhanced, and TransformJS. I also tried coding my own custom transition handler within jQM and it didn’t render properly (although it came the closest: worked in Firefox and Safari, just not in UIWebView).

I’ve had inconsistent results regarding transitioning elements with different types of content too: video, images, floated elements, multiple paragraphs are all I’ve tested. There was at some stage too issues with transitioning to/from the extreme ends of the sequence (i.e. 1 and 5), but now my issue is just that ascending page transition (i.e. 1 to 5).

I’ve spent a number of days on this, trying to address this seemingly small issue, but it’s quite integral to the user experience having slide transitions like this which are contextual based on the direction the user is moving through the app. The easiest solution is to just remove the transitions, but if there’s some way to understand what exactly Webkit/UIWebView is having trouble with, to create some solution for it. It’s no doubt related to the flicker issue jQM experiences with transitions too. A lot of the CSS fixes for those people suggested on the web didn’t work either, such as -webkit-backface-visibility: hidden and setting a default transform property -webkit-transform: rotateX(0).

Answer 1

It looks like the problem is that it’s not animating the next item from the right (when moving up 1->2->n) but rather just ‘.show()’ing it when the previous item is done animating out.

Gimme a sec to step through the .js

...

Ok, I think I know what it is, what it’s doing is, when it’s moving right-to-left (numerically up) you can’t see the new page (higher number) coming in from the left, because ‘left’ is moving from 100% -> 0% because the smaller number is moving out.

Ok, I think that’s wrong…

If you change the 100% to 92.5% in the @-webkit-keyframes slideinleft and @-moz-keyframes slideinleft declarations then it should work for you, unfortunately I can’t tell you why exactly as we (work) avoid CSS transformations as we do a lot of corporate work and so still need to support IE7+ and sometimes even 6 🙁

Note that you can try a value other than 92.5% I just wanted to find the lowest sensible value that worked for you (95% didn’t work)