Let’s just get right into it, shall we?
What is an abstraction?
Before we can talk about leaky abstractions, and why they are bad, let’s define what exactly an abstraction is.
An abstraction is when we take a complicated thing and we make it simpler by generalizing it based on the way we are using or talking about that thing.
We do this all the time. Our brains work by creating abstractions so that we can deal with vast amounts of information.
We don’t say “my Dell Inspiron 2676 with 2 GB of RAM and a 512 GB hard drive, serial number 12398ASD.” Instead we say simply “my computer.”
In programming, abstractions let us think at higher levels. We create abstractions when we give a bunch of lines of code that get the username and password out of a textbox and log the user into our system, a name; login. And we use abstractions when we utilize APIs or write to the “file system.”
Without abstractions we’d have a really hard time managing the complexity of any software system. Without abstractions, it would be pretty difficult to write any kind of non-trivial application.
If you want to feel what it is like to program with fewer abstractions, try serving up a web page with assembly code. If you want to feel what it is like to program with almost no abstractions, do the same exercise, but write your code in machine code… 0s and 1s.
How do abstractions “leak?”
This means the details of the lower level concept that are being abstracted away are leaking up through the higher level concept.
Ever get a cryptic error message when an application you are using crashes? That is a leaky abstraction.
“Memory location 00FFE133 could not be written to” is just bad news. It means that in order to understand what is going wrong with your application, you now have to understand the inner workings of the application. Your abstraction has effectively “leaked.”
Abstractions leak whenever you are forced to pull back the curtains and see the gnomes running in hamster wheels spinning the cogs.
Why “leaking” is bad
I’m often surprised when I hear someone say “I like how you didn’t hide the details of what is really going on” about some code or some technology.
Not being able to hide the details of what is really going on points to an inability to create the proper abstraction, which in my opinion, should be water tight.
It is not a good thing that we can see what is really going on. No, it is actually a very bad thing, because it forces us to think at the higher level some of the time, but have to switch gears to drop down to the lower level some of the time.
Leaky abstractions prevent us from being able to comfortably drive the kids to school; we have to constantly keep looking in our rear view mirror and checking to see if those rowdy kids who insist on sitting at the back of the bus are causing trouble again.
A good abstraction hides the details so well that we don’t ever have to think about them. A good abstraction is like a sturdy step on a ladder, it allows us to build another step above us and climb up it. A leaky abstraction is like a wobbly rotted wood rung; put too much weight on it and you’ll fall through—it traps you at the current step—you can’t go any higher.
Shouldn’t we understand how things really work?
No! Stop already! Enough with the guilt trips and unnecessary burdens. You are defeating the purpose of abstractions and stunting further growth.
You only “need” to know the details if you want or need to know the details.
I know that many people want to know how things really work, myself included for many things, but in many cases I don’t need to understand how things really work.
For example, I can buy and use a watch without understanding how the watch works. It really is better that way. I don’t need those details floating around in my head. I get no benefit from knowing what is going on inside my watch and frankly, I don’t want to know.
When I buy a car, I just want to drive it. I don’t want to remember when to rotate tires, check fluids, etc. Waste of my time. I don’t want to know how the engine works or anything about it besides I have to put gas in, turn a key, and press a few pedals to make the thing go. I know I could save money servicing my own car, but you know what? I’d rather pay money than have to do it myself. And since that is the case, as long as I can find someone I trust to work on my car, I don’t want that abstraction to leak.
The same with machine code and compilers. Yes, it is interesting to understand how memory mapping and CPUs and logic gates work and how my C# code gets compiled into MSIL which is interpreted by the CLR and so forth and so on, until it ends up as electrons operating on registers in my CPU. And I learned about how much of that works, because I was interested, but let’s be completely honest here. That abstraction is really good. It is so good and so air-tight, that I don’t need to know all of that crap in order to write a web app. If I did, I’d probably find another profession, because my brain would explode every time I tried to write the simplest piece of code.
By the way, if you do want to understand what is happening behind many of those abstractions, I highly recommend Charles Petzold’s book: “Code: The Hidden Language of Computer Hardware and Software”. (This is the kind of book you read for “fun,” not because you need to know this information.)
Leaking abstractions in technology
Too many technologies and frameworks today are leaky, and this is bad– really bad.
Want an example? Whatever happened to ORM (Object Relational Mapping?)
Nothing. Nothing happened. It was invented, oh, like thousands of years ago and while lots of other technologies and frameworks grew up and got jobs and left the nest, ORM is still a fat unemployed balding 30 year old man playing WOW all day in his mum and dad’s basement.
Don’t get me wrong. I use ORMS; I sometimes like to think I like ORMS. But, I can’t use them well because I have to understand SQL and what is going on in my relational databases anyway. What good is that?!
Seriously. This is a completely failed abstraction. The leakiness of it makes it actually more work to learn an ORM framework, because in learning one you must also learn hairy details of SQL and understand what N+1 means and so on. You can’t just fire up your ORM and put in the Konami code, you have to blow the cartridge out and fiddle with the reset button half a dozen times. Useless.
I could go on and on with examples of leaky abstractions in technology and how they are holding us back, but I think you get the point by now.
Leaky abstractions are bad. They don’t do anything good for us besides require us to know more and add more points where things can go wrong. We’ve got to seal up these abstractions, if we want to be able to keep climbing the software development ladder. You can’t keep building a tower if you have to worry about every single block beneath you. You eventually get to a point where you can’t keep it all straight and your tower comes crashing to the floor.
Leaky abstractions and you
So, what can you do about it?
You really have to make a choice. Am I going to hide the details so that no one needs to know about them, or am I going to expose the whole thing? Either make a really good abstraction that doesn’t require someone using your code to understand what is going on underneath that abstraction or don’t attempt to make an abstraction at all.
It takes more effort to create good abstractions. Sometimes the effort required is orders of magnitude greater. But, if other people are going to use your abstraction, to save the mental burden of understanding what is happening beneath it is going to go a long way in making up for your extra effort.
10 Steps to learn anything quickly course
By the way, I just released by 100% 10 Steps to Learn Anything course. If you want free access to it, just signup to my newsletter here. It is a totally free video course and optional workbook that reveals the process I use to quickly learn technologies.
I have another new course on Pluralsight, check out:
I am very excited to finally get this course out. Many viewers have been asking me to make a comprehensive course that actually shows you how to build a real automation framework, and I finally did it in this course.
I reveal all my tricks and tips here and tell you exactly what I have done in the past to build successful automation frameworks. It was a lot of fun to create this course and it brought me back to the glory days of creating automated tests. (This is some serious fun!)
Anyway, check out the course and let me know what you think.
Here is the official course description:
Learning how to use a tool like Selenium to create automated tests is not enough to be successful with an automation effort. You also need to know how to build an automation framework that can support creating tests that are not so fragile that they constantly break. This is the real key to success in any automation effort.
In this course, I will reveal every secret I know from creating several successful automation frameworks and consulting on the creation of others. I will show you exactly, step-by-step how to create your own automation framework and I will explain to you the reasoning behind everything we are doing, so you can apply what you learn to your own framework.
We’ll start off this course by going over the basics of automation and talking about why it is so important as well as discuss some of the common reasons for success and failure.
Then, I’ll take you into the architecture of an automation framework and show you why you need to pay careful attention to the structure of any framework you build and give you some of the underlying design principles I use when creating an automation framework.
After that we’ll be ready to start creating a framework. In the next few modules, I’ll show you how to create a real automation framework capable of automating the WordPress blogging platform administrative console. We’ll start off by creating smoke tests and using those smoke tests to build out our initial framework.
Then, we’ll expand the capabilities of our framework as we create more tests and learn how to use techniques like dummy data generators to make our tests as simple and easy to read as possible.
Finally, I’ll take you through some best practices and tips that cover topics like scaling out, working in Agile environments and other important issues you are likely to face. If you are responsible for an automation project for a web application or you want to start using automation, you’ll definitely want to check this course out.
(Update: My view is starting to change from when I first wrote this article. Thanks to numerous people who have pointed out many things I overlooked and have helped me to understand what responsive design really is and what goals it is trying to achieve, I am getting a better understanding of why it may indeed not be a waste of time. There are still some valid points that I brought up in this article, but I don’t stand by my original blanket statement that it is a waste of time. It especially seems like responsive design may be worth the “trouble” for ecommerce sites where you are more much likely to make the sale if someone can easily purchase and browser your products from a mobile device. Thanks everyone who has been giving me great advice and resources for better understanding responsive design and for mostly doing it in a “nice” way.)
