Monday, November 25, 2013

Sample Android Project: Movies

There's a sample application that I worked on recently with talented designer Chris Arvin for a talk we gave at DroidCon UK 2013.  I also repurposed part of it for the talk about animations I gave at AnDevCon a few weeks later.

The app is open source and you can check it out here:

You can get the APK here:

The interactions required a bit of interesting engineering so I wanted to discuss some of it here.

ViewPager and Decor

One of the key interactions we wanted was the app to feel like a ViewPager so that it was an interaction users were familiar with except that the result isn't quite the same - instead of paging, content would slide out from underneath other content.  This is not something ViewPager is inherently designed to do.

There were two options open to me.  One was to rip out all of ViewPager's event handling code and make my own custom View, which was not an appealing prospect.  The other was to find some way to manipulate ViewPager itself.

It turns out there is an interface ViewPager finds special called Decor.  If a View implements Decor, then ViewPager treats it as a special View that can remain on top of it the ViewPager.  This is normally for implementing your own tabs, but in this case I imported my own copy of ViewPager (since Decor is hidden normally) and made my entire UI a Decor View.

It's a neat trick that would work for any app that wants ViewPager event handling without paging Views, though I think if you wanted a less hacky solution you'd write your own event code.

Custom Views Everywhere

I've gone from avoiding custom Views like the plague to fully embracing them.

The key realization is that custom Views do not need to handle every situation.  If you're looking at framework custom Views as reference you will be overwhelmed quickly.  For example, during measurement framework Views have to handle WRAP_CONTENT, MATCH_PARENT, and everything in between; but if you know your View is always going to be MATCH_PARENT you can greatly simplify all your code.

The movies sample app is a pile of custom Views.  First there's SlidingRevealViewGroup, which generically shows one View slide out from another.  On top of that is built MovieRowView, which has the specific Views that we want to display.  Then there's SlidingPairView, which is a set of two SlidingRevealViewGroups that creates the side-by-side effect seen in the app.

I also needed a few other custom Views to shore up some other issues.  CenteringRelativeLayout just adjusts the film cover so that, as it shrinks in size, it still looks centered.  SlidingListView was required for performance; we needed to manipulate the rows directly when sliding, instead of constantly notifying of data changed.

Performance Tricks

The coolest part of the whole app is the slide: how the cover becomes smaller and the content slides out from underneath.  All of this was achieved through translation of Views.  I took advantage of the fact that ViewGroups don't render their children outside of their own bounds.  By translating content I could hide them outside the clip bounds.

When a slide starts, it throws practically everything into hardware layers then just slides Views left/right.  Moving around pre-rendered content is fast and as a result the paging is silky smooth.

Nowadays I'm a huge fan of those basic properties of Views (translation, scale, rotation and alpha).  They're the core reason to support ICS+ only; with these properties you can take your interactions to the next level.


The drawback of the solution I came up with is that it creates a ton of overdraw.  This hurts performance, especially (as far as I can tell) when rendering a new row while scrolling up/down.  I sacrificed scrolling performance for paging performance.  Perhaps there is a way to achieve both, but I'm pretty much done with this sample for now.

The rounded corners could've been implemented in a much better fashion than an overlaid Drawable, but with the limited time before the presentation I had, I could not come up with a better solution.

Thursday, November 21, 2013

Is findViewById() Slow?

One of the presenters at Droidcon UK 2013 made the point that View.findViewById() has been unjustly villainized and that the ViewHolder pattern recommended by Google is unnecessary.  After hearing that I studied the code (in View and ViewGroup) and did not observe any heinous performance issues contained therein.  It's a pretty simple recursive lookup that walks the View hierarchy for an id.

As this article points out, findViewById() is necessarily slower than ViewHolder because it's O(n) vs. O(1).  But that doesn't mean findViewById() is slow.  If the operation is sufficiently fast and n is low then it doesn't matter if it's O(n).  Who cares if I have to do a 1 nanosecond operation a thousand times?

My hypothesis is that findViewById() is fast enough as to be negligible.  With that in mind, I cooked up a sample application that tests findViewById():

You can dynamically create a hierarchy of Views, either adding to the depth or the number of children at each node.  When you hit "run" it searches for the furthest away id a number of times, then averages the time taken.  I ran the test on my Nexus 1, to somewhat recreate the situation described when Google explicitly recommended the ViewHolder pattern three years ago.

Here's the results from a test of many possible depth/views per node values:

As  can be seen from the chart, the time taken to use findViewById() is roughly linear with the number of Views in the hierarchy.  What's key here, though, is that for a low number of Views it barely takes any time at all - usually a matter of several microseconds.  And remember, this is on a rather old phone, the Nexus 1.

That said, if you were to call findViewById() many, many times it could cause a problem.  Admittedly this part of the post is going to be a bit more hand-wavy than the rest, but: I tried out scrolling on a ListView on my Nexus 1 - there is an upper bound to how fast I can scroll and I can't seem to break 60 getViews() per second.  That means that getView() is called once per 16ms frame.  Given that we're measuring findViewById() in microseconds, even calling it a dozen times (with a reasonable number of Views) shouldn't cause performance issues.

So my conclusion is that findViewById() is harmless in most practical circumstances.  I'll probably still use ViewHolder to avoid casting Views all the time, but performance will not be the main purpose anymore.

Disclaimer: I'm not a performance expert so I suspect there's something I'm missing (especially since I'm going directly against Google advice here).  If someone knows of a missing link let me know!

Monday, November 18, 2013

Android Talk Slides from 2013

I'm done with talk season!  In case you missed them, here's the slides:

Crafting Unique, Delightful Apps

I Can Animate and So Can You

I suggest downloading the Keynote files for the presentations.  They animate!