Controlling Time

Disclaimer: This technique does not work outside of programming. Do not try this on your neighbours, kids or pets…

What’s wrong with time dependent tests?
It’s easy to write tests that are far too flaky and intermittent. Worse yet, a quick fix often results in putting a pause to tests that make them drag out longer and longer. Alternatively, unneeded complexity is added to try to do smart things to poll for tests to time out.

What can we do about it?
I’m all about solutions, so I’m out about to outline the secret to controlling time (in at least unit tests). Here’s a situation you might recognise. Or not. Sorry to anyone vegetarian reading this entry.

We have a class called Beef that knows when it’s past its prime using the Joda Time libraries.

package com.thekua.examples;

import org.joda.time.DateTime;

public class Beef {
	private final DateTime expiryDate;

	public Beef(DateTime expiryDate) {
		this.expiryDate = expiryDate;
	}

	public boolean isPastItsPrime() {
		DateTime now = new DateTime(); // Notice this line?
		return now.isAfter(expiryDate);
	}
}

Surprise, surprise. We also have a unit test for it:

package com.thekua.examples;

import static org.junit.Assert.assertTrue;
import org.joda.time.DateTime;
import org.junit.Test;

public class BeefTest {
	@Test
	public void shouldBePastItsPrimeWhenExpiryDateIsPast() throws Exception {
		int timeToPassForExpiry = 100;

		Beef beef = new Beef(new DateTime().plus(timeToPassForExpiry));

		Thread.sleep(timeToPassForExpiry * 2); // Sleep? Bleh...

		assertTrue(beef.isPastItsPrime());
	}
}

Step 1: Contain time (in an object of course)
The first step is to contain all the use of time concepts behind an object. Don’t even try to call this class a TimeProvider. It’s a Clock okay? (I’m sure I used to call it that in the past as well, so don’t worry!). The responsibility of the Clock is to tell us the time. Here’s what it looks like:

package com.thekua.examples;

import org.joda.time.DateTime;

public interface Clock {
	DateTime now();
}

In order to support the system working as normally, we are going to introduce the SystemClock. I sometimes call this a RealClock. It looks a bit like this:

package com.thekua.examples;

import org.joda.time.DateTime;

public class SystemClock implements Clock {
	public DateTime now() {
		return new DateTime();
	}
}

We are now going to let our Beef now depend on our Clock concept. It should now look like this:

package com.thekua.examples;

import org.joda.time.DateTime;

public class Beef {
	private final DateTime expiryDate;
	private final Clock clock;

	public Beef(DateTime expiryDate, Clock clock) {
		this.expiryDate = expiryDate;
		this.clock = clock;
	}

	public boolean isPastItsPrime() {
		DateTime now = clock.now();
		return now.isAfter(expiryDate);
	}
}

If you wanted to, the step by step refactoring would look like:

  1. Replace new DateTime() with new SystemClock().now()
  2. Replace new instance with field
  3. Instantiate new field in constructor

We’d use the RealClock in both the code that creates the Beef as well as our test. Our test should look like…

package com.thekua.examples;

import static org.junit.Assert.assertTrue;
import org.junit.Test;

public class BeefTest {
	@Test
	public void shouldBePastItsPrimeWhenExpiryDateIsPast() throws Exception {
		int timeToPassForExpiry = 100;
		SystemClock clock = new SystemClock();

		Beef beef = new Beef(clock.now().plus(timeToPassForExpiry), clock);

		Thread.sleep(timeToPassForExpiry * 2);

		assertTrue(beef.isPastItsPrime());
	}
}

Step 2: Change the flow of time in tests
Now that we have the production code dependent on an abstract notion of time, and our test still working, we now want to substitute the RealClock with another object that allows us to shift time for tests. I’m going to call it the ControlledClock. Its responsibility is to control the flow of time. For the purposes of this example, we’re only going to allow time to flow forward (and ensure tests use relative times instead of absolute). You might vary it if you needed very precise dates and times. Note the new method forwardTimeInMillis.

package com.thekua.examples;

import org.joda.time.DateTime;

public class ControlledClock implements Clock {
	private DateTime now = new DateTime();

	public DateTime now() {
		return now;
	}	

	public void forwardTimeInMillis(long milliseconds) {
		now = now.plus(milliseconds);
	}
}

Now we can use this new concept in our tests, and replace the way that we previously forwarded time (with the Thread.sleep) with our new class. Here’s what our final test looks like now:

package com.thekua.examples;

import static org.junit.Assert.assertTrue;
import org.junit.Test;

public class BeefTest {
	@Test
	public void shouldBePastItsPrimeWhenExpiryDateIsPast() throws Exception {
		int timeToPassForExpiry = 100;
		ControlledClock clock = new ControlledClock();

		Beef beef = new Beef(clock.now().plus(timeToPassForExpiry), clock);

		clock.forwardTimeInMillis(timeToPassForExpiry * 2);

		assertTrue(beef.isPastItsPrime());
	}
}

We can even further improve this test to be more specific by forwarding time by simply adding one rather than multiplying twice.

Step 3: Save time (and get some real sleep)
Although this is a pretty trivial example of a single use of time dependent tests, it shouldn’t take too much effort to introduce this concept to any classes that depend on time. Not only will you save yourself heart-ache with either flaky, broken tests, but you should also save yourself the waiting time you’d otherwise need to introduce, leading to faster test execution, and that wonderful thing of fast feedback.

Enjoy! Let me know what you thought of this by leaving a comment.

Practicing Root Cause Analysis: An example for programmers

Adding complexity in code is easy. Removing it often takes a little bit more effort. Using root cause analysis, we can sometimes remove this unneeded complexity. Here’s an example my pair and I went through today.

Me: Why is this test breaking?
Pair: Because we have to specify a different value than that one.

Me: Why do we have to a specify a different value than that one?
Pair: Because we have different values in this configuration file for tests.

Me: Why do we have different values in the configuration file for tests?
Pair: So that we can bind services to a different name?

Me: Why do we want to bind services to a different name? (since we have a contract with our clients that never change)
Pair: I’m not so sure… Let’s ask our BA.

Me (to the BA): Why do we want to bind services to a different name?
BA: I don’t know. (Thinks about it for a while…) I don’t think we would.

Now knowing that we wouldn’t ever deploy our application with a different configuration, we got to delete a whole lot of code, and an unnecessarily complicated configuration file. Awesome!

Singleton vs Single Instance

I’ve been fighting the fight against the Singleton pattern on a new code base and got me thinking about easy it is for developers to equate Singleton with Single Instance and, in my experience, more often than not, all they really want is a Single Instance. Of course, you may have proper uses for the Singleton pattern (all listed in the Gang of Four book I’m sure) yet there are many reasons why you should default to Single Instance.

Singletons often tend to be a pain because their “singleness” often makes testing a pain, with shared state persistent across different tests, and additional code to “reset” it. Worse still is the way that their global state makes them just a little bit too easy to stick into the middle of a block of code, without regard of whether or not it belongs. It often takes a while to unwind any dependencies and identify appropriate responsibilities. It’s global nature also often makes it all too tempting for it to evolve into a bit of the god object.

When Retrospectives Go Wrong: Too Many Goals

signpostsTeams run retrospectives for many different reasons. I’ve found that trying to meet too many goals in a heartbeat retrospective severely limits their effectiveness. When I prepare for retrospectives, I normally ask the sponsor (the person who asked me to facilitate) what they want to achieve. Sometimes they don’t know themselves, so it’s a useful exercise, by itself, to clarify their intended goals.

When I’ve sat in teams new to retrospectives and the goal is not made clear, people start to bring up too many different issues, and it’s difficult to resolve anything. One hour seems to be the maximum that teams are willing to set aside, and when you’re dealing with team issues, technical issues, process issues and more, all that time literally flies by. The result… nothing is improved and people get frustrated with the vehicle that brings some visibility (the retrospective).

What you can do about it
I use a rule of thumb, “One Goal Per Heartbeat Retrospective”, and make it clear at the start of the retrospective. It helps, if you’re the facilitator, to agree on what that goal is before you start, and do whatever you can beforehand to make sure everyone agrees. It helps if you’re working with a team that is already is formed (see the Tuckman Model) and you confirm this before hand.

Place the goal in front of everyone where they can see it, both as a subtle hint, and as an aid if you feel the retrospective veering off. Run retrospectives with a different goal in mind to address those that you don’t get around to talking about.

Photo of the cross signs taken from Gregkendallball’s Flickr stream under the Creative Commons licence.

Estimates degrade over time

Project management is a funny thing. It seems like once something is converted into a number, the number is often all that matters to traditional schools of project management. Plans get created, promises get made, yet little diligence goes into ensuring the number is based on sound thinking, and that the basis for that number remains correct. Maybe that also explains the state of the current banking industry, though that’s another post entirely.

Most people fail to understand that estimates have a half life, and often a very short one. Even results produced with agile estimation techniques, have an expiry date that is often ignored. Magic numbers (aka, the estimates) have an expiry date because in order for them to be even slightly reasonable, they need assumptions to be made, and it is often these assumptions that expire. Gantt charts are awful at highlighting what those assumptions are, let alone tracking when they fail to be met.

burntnumbers
Image of burnt numbers provided by Dead Scene under the Creative Commons licence

A recommended practice for agile estimation is to get the people who are going to do the work, to estimate. Put these estimates on a shelf to ripen for a year and, presto, you have a new team to which, the original estimates no longer apply. Even given the same team, set to work on something else only to return to the system, will have lost some flow, forgotten some things and need more time than if they just continued to work on the same system. Environments often change, and the underlying needs for the business change even more frequently.

What can we do about it? I refer to Lean Software Development’s principle of the Last Responsible Moment, to not only make decisions, but when collecting (and keeping) estimates. Collect estimates now if it helps you make a better decision, discarding the estimates if you choose not to immediately implement the project. You set yourself up for failure if you choose to estimate now, knowing that the project is to be shelved for “only three” months and keeping those numbers as if they are in any way relevant. If you don’t need to make a decision now, work out when you need to make the decision and defer collecting estimates until just before then.