Response design seems like a good idea.
At first look, it makes sense. Why wouldn’t you want your same website to scale down to a mobile device and display differently?
For example, there are many WordPress themes now that offer responsive design baked right in. If you are using one of those themes, you might as well take advantage of the hard work someone else has put into making the design responsive.
But, if you are creating a new website or you are considering converting your existing website to a responsive design, I’d suggest you think twice before making the plunge.
The extra work of responsive design
The biggest problem with responsive design is the extra work involved. Sure, there are frameworks and tools to make responsive design easier, but no matter how good those frameworks and tools get, to make a site adapt from the desktop to a mobile device, it is still going to require significant work.
If you have a really simple site with just a few pages or pages that pretty much all use the same layout, but just differ in content (like a blog for example,) then the work might not be that much, so go ahead, make it responsive.
But, if you have a website of any considerable size, you’ll probably find that responsive design is going to add a non-trivial amount of extra work to your project.
It actually turns out the adding responsive design to a web site can add more work than creating a separate desktop and mobile site, because you are creating a user interface that morphs or adapts to the mobile form factor–which, by the way, is ever-changing.
The problem with one-size-fits-all
Consider this. Which is more difficult to create: a car that turns into a motorcycle, or a car and a motorcycle?
Here is another one. Which is more difficult to create: a multi-function printer or a printer and a scanner?
When you try to make a product that can do multiple things, the complexity of creating that product increases exponentially. It is almost always easier to create two separate products that do two different functions or have different form factors, than it is to create an all-in-one product that is able to do many things.
Besides the increased difficulty, trying to create products that do more than one thing usually results in them not doing either function that great.
I’m not saying you can’t design a great responsive website that serves desktop clients and mobile clients wonderfully and makes their eyes light up with joy every time they see your beautiful site. I’m just saying you won’t.
I’d also rather not be handcuffed to my original design because changing anything will incur the cost of making changes to the reactive layout as well, that scales down to mobile.
The reuse fallacy
What are you really trying to do by creating a reactive design, instead of just creating a desktop version and mobile version of the site? Reuse some HTML and CSS?
When you think about it, it doesn’t make much sense. I’m all for code reuse, but responsive design is like putting ifdefs all over your code base. Are you really saving much time and effort by reusing your front end code for both platforms?
It seems to me, that if you have to have your site display differently on a mobile device you are better off just forgetting about trying to reuse the HTML markup and CSS, and instead focus on reusing the backend code for both the mobile and desktop versions of your site; that is where you’ll actually get the biggest bang for your buck.
Again, I don’t know about you, but I’d much rather maintain two front end codebases that are simple than one monstrous complicated front end codebase.
It is good enough already?
Most desktop websites actually look fine on a mobile device without any reactive design. To me it makes much more sense, in most cases, to focus on just creating a desktop design that looks good on mobile or even a mobile design the looks good on the desktop as well.
Sure, responsive design is pretty cool, but do you really need it?
It is the browser’s responsibility anyway
Aside from the practical reasons I’ve talked about so far, there is another major reason why responsive design might just be a waste of time anyway: this should all be the browser’s responsibility, not the web developer’s.
This is even already starting to take place. Consider the special input types in HTML5, like the number input type. On a mobile device it brings up the numerical keyboard in mobile browsers.
I predict as more and more websites are viewed from mobile devices, the types of things that web designers are doing to make desktop sites responsive right now will be incorporated directly into the HTML markup and the browser itself. So, creating a responsive design manually might end up being a big waste of time.
I’ll leave you with one final point, then I’ll get off my responsive design soapbox.
We are in a time of constant flux right now
Things are changing rapidly. Mobile devices are getting bigger screens and higher resolutions and becoming much more powerful. While two years ago it was a pain to visit a desktop site from a mobile browser, it is much less painful now. With my new Galaxy S4, I hardly feel the difference at all and I even find myself specifically requesting the desktop version of a mobile site from time to time.
The trend says that mobile browsing will eventually take over desktop browsing as the main source of visitors to websites. If this becomes reality—which I see no reason to doubt—responsive design won’t make much sense at all. In that case, it will make much more sense to build websites optimized for mobile and not really worry so much about desktop.
If you are going to do responsive design, at least get a good book on the subject. There are many subpar books floating around, because responsive design is so hot right now.
What do you think? Am I off my rocker? If you have a convincing argument as to why responsive design is worth the effort or why it isn’t as much extra effort as I have painted it to be, I’d love to hear it. Let me know in the comments below.
The goal of software development is to solve problems.
At its heart, software development is really about solving problems through automation.
Many times we have the tendency to make software development about creating solutions. It may seem that these are the same thing, but there is a subtle difference.
The difference is the focus
When we are trying to solve problems, we are obsessed with the question of why. It is only by understanding why that we can know what to build.
When we are trying to build a solution, we are obsessed with the question of what. We want to know what to build so that we can build it.
This is fairly natural for software developers since the what is something we can control. Any fairly skilled software developer can build any what you can describe to them. Just like any skilled carpenter can build any type of wooden furniture you desire.
The difference is that software development is about more than just building things. A carpenter doesn’t have to focus so much on the why, to build you a piece of furniture. But a truly skilled carpenter will ask you why. A skilled carpenter might make a house call to the project to custom build an entertainment center suited to your needs.
A truly skilled craftsman of any craft, needs to know why.
It is odd that we constantly seem to neglect the why when building software (at least I do), primarily because we think we don’t have time for the why, when in the end solving the why is the only thing that really matters.
Do you really want a carpenter to build you a perfectly crafted entertainment center with plenty of custom shelves for all your music CDs when you don’t own any music CDs?
Why focusing on what doesn’t work
Consider for a moment what would happen if your GPS system stopped showing your route and the directions you were going to take ahead of time, but instead abruptly told you to turn right when you were supposed to turn.
One of the early GPS systems I used for navigation did exactly this. It was very frustrating! I would constantly shout at it that telling me what to do exactly when I am supposed to do it wasn’t any help. I needed some forewarning so I could switch lanes and prepare to turn.
Also, have you ever asked your GPS the question “why are you taking me this way?”
Focusing only on the what results in split-second decisions and avoidable mistakes. When you are focusing on the what you are not thinking, just doing.
Contrast this with a good GPS navigation system.
What makes a good navigation system good?
I’ve found that good systems will alert me far in advance to what my next move is. This gives me time to switch lanes or to have a better understanding of the bigger picture of the trip.
Now this allegory won’t carry us very far in the software development world. You don’t really need to know why your GPS system is taking you a particular route in order to get there, but it does demonstrate how focusing only on the immediate what can lead you astray.
Perhaps a more direct analogy related to software development is outsourcing.
If you’ve ever worked on a software development project that has had a portion of its work outsourced, you may have felt the pain of what happens when oftentimes perfectly competent programmers focus completely on the what and might not even have the slightest clue about the why.
I’m not trying to knock outsourcing, and I’m not even specifically talking about outsourcing to different countries. (The same problems exist in outsourcing whatever the conditions are.)
Oftentimes outsourced projects are treated like a set of blueprints that need to be built. The poor developers working on the outsourced project tend to get thrown under the bus when they build something based off just the what and can’t anticipate the why.
Why focusing on why is better
Focusing on the why is focusing on providing holistic cures to software development problems rather than treating each symptom individually.
Just like you’d like to have a doctor set a broken leg and put it in a cast so that it can be wholly cured, rather than give you some pain medicine to manage the pain, give you crutches to help you walk, and send you to therapy where you can learn to live without the use of your leg, your customers would probably rather you built something that actually is focused on solving their problem, not something that makes their symptoms easier to live with.
If we start building software from the what that is given to us, versus the why, we are at a distinct disadvantage because we only have the ability to treat the symptoms of a problem that we don’t even understand.
Focusing on the why is vitally important because it helps us to better think about and design the what.
Reuse is also much more likely to be present on why focused solutions than what focused ones.
Which software component is more likely to be adopted for reuse: a component that addresses a specific set of account balancing steps or a component that addresses the general problem your customer has of doing accounting?
Why focused solutions are much more likely to be shared across many customers, while what focused solutions are often very specific to a single customer.
Let’s not forget one of the most important reasons why it is important to focus on the why.
Missing the why is costly!
You can build the what someone is asking for, and you can do it perfectly, but at the end of the day if what you built doesn’t solve the problem (the why,) it’s going to have to be redone or thrown out completely.
You may think you are saving time by starting with the what, and you may actually build your solution faster by not taking time for everyone on the project to understand the why, but chances are you’ll build the wrong thing, and when you do, you’ll lose any time benefit you may have accrued.
It is very hard for a team to focus on the what. Focusing on the what tends towards segregating the responsibility of the team. The team members end up only having ownership for their own parts of the solution, rather than solving the problem at whole. So, when a team that focuses on the what ends up with a problem, finger pointing a shirking of responsibility inevitably ensue.
By instead focusing the team on the why, every member of that team becomes responsible for solving the problem rather than solving their part of the problem.
The most compelling reason to focus on the why?
For many problems, just having a thorough understanding of the problem takes you 90% of the way to the answer. I have spent countless hours trying to design and architect a clever solution to a problem without having a really good grasp of the problem that I’m trying to solve, only to find that 20 minutes spent actually fully understanding the why of the problem resulted in the answer becoming completely apparent.
How to know if your focusing on what
I’ve got a pretty good idea of how to know when I am focusing on what instead of why. Not because of any special wisdom on my part. No, it is because I fall down in this area quite a bit myself.
I’ve learned the hard way, and I am still learning how to identify when my focus needs to shift from what to why.
Here are some simple questions that you may find helpful to evaluate your own situation. I’d be glad to hear any others that you can can think of or use routinely yourself.
- Do you’re backlogs, stories or general work items state a solution instead of a problem?
- Do you know how to actually use the thing you are building? (Whoops, I am so guilty of this one so many times.)
- Can you define in simple terms the problem you are trying to solve? (Einstein said “If you can’t explain it simply, you don’t understand it well enough.”)
- Can you clearly define why the what your are building is needed and what purpose it will solve?
- Can you walk through the manual steps of the process you are automating, understanding each step and why it is necessary?
Taking the time to focus on the why instead of jumping into the what can be difficult to get accustomed to, but it is worth the effort. You’ll find much more satisfaction in your work when you understand the purpose it is going to serve, the true problem you have solved. In addition your customers will thank you!
What slows down the development of software?
Think about this question for a bit. Why is it that as most software evolves it gets harder and harder to add features and improve its structure?
Why is it that tasks that would have at one point been simple are now difficult and complex?
Why is it that teams that should be doing better over time seem to get worse?
Don’t feel bad if you don’t have an immediate answer to those questions. Most software practitioners don’t. They are hard questions after all.
If we knew all the answers, we wouldn’t really have these problems to begin with.
Regardless though, you’ll find many managers, business owners, customers and even software developers themselves looking for the answers to these questions, but often looking in the wrong place.
Process is almost always the first to be blamed. It stands to reason that a degradation of process or problems with the software development process are slowing things down.
Often there is some merit to this proposition, but I’ve found that it is often not the root cause. If your team is not sitting idle and the work that is important is being prioritized, chances are your process is not slowing you down.
Now don’t get me wrong here. I am not saying that these are the only two important aspects to judge a software development process, but I am saying that if generally your team is working hard on important stuff most of the time, you can’t magically improve process to the point of increasing productivity to any considerable order of magnitude. (In most cases.)
Often questions are asked like:
- Should we pair program or not pair program?
- Should we be using Scrum instead of Kanban?
- Should we be changing the way we define a backlog?
- Should we use t-shirt sizes or story points or make all backlogs the same size?
- Do we need more developers or more business analysts?
- Do we need to organize the team differently?
Now these are all great questions that every software project should constantly evaluate and ask themselves, but I’ve found over and over again that there is often a bigger problem staring us in the face that often gets ignored.
Let’s do a little experiment.
Forget about process. Forget about Scrum and backlogs and story points and everything else for a moment.
You are a developer. You have a task to implement some feature in the code base. No one else is around, there is no process, you just need to get this work done.
It might help to think about a feature you recently implemented or one that you are working on now. The important thing with this experiment is that I want to take away all the other “stuff” that isn’t related directly to designing and implementing that feature in the code base.
You will likely come to one of these conclusions:
1. The feature is easy to implement, you can do it quickly and know where to go and what to modify.
Good! That means you don’t really have a problem.
2. It is unclear what to do. You aren’t sure exactly what you are supposed to implement and how it fits into the way the system will be used.
In this case, you may actually have somewhat of a process problem. Your work needs to be more clearly defined before you begin on it. It may be that you just need to ask more questions. It may be that half baked ideas are ending up in your pipeline and someone needs to do a bit more thinking and legwