Unit testing in Android: Design Patterns to the rescue!

Recently, I have been playing around with Android. Unit-testing your application on Android is slightly cumbersome: the class that you test cannot use any class from AndroidSDK, e.g. View or Activity.

As we will see in the example below, in many cases you can achieve a reasonable testability of your code.

We wish to create a simple game. The view should show playground in which there is several balls. For now we are interested in a snippet that would allow us moving those balls freely around playground.

To that end we create classes Playground and Ball. Playground serves as a facade for accessing Balls. Ball is an object that knows it's position, size and identity.

A naive implementation would look like this: As noted above such code is not unit-testable. A way to work around that is to separate the infrastracture classes (PlaygroundView) from model (PlaygroundModel) by a bridge (Hand). Now the implementation Hand is Android independent and we can finally test it :-)

Understanding HTTP protocol by using telnet

After few incentives during GeeCON and other conversations and conferences I decided to dive into HTTP details finally. I.e. to read the HTTP protocol in the raw form of RFC2616

Actually I wanted to understand the GET, POST, PUT, DELETE methods but before I got there I found it pretty amusing to hack a bit using telnet. And that post is exactly about this: hacking HTTP using telnet.

Prerequisites

I assume that you have access to:

• telnet from command line
• Chrome web browser. In principle you could use Firefox with Firebug plugin installed but I will focus on Chrome

First hack

On the command line write:

 
telnet most-recently-used.blogspot.com http

The google server should answer with something like

Trying 209.85.148.132...
Connected to blogspot.l.google.com.
Escape character is '^]'.

and then paste this magic into the terminal

GET /2012/05/geecon-2012-review.html HTTP/1.1
Host: most-recently-used.blogspot.com

And tap Return/Enter once or twice.

This will request a google server to fetch for us a document called /2012/05/geecon-2012-review.html

The HTTP protocol specifies that the response starts with header like the one you can see at the very beginning of the server's response:

HTTP/1.1 200 OK
Content-Type: text/html; charset=UTF-8
Set-Cookie: blogger_TID=136dec288ddf5a27; HttpOnly
Expires: Sat, 26 May 2012 15:34:12 GMT
Date: Sat, 26 May 2012 15:34:12 GMT
Cache-Control: private, max-age=0
Last-Modified: Sat, 26 May 2012 15:34:11 GMT
ETag: "cf88fe06-51da-4158-aa3f-9d374ae09058"
X-Content-Type-Options: nosniff
X-XSS-Protection: 1; mode=block
Server: GSE
Transfer-Encoding: chunked

What follows is already a html file of an article of my blog.

Go ahead and check what happends if you:

• tell the google server that you speak the predecessor of HTTP/1.1 protocol, namely HTTP/1.0 (diff or vimdiff is your friend here)
• make a typo in the host name
• change GET method to HEAD, PUT, DELETE or anything else

Second hack

Now, I always wondered what is the difference between refreshing a page using Ctrl+F5 and F5. Let's

telnet www.w3.org http

GET /Protocols/rfc2616/rfc2616-sec9.html HTTP/1.1
Host: www.w3.org
If-Modified-Since: Wed, 01 Sep 2004 13:24:52 GMT

Which means: get this document for us if it was modified since the date given. Of course it wasn't modified and therefore the answer is a bare HTTP header saying:

HTTP/1.1 304 Not Modified
Date: Sat, 26 May 2012 15:58:52 GMT
Server: Apache/2
Connection: close
ETag: "40d7-3e3073913b100"
Expires: Sat, 26 May 2012 21:58:52 GMT
Cache-Control: max-age=21600

Browser details

Now that we know that browsers do no magic but communicate with servers using a simple protocol we may wish to inspect what requests they actually do.

• open Chrome
• open new tab
• launch Javascript Console
• in the console open Network tab
• direct your browser's tab to www.w3.org/Protocols/rfc2616/rfc2616-sec9.html

Have a closer look on the rfc2616.html document and you will see similar things that we made telnet to write for us. The browser puts much more informaction into the header. See what's the difference if you refresh the page using F5 and Ctrl+F5. Did you notice the following lines in the header?

If-None-Match: "40d7-3e3073913b100"
If-Modified-Since: Wed, 01 Sep 2004 13:24:52 GMT

The reason for that is: if we press F5, the browser believes it has the right version of the file to be rendered and asks the server if the file changed since a date specified. If it didn't only a short response message is sent back. Otherwise whole html file would be sent back. If, however, Ctrl+F5 is pressed there is none of the lines above and therefero the server eagerly serves html content right away.

Summary

Now go ahead and use Chrome's Javascript Console (Network tab) to see more examples of requests done in your name by the browser!

GeeCON 2012: review

A week ago I took part in a Java conference held in Poznań (Poland) called GeeCON . I didn't spend much time considering if it is worth coming to Poznań for that conference because my Java guru told me that it is and on the other hand I saw Bruce Eckel among the keynote speakers.

And so I went, unprepared. As it often turns out in such situations, it wasn't Bruce that rocked my world (though his speeches were cool too and what OpenSpaces even cooler) but other speakers I had the opportunity to listen to and funnily enough, not necessarily those the most known.

I will definitely remember Kevlin Henney who gave a hilarious and yet deep and insightful speech about code coolness. Or should I say art of code ? Lesson learned: read code of the libraries you use. Even if it's as heavy as Hibernate...

An other name to be remembered: Sam Newman from ThoughtWorks. Essentially, he spoke about Software as a Service architectures and what are key benefits of adapting this approach instead of the traditional one.

Tomasz Kaczanowski ("good tests, bad tests") and Keith Braithwaite (on TDD) are other speakers that made great impression on me. Or maybe I just agree with their thesis and that's why I liked their talks?

As of Bruce Eckel, I didn't really enjoy his keynote speech on thursday. On the other hand I heard that his "Reinventing business" speech was marvelous. Well, I can imagine because it turns out that Bruce Eckel invests recently a lot of time into the reinventing business project and OpenSpaces at the same time. You can find this talk as conducted at Heroku (though I heard that Bruce did better in Poznań).

As to the organization and stuff -- no complains whatsoever. Everything went smoothly, on time and I felt well informed. If I were to sum up the conference in one sentence: best spent 150 euro for a conference ever! Seriously, I liked each talk I listened. Both in terms of quality of the speech and contents.

And then on Saturday Geecon OpenSpaces happend. I fell in love with the idea of self-organizing conference from the very beginning but seeing it in action was a better yet feeling (if you want more on OpenSpaces than you should see this short video). The best part for me was that we actually talked about stuff that we deal with every day and not on stuff that speakers came to talk about. Not to say that the opportunity to actually meet someone (by which I mean: exchange business cards and talk) was much greater than on regular conference.

The other best part: I supose that organizing this event was 30 times cheaper in terms of human resources then GeeCON itself. Which basically means that everyone can do it in almost no-time.

After the OpenSpaces I must seriously rethink if going to regular conferences is worth my money? But GeeCON rules so I will go there next year for sure. The other thing I'm sure about: I will not go to Kraków's JDD conference any more.

Spring IoC with annotations only and without any configuration!

The purpose of this post is to show on a simple example how to create an application using extensively Spring IoC and yet configured only by means of Java5 annotations.

I believe a short and concise example is worth more then theoretical divagation. Therefore I'll strive to be brief!

Prerequisites

* A subset of these Spring libraries: spring-core, spring-aop, spring-beans, spring-tx
Obtainable from the Spring's homepage

* Java 1.5

Main

Let's go directly to the code:

import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.context.annotation.AnnotationConfigApplicationContext;
import org.springframework.stereotype.Service;

@Service
public class Main 
{
    
    @Autowired
    private Runnable proxy;
    
    public void run(){
        proxy.run();        
    }
    
    public static void main(String[] args) {
        AnnotationConfigApplicationContext context = new AnnotationConfigApplicationContext("the.package.to.be.scanned");
        ((Main) context.getBean(Main.class)).run();
    }
}

Now, the most important line for us is "new AnnotationConfigApplicationContext("the.package.to.be.scanned"). It says:

Check every class in the package "the.package.to.be.scanned" and look for those which are annotated by @Service, @Component, @Controller or @Repository and register them as beans. Once it's done, instantiate all the not-lazy singletons and autowire any dependencies needed.

Of course, registered but not instantiated beans may be instantiated later. Note, that by default beans are not-lazy singletons.

DummyThread

Up there you could see that we autowired an instance of a class implementing the Runnable interface (it's not a good idea in a production code; we do it for the sake of simplicity here!)

import org.springframework.stereotype.Service;

@Service
public class DummyThread implements Runnable{

    public void run() {
        System.out.println("run()");
    }
}

And yes, you should expect "run()" on the conosle as a result of this simple app!

AFactory

Now, you may ask what if I want to register a factory? Is it possible using annotations only? Well... it is!

@Service
public class ConfWithProfiling extends AFactory{
    
    @Bean(name="proxy", autowire=Autowire.BY_NAME) public Runnable proxy(){
        return new DummyThread();
    }
}

The magic is done by @Bean annotation, it says:

This method knows how to create a bean named "proxy".

Note also the @Service annotation. We need that for our scanner in Main class. Without it, Spring context wouldn't register AFactory class and therefore, the "proxy" bean.

Sum up

It is possible to use Spring without any configuration file whatsoever. You must take into account however, that without a @Configuration class or .xml configuration file it might be hard to understand what happends under the hood of Spring. In particular, which classes get instantiated and in what order! More on that later.

SpringIDE 2.6.0 dla Eclipse - problem z org.apache.xerces.impl.dv.DVFactoryException

W SpringIDE 2.6.0 jest bug, na który łatę trzeba pobrać stąd, używając zwykłego mechanizmu instalacji w Eclipse (Help -> Install New Software).

http://dist.springsource.com/release/TOOLS/patches/e3.6

Po instalacji na moim Eclipse 3.6.0 Helios SpringIDE ruszył :-)

Mylyn + JIRA -- instalacja i konfiguracja connectora w Eclipse

Celem tego posta jest możliwie zwięzłe przedstawienie od początku do końca procesu instalacji JIRA-connectora dla Mylyn w Eclipse.

Założenia wstępne

Używasz Eclipse 3.6.
Używasz JIRA v 4.0.* lub 4.1.*

Instalacja

Install Available Software w Eclipse;
Work with: http://update.atlassian.com/atlassian-eclipse-plugin/e3.6

W oknie wyboru oprogramowania wybierz do instalacji tylko to, co jest Required lub Recommended (w szczególności: Atlassian Connector), resztę możesz śmiało zignorować.

Konfiguracja

Otwórz widok "Task List" (Alt+Shift+Q, K)



Rozwiń menu z "bialej kartki" (guzik najbardziej po lewej) i wybierz "add repository".



Wybierz JIRA i Next.



W polu Server wpisz adres do Twojej JIRE (np. https://twoja.firma.com/jira), uzupełnij pozostałe dane i wybierz Finish (możesz też sprawdzić poprawność wprowadzonych danych używając "Validate Settings").

Mierzenie czasu

Window --> Preferences --> Tasks (lub Mylyn):
zaptaszkuj:
"Track time spent within Eclipse when a task is active"
"Enable inactivity timeouts"

To tyle podstawowej konfiguracji.


Ważne skróty klawiszowe :
Ctrl+Shift+Alt+right arrow
Ctrl+Spacja --> "górne propozycje" pochodzą z kontekstu.


Czytaj więcej:
Mylyn 2.0, part I: Integrated task management
Mylyn 2.0, part II: Automated context management

Java -- tricks

Drobne tricki niespotykane w C++:

1. inicjalizacja atrybutów klasy: jak wiadaomo bywa to problematyczne. Ale jest magiczny trick

class Spoon {
  static int i;
  static {
    i = 47;
  }
}

który można również zastosować dla niestatycznych atrybutów.

2. metoda finalize() jest wywoływana _przed_ zebraniem przez Garbage Collector. Nie można nią więc czyścić elementu z ekranu (trzeba to zrobić manualnie), ale można np. sprawdzić czy obiekt został wyczyszczony (i znaleźć tym sposobem ew. pluskwiaka
).

Źródło: Thinking in Java 3rd edition