HOWTO: Build node.js on a CentOS machine

I recently had the “pleasure” to build node.js on a relatively vanilla CentOS machine. I was a little surprised at the number of dependencies which were missing.

So, I wanted to take a little time to document the experience. I don’t know whether this is successful for anyone else. It will depend on your initial CentOS set-up, and what yum repositories you have access to.

My first surprise during installation was that I did not have the expected C++ packages installed, even though gcc was installed and available on my path. I had to grab the following package:

sudo yum install gcc-c++

Then, node.js began complaining about openssl not being available, even though it was. What node.js meant is that it wanted the SSL libs that are packaged in openssl-devel. So…

sudo yum install openssl-devel

After that, everything went as advertised. Here’s all the install steps together:

sudo yum install gcc-c++
sudo yum install openssl-devel
sudo yum install python
sudo yum install git 
git clone git://github.com/joyent/node.git
cd node
git checkout v0.4
sudo ./configure
sudo make
sudo make install
curl http://npmjs.org/install.sh | sh

My Dream iPad

There are rumors that Apple is on the verge of updating the Macbook Air. Will it be an A5 based Macbook Air?

I have been reluctant to get an iPad so far, mainly because it doesn’t offer enough differentiation from my iPhone, and it doesn’t enable me to do anything close to the amount of activity I can achieve with my laptop. It doesn’t justify the need to have one more device.

What I want is for the Macbook Air and the iPad to come together. There are a number of hybrid notebook/tablets on the market. One of the slickest is Dell’s Inspiron Duo.

I want the best of both worlds. I want a Macbook Air laptop…

• That can run OS X and OS X applications
• That has a multi-touch screen which can pivot into a tablet form
• That can run iOS applications

This would be the ultimate road warrior machine for me as a telecommuting employee: lightweight, powerful, and diverse.

Reasons why it may not happen?

• Apple would have to reconcile iOS and Mac OS X in some way. Either by porting OS X to A5 architecture and providing an emulator for OS X apps, or by providing an iOS emulator (something they kind of already have in the iOS SDK) for OS X running on intel.
• Apple would be cannibalizing some of their device stack. They’ve put a lot of marketing effort behind the iPad in trying to convince consumers that they need a laptop/desktop, an ipad, and an iphone (plus an Apple TV and an Airport Express). 

Reasons why it may happen?

• It would be an AWESOME device. Apple enjoys building awesome devices.
• The market will force Apple to make this convergence some day. As Android improves for tablet devices, you’re going to see more hybrid form factors that make Apple’s notebook entry more irrelevant than it is already. And as notebooks become more powerful, in terms of the things a variety of users can do on them, the Macbook and Macbook Pro product lines will begin to seem like bulky dinosaurs compared to the laptop market of the future.

I don’t think we’ll see this dream device this month, but we may see an A5-based Macbook Air running OS X which would probably be the first step. 

Non-Destructive Spies in Jasmine

Over the past few months I’ve been building out our UI test framework tools at AppNexus. This gave me the opportunity to research and play with the latest automated testing tools currently available for JavaScript.

I’ve been extremely impressed with the Jasmine BDD framework which is maintained by Pivotal Labs. We’ve begun incorporating it heavily in our development processes and it is helping our code base to mature.

Why Jasmine?

I love the BDD-style interface, both for writing tests, and the reporting output. Jasmine patterns itself after RSpec. Jasmine is designed to take advantage of JavaScript’s strengths. The assertion API has a fluent interface, and anonymous functions are leveraged throughout. Nested describe blocks make it easy to encapsulate repetitive test set-up actions into limited-scope utility functions, which keeps my tests DRY.

Jasmine is extensible. It’s easy to add additional matchers, and since we utilize jQuery heavily at AppNexus, the jasmine-jquery package is extremely helpful.

Jasmine provides a lightweight spy interface (i.e. mocks, test doubles). This is an essential tool for testing our JavaScript in isolation from its dependencies. These dependencies are usually an AJAX back-end. Using Jasmine’s spies I can fake calls to a non-existent server in a test environment. This reduces the complexity of my test environment, and allows me to run tests faster.

One of the first issues I ran into with Jasmine was the destructive nature of its spy creation.

In Jasmine, you create a spy like so:

spyOn($, 'ajax');

In the above example, I’m replacing the $.ajax() method with a spy. The spy will record when $.ajax() is invoked in my application code, and it will record what input was given to $.ajax when it was called. After I’ve executed my application code I can verify that the application code correctly invoked $.ajax()

expect($.ajax).toHaveBeenCalledWith(someParams);

In JavaScript, functions can have properties of their own. I can define a function like so:

function sayHello() {
	console.log('Hello');
}
sayHello.makeGuestsComfortable = function() {
	sayHello();
	console.log('Please sit down');
}

In this example, the sayHello() function has a method called makeGuestsComfortable().

Here’s a Jasmine spec that spies on sayHello():

describe("Greeter", function() {
   it('says hello', function() {
	spyOn(window, 'sayHello');

        sayHello.makeGuestsComfortable();

	expect(window.sayHello).toHaveBeenCalled();
    });
});

But this would fail with something close to the following error: TypeError: Object function () { … } has no method ‘makeGuestsComfortable’. This is because the spyOn() call destroyed the properties of the sayHello function.

We can avoid this by wrapping our call to spyOn in a function which preserves the properties of the spied function.

describe("Greeter", function() {

  var niceSpy = function(obj, funcName) {
	var original = obj[funcName];
	var spy = spyOn(obj, funcName);

	for (var i in original) {
		if (original.hasOwnProperty(i)) spy[i] = original[i];
	} 

	return spy;
   }

   it('says hello', function() {
	niceSpy(window, 'sayHello');

        sayHello.makeGuestsComfortable();

	expect(window.sayHello).toHaveBeenCalled();
   });
});

Now the test will pass because the makeGuestsComfortable() function has been preserved.

Jasmine is a fantastic testing tool, and part of a new breed of JavaScript testing frameworks that allow software developers to keep their JavaScript code tested. I’m really excited to see this project grow and help in the development of quality software.

Getting Productive, Staying Productive with AutoFocus 4 and The Pomodoro Technique

I’m continually evaluating ways to stay productive in both my professional life as well as my personal life. Over the past two years I’ve come to rely on Mark Forster’s AutoFocus system (v4) in conjunction with the Pomodoro Technique to manage my time effectively.

Adopting these processes has helped me become more productive and helped to reduce the amount of stress and drain I experience throughout my work day.

AutoFocus v4

AutoFocus really just starts with a long list of the things I need to do. The system itself is actually a method for managing that list.

Building the List

I start the process by listing out in no particular order the tasks I need to get done. Once I’ve listed out everything I can think of I draw a line across the bottom of the list. As I think of new items to put on the list they are written below the line.

Moving through the List

I start my day by reading through each of the tasks that is above the line. I don’t stop until I’ve read through them all. I then pick one of the tasks and work on it. If I read through the whole list and cannot start on any of the tasks, then I read through the list of tasks below the line and start on one that is ready.

If I read through the list of tasks above the line and don’t work on any of them, then I highlight them. Next time I read through the list above the line I either have to work on a highlighted item or cross it off and either leave it off, or add it onto the bottom of the list.

Once all items above the line have been worked on then a new line is drawn at the bottom of the entire list and the process repeats itself.

In addition to this workflow, here are some extra tips that I’ve found extremely helpful

• Always read fully through the list before starting on a task
• After completing a task, always come back and read through the list
• Keep separate lists for work and for home/personal projects. Don’t allow one to interfere with the other
• If I find myself continually crossing off a task and re-adding it (and it’s not a recurring task like ‘Check e-mail’) then either this task is too large in its current form and needs to be broken down, or it’s not high enough priority to be on the list.

I wouldn’t use AutoFocus for team project management. And I wouldn’t use it for long-term planning. This is more of a day-to-day management tool. Items you’re unlikely to find on my list are: Buy a vacation home. Manage retirement. Start my own business. These are all long-term goals or plans that are composed of hundreds and thousands of tiny tasks which AutoFocus can help me manage.

The Pomodoro Technique

When I worked at Adchemy one of my co-workers turned me on to The Pomodoro Technique which is popular among pair programmers as a way of time-boxing programming sessions to avoid burnout. But it’s equally effective working on your own.

The technique is very simple. It simply dictates that you work for 25 minutes focused exclusively on a certain task. After that 25 minutes is up, you take a 5 minute break. This can be a 5 minute physical and/or mental break.

A good pomodoro is a little tricky to pull off. First I should note that I really only use Pomodoro for tasks at the computer. I don’t use it for gardening. Since I only use the technique at the computer, it’s nice to have an app for it. This is my favorite.

It has a lot of configurable options

And it even has some statistics to show how effective I’ve been throughout the day.

Here are some tips I recommend for a successful pomodoro

Close all programs on your computer that aren’t essential to the task at hand, especially anything that’s likely to distract you including non-essential chat windows and e-mail. I do leave my IM connection open, but I will generally sign out from non-work accounts or mark myself busy on those accounts to minimize their distraction.

E-mail is a constant attention drain. I generally close my e-mail client during a pomodoro and open it back up during a pomodoro break to catch-up.

Reset the pomodoro when you’re distracted. If you do find yourself getting distracted, either by your own wanderings, or the outside world intruding, then simply reset the pomodoro and restart later.

Honor the 5 minute break. One of the seemingly more annoying things about the Pomodoro Technique is the arbitrary time limit. What about when a pomodoro ends right when I’m in the middle of typing a line of code? Well, I find the best thing to do is finish the immediate train of thought I’m on and then start my break.

The time limit is not that arbitrary to begin with. There are studies which state the average sustained attention span of a healthy adult to be around 20-30 minutes.

Despite the annoyance of having to take a break in the midst of working on a project, I find that if I stick with a strict pomodoro schedule throughout the day I am less mentally drained at the end of it. I also notice less of a decrease in my mental capacity as the day goes on.

Use the break for mental and physical respite. The first thing I do when a pomodoro ends is I usually refill my water glass or coffee cup. Getting away from the computer and moving the legs is a good physical and mental break. I usually use the remainder of the break to peruse e-mail or my news reader before starting in on another round. Since I work from home I also occasionally use the breaks to do some brief physical exercise, a few push-ups, sit-ups or deep stretches help me maintain some semblance of decent posture and increase blood flow to the brain. Also, if the sun’s out and it’s warm I’ll usually step out my backdoor to get a little vitamin D intake.

If I ignore the pomodoro technique and work on a project until I can’t see straight my hunch is that the final several hours of that work were largely distracted and unproductive.

AF4 + Pomodoro

So how do I use these techniques together? You may have already guessed it. For most tasks on my list I devote a pomodoro to them. If I finish the task before the pomodoro is up then I start on a new task and take a break once the pomodoro is completed.

If a task requires more than a full pomodoro, then I usually put a checkmark next to the task for each pomodoro. If I spend more than four pomodoros on a task then I consider breaking up the remainder of the work into new tasks. That’s a sign that the task I have written down is more involved than I initially anticipated.

Wrap-Up

I hope you find one or both of these techniques useful in your own life. If you find tips or tricks which improve your productivity, please share them!

Keeping Things in Sync with Syncer

In almost every company I’ve ever worked as a developer I’ve had both a personal work computer as well as a development machine. The development computer may be a virtual machine sitting in a datacenter somewhere or it may be a Dell desktop tucked under my desk next to the file cabinet.

In all cases, I’ve preferred to work on my code locally and sync it with my development machine for testing. This allows me to write code in an off-line mode, and allows me to take advantage of many of the bells and whistles of my IDE.

The process of keeping files in sync between my local and remote computers has always been different, and never ideal… until now.

One of my co-workers recently clued me in to Mac OS X’s File System Events API. This is a low-level OS API that emits events corresponding to very common actions on files, like saving, opening, and closing.

I did some hunting and found this great Ruby script which takes advantage of the API. When the script is running it listens for save events in the current working directory tree and syncs the entire directory tree to a remote location using rsync.

I made some improvements to the script so that it can take command-line options like username, hostname, and a remote directory location. The updated script is available in its own repository from GitHub. Feel free to grab Syncer and play with it.

As it stands right now Syncer makes two assumptions about your operating environment:

1. You have an ssh-agent running which allows you to connect to your remote server without having to type a password every time. If you don’t already have this, read this great tutorial on ssh-agent. It will tell you everything you need to know.
2. The script takes advantage of growlnotify, a command-line interface for Growl. growlnotify is an extra in the Growl distribution. So, you need to install it separately.

If you find this script useful let me know.

View Files Beginning with a ‘.’ in Eclipse

I recently rebuilt my eclipse installation on Eclipse Helios and had set up a new workspace. In the process, I lost one of my long-time PHP Development Tools (PDT) settings where I can view files that begin with a . in the PHP development perspective file browser.

It took me a while to find the setting again. So, I thought I’d blog about it in case any others out there are having trouble finding this.

In the PHP Explorer View, click on the inverted triangle the upper-right hand corner of the pane to reveal the view’s contextual menu.

The .* resources item may be in the menu. If it is, click on it to remove the filter. If it isn’t then select the ‘Filters…’ option.

Uncheck the .* filter in the Filters… dialog window.

Now you should be able to browse and edit .htaccess files in the PDT perspective.

Bill Gates and the Public Schools Pay Structure

This week Bill Gates and Secretary of Education Arne Duncan argued for merit-based rewards for public school teachers, and criticized a system which gives automatic pay increases to teachers who have earned master’s degrees in education, regardless of how it improves teachers’ abilities.

Obviously teachers’ unions are concerned about the criticism as their task is to protect teachers’ interests and welfare. What I see missing in the debate is the question of what drives education institutions to reward master’s degree candidates in the first place?

My assumption is that there is a wide variety of young teaching candidates coming out of four year universities across the country. Some of these candidates may not have had education as their primary focus of study. A master’s degree in education would help solidify their training and make them stand out in a crowded job market.

Additionally, teachers are chronically underpaid. A two year master’s program that can be done at night and on the weekends may be a good investment for most teachers who want to get paid a reasonable wage for their work.

Gates and Duncan are probably right that for our public education system to be successful there needs to be an emphasis on merit-based performance rewards, although determining the nature and structure of such rewards is a thorny issue.

But, what is missing in this recent discussion is an interrogation of why this problem exists in the first place. We need to determine what pressures are causing school districts to reward master’s degree trained teachers more, regardless of experience or ability.

If master’s programs aren’t turning out better candidates, why are districts rewarding them? If master’s programs ARE turning out better candidates, then what is wrong with our bachelor’s degree system for teachers that is turning out sub-par candidates?

 

Google invests in off-shore wind-farming. What’s the risk to wildlife?

According to Google, the Atlantic coast is particularly suited to this project because the grid is overstretched see the sweeping 2003 Northeast blackout and it doesnt yet have access to “high-quality” wind power. It also offers relatively shallow water that extends well in the ocean. “These shallow waters make it easier to install turbines 10-15 miles offshore, meaning wind projects can take advantage of stronger winds and are virtually out-of-sight from land,” Google said.

via Google and friends scheme offshore windmill army • The Register.

It would be interesting to hear what the project planners are doing to mitigate damage to migrating seabirds and aquatic life.

AppNexus is Growing

This isn’t any secret to those of us inside the company: AppNexus rocks! We got some public recognition today with the announcement of a new round of investing totaling $50 million.

If you’d like to become part of a growing, successful tech company filled with proven winners committed to their work and their team then I suggest checking out the AppNexus web site and our available positions. We’re hiring and maybe you can be a part of the action.

Climate Change and Ocean Acidification

Here’s a great TED talk by Rob Dunbar on the impacts of human-caused CO2 emissions on ocean acidification and the impact that may have on marine systems.