Monday, November 9, 2015

New Adventures in Software Testing

I recently described my past years in testing, but this post is more about the future, using data to back it up.  However, we can only look into the future using the mirror of of the past.  In writing my last 70 or so articles, I have gather a good number of thoughts, and I have posted on average 1-2 in depth posts a month.  In using blogger's data, I had assumed that I was reaching an audience of substance.  However, in investigating my actual reach, it is substantially smaller than I thought.  I have two purposes for writing.  One is to collect my own thoughts, the other is to document my ideas for others to use.  Until today.

I am backing away from my editorial adventures for the time being in order to focus on another adventure.  But wait, don't leave yet.  You've not even asked what it is.  Well, here is the last article I plan to write in the near future.

Often times, I have been asked what sort of code someone should write when they want to learn to write code.  I often come back with the reply, well what sorts of problems do you want to solve?  Why write code at all?  If you have no vision for what to write, being motivated to write code can be incredibly hard.  Even if the answer is I want to write web applications to get a job, that at least provides a hint into the motivations and what sorts of study you need.  Some people have no focus at all and just want to do "something cool".  I have found these people rarely have the interest in completing projects, as cool moves so fast.

In learning to code, I started out with games, like most other people my age.  I built up a set of programs over the years, some of which I spent years developing.  Often times, these programs were designed for a particular need I experienced.  Like my advice, I work on what I find I need rather than what is cool.  Several years ago I became a land lord and found I needed a way to track my write offs.  I also had a job interview that involved PHP, so I wrote a simple tool to track write offs using PHP and Postgres.  I didn't get the job.  My PHP application was clunky and required me to record all the data by hand, yet my bank knew about these transactions.  Why couldn't I just automate the task?  I didn't enjoy working in PHP so I moved to .net and started building a windows application.

I worked for about 3 weeks and had a rough framework designed, primarily using reflections and a Excel-like control.  I started to refine it and within a month I had an application I was happy with.  I kept refining it over a few years, but it was always just meant for me.  I fixed big bugs, but it was meant to please me, so I didn't worry too much about it.

Then, just this last year, my boss, Isaac asked me a question.  He asked me if I was ever going to sell my project.  I thought a while around this and decided it would be a interesting effort and a good learning experience.  I have always wanted to go to work for myself, so why not?  I started in on my adventure.  My adventure included multiple iterations and efforts.  Many of these efforts have been running parallel over the last year.

What Features Do I Need?

I knew what I needed.  I needed one serious feature, the ability to mark items as write offs.  No one else except for the massive applications supported that sort of feature.  These mazes of confusion and madness were more complex than I wanted.  Most of the small open source applications either didn't support the feature or were equally complicated.  Furthermore, the thing I personally needed only a little is one of the big words in personal financing, budgeting.  So, like the heuristic of comparable products, I started examining other products.  I worked on developing features and creating test maps around what would need testing.  But more on that later.

This is a more complex question than you might otherwise think.  For example, what does your EULA look like?  How do you authenticate your key?  What other non-functional requirements am I missing?  I developed a large list of requirements and had to build them all out.

How Do I Start a Business?

I had to start investigating how to license software.  Even the word license is tricky, as I don't sell software, I license its usage.  That affects taxes.  You have to gather all sorts of documentation for the government to allow you to start a business.  Then there is the logo.  What about the store front?  The list goes on and on.  On the good side, I have no employees, so I that simplifies a great deal.


Let me be frank, to the few people who read this regularly.  I'm not good at marketing.  I choose privacy over announcing my existence over and over again.  To make even a personal blog big, it appears one must do lots of advertising.  Isaac suggested it takes 2 years of effort to maybe hit it big, ignoring Black Swans.  I still don't have a personal twitter handle, although I do have one for business to my chagrin.  This will likely be the hardest part for me.  I have a few tweets queued up, and several blog posts about personal finance written.  Hopefully my efforts will pay off, but one never knows.

I also am starting to learn about Google Ads and Bing Ads.  In reality, things are way more complicated than you might imagine.  Google's Ad system is crazy-complicated.  Campaigns connect to Ads which tie to keywords or to a screen scrapper Google uses.  Google Ads use a complicated bidding system where the winner doesn't pay what they bid but the max the last guy bid.  The more I learn the more I learn I know nothing.


I grabbed a free for a year AWS instance of a wiki system and started writing.  I kept writing and writing.  I have written documentation for several open source systems and have done some fairly extensive documentation in the past.  In this case, I tried tackling the problem from several different ways.  The first way I tried was to write out different features of the system.  Then I tried writing things from a different tact.  I wrote about how you can accomplish tasks.


As I built out new features, I discovered how complicated my code base started to get.  I had never done any serious regression on my work since it was built just for me.  I had added features I thought might be useful, but it was still poorly tested.  So I built up a list of all the features and major feature interactions and kept a copy in google docs.  Once my program was in a mostly stable state I started rigorously testing it.  I wish I could claim I didn't find any bugs, but that would not be true.  Instead, I found more than two dozen bugs.  Many of them were relatively minor, others were usability.  My list of needed features grew even larger.  I spent 10-20 hours a week for several weeks testing while I spent my normal time on my day job.

One interesting thing I want to note to all my fellow testers is how hard it is to do both the software development and testing.  While I certainly did testing, I found giving some time between developing the code and doing in depth testing gave me some time to forget exactly how I coded something.  That way, my bias in expectations would be gone, allowing me to note usability issues much more clearly.  I also noticed how it was hard to remain motivated.  Testing for 40 hours a week and then going home and doing more of the same is a challenge.  If you're on a death march, imagine how unmotivated a developer is to do any testing of their software.

Shipping is a Feature

Ultimately, I got my application to a stable state.   While I do still occasionally find a bug, I have not found any big bugs in weeks.  I am still doing occasional testing in that I use it for myself.  I do find value in some of the new features I have added in the past 12 months, so I spend more time in my application than I use to.  It's still fairly trim compared to many of my competitors, but I think that is a feature.  Simpler system often have their own sort of value.  I think it is ready for release.  So, on Oct 31st, I released my program, Money Pig, into the world.

I'm also stepping away from our blog for a while, while I work on Money Pig.  I shall return!


1 comment: