…download the 0.1 Early Access Preview of Android Studio
Here’s my quick notes on a git workflow to create a branch, merge and clean up:
Create a branch named “x”:
git checkout -b x
git branch x
git push origin x
git push -u origin x
git checkout x
git push --set-upstream origin x
Merge a branch “x” back into master:
git checkout master
git merge x --no-ff
Note that this creates a merge commit to make it easier to find where branches are merged into master.
Clean up branch “x”:
git branch -d x
git push --delete origin x
Update [2013-05-15]: Used the “-u” option with git push so we have one less line when creating a branch. Thanks to @jdriscoll for that tip.
Update [2013-07-11]: Used the “-b” option with git checkout so we have one less line when creating a branch. Thanks to @bobz44 for that tip.
I’ve been starting to use calabash-android, which is a way to run cucumber tests on Android. It requires Ruby Gems and Xcode Command Line Tools on Mac, which installed fine. However, when I ran the first sample test, then I noticed the app did not start. I looked at the output and saw that this error showed up:
App did not start (RuntimeError)
I looked around for the solution and did not find, so I tried a few things. It turns out to be an easy fix. In your Android app’s manifest, aka AndroidManifest.xml, include the line:
<uses-permission android:name="android.permission.INTERNET" />
inside of the manifest tag, after your application tag.
Why does this work? The reason is because Calabash-Android (and Cucumber in general) uses HTTP to communicate between the host computer and the target application. Since Android apps do not have permission to communicate via HTTP by default, the test fails to start.
So you’re being a good Android developer, using JUnit 3 tests [Sidenote: JUnit 4 does not play well with the JUnit runner in Eclipse for Android currently] and using Mockito to create mocks to make sure you’re focusing on the class that you’re testing. However, you run into a java.lang.ExceptionInitializerError when running your tests. This is pictured below. What should you do?
Put those 3 JARs linked above in a “libs” directory in your test project.
The root cause of this is that the classes that Mockito makes needs to interact with dex, which is what Android uses to create its classes. This can be done with Dexmaker and the Dexmaker/Mockito integration. Once you add in those 3 JARs into your libs directory, just do the refresh/clean project dance and then you should be good to go.
Special thanks to Jesse Wilson (@jessewilson) for making DexMaker and integrating Mockito with Android!
If you have ever had to implement the Parcelable interface for one of your Android classes, you will find that it is tedious. It is especially tedious the more fields your class has. So usually you don’t do it unless you have to pass it in an Intent to an Activity or as a Bundle of arguments to a Fragment.
Enter Parcelabler, which is a web-based tool that creates Parcelable implementations for you. Just copy the portion of your class which has the fields into the Code textbox and then press the Build button. I’ve had good results with it, although it is best at primitive fields. I usually edit the class in a text editor like Sublime Text to make it so that only the class and fields are entered.
Thanks to Dallas Gutauckis for creating this helpful tool!
Hey look – I earned the android badge on StackOverflow!
android badge: “Earned 100 upvotes for at least 20 answers in the android tag.”
I’ve recently been having problems with my Nexus 7 not booting up after it completely runs out of battery. Specifically – even if I have charged it overnight, if the power is off and I click the power button, then it does not turn on!
I’m not sure if this is related to 4.2.1, but this problem didn’t seem to occur with 4.1.1. I’ve got the 8GB version of the Nexus 7.
To solve it, press and held on to the power button until it starts up (for approximately 30 seconds.)
Note that Asus has a support page which mentions this titled Nexus 7 won’t start up. I agree with their recommendations, but I’d add that you’ll probably need to press and hold it for longer than what they recommend (around 30 seconds vs their recommended 15).
Update: after a quick search, a thread on Android Enthusiasts titled Nexus 7 wont boot after complete discharge corroborates my findings.
I attended AnDevCon III in May 2012 as my first Android-related conference, about a year and a half into my Android experience. I have attended many developer conferences before so I thought it’d be interesting to compare it to those. I’ll be attending AnDevCon IV later this week, so obviously I was pleased with what I experienced, but I wanted to note down what my thoughts were from that first experience.
I’ve been to a variety of conferences:
- JavaOne, one of the biggest but also one of the most “corporate”
- WWDC, the best place to interact with Apple engineers and other Mac and iOS developers
- EclipseCon, which was focused on a single open source project, Eclipse
- No Fluff Just Stuff, sort of an anti-JavaOne, where the speakers are all practitioners
- C4, sort of an anti-WWDC, where the speakers were mostly Mac indies
AnDevCon feels a lot like EclipseCon. Partly because this is because it was held at the Hyatt Regency in Burlingame, where I once attended EclipseCon 2005. More substantively, a lot of companies participate in the conference, which gives it an interesting and varied vibe. This is compared to single-vendor conferences like WWDC or Google I/O, where you get the perspective of only one company mainly.
Google engineers and evangelists do play well with AnDevCon, however. They present some of the sessions, which are prominently marked on the scheduled as “Google CLASS”. Kirill Grouchnikov‘s Responsive Mobile Design in Practice was particularly insightful in the project that I was working on at the time. I had seen his slides, but they did not make as much sense as when I had heard him speak live about responsive design. Now that we can nest fragments within fragments as of the 4.2 SDK, I wonder what he would say about that.
There are a large amount of sponsors. You might think that this is a bad thing if you have attended a conference that had lots of sponsored talks. However, the organizers seem to be aware of this and have helpfully marked all the talks that as “Sponsored by XYZ”. I actually liked some of those, particularly the ones sponsored by Intel, where I learned more about the HAXM-accelerated emulator and Sony, where they taught us about the Sony SmartWatch SDK.
I normally don’t like Exhibit Halls at conferences. Sure, you can get a lot of free goodies. OK, I admit it – half my wardrobe is conference t-shirts. But the booths usually are kind of boring. However, at AnDevCon, there were a lot of interesting exhibits. I’m not sure if it is because the industry is expanding so rapidly or because I just like gadgets, but there were some things I hadn’t seen before there, like the Epson Android-powered glasses (sort of a bulky early release of Google Glass) or the Qualcomm developer boards, which are entire Android systems on a large PCB.
I met a lot of interesting people, especially at Square’s Android Dessert Bash. Everyone was fired up about Android and there were a lot of different perspectives there. There were folks from the phone manufacturers like HTC, independent consultants, authors, trainers, regular developers and also platforms like Nook. It was pretty friendly atmosphere but I didn’t see as many night parties as I have seen at other events, but maybe it’s because I didn’t know as many people.
I honestly did not know what to expect and only knew one person who was going to the event, Dave Smith, who gives talks on Android accessory development. I was pleasantly surprised with my overall experience. The only con is that the official conference t-shirt was sponsored – in my opinion, it’d be nicer if it just had AnDevCon on it. However, Square gave away one that was more subtle and had all the different dessert icons on it, which was my favorite.
I’m looking forward to AnDevCon IV, happening later this week, which promises to be even bigger and better, featuring keynotes from Amazon, Google and Facebook.
I use Eclipse for Android development and one of the issues I have come across is that sometimes Command-C on Mac (Control-C on Windows) doesn’t work properly. I will copy, but paste will paste what was previously in the buffer, not what the latest I wanted.
My previous workaround was to hit Command-C twice which always seems to work. This is a pain, however.
Fortunately I came across this blog post: Ctrl+C to copy does not always work in Eclipse which describes a simple fix: Unbind the copy command in Eclipse. Note that this also applies to the Android Developer Tools distribution, since it is built on Eclipse.
The workaround is to unbind the Copy command in Eclipse. The reason this works is that the native Copy will be used instead.
To do so, go into the menu:
ADT or Eclipse
Search for Copy
Press the button “Unbind Command”
Here are some screenshots:
Before (Copy is bound to a shortcut key):
After (Copy command no longer bound to a shortcut key):
I have an LG Optimus V that I use for testing. Its battery had run down and when I plugged it back in, the buttons would light up but the screen was black. So what to do with a blank screen?
Unplug from wall. Remove the back cover. Remove the battery. Put the battery back. Replace cover. Plug it back in. The screen is back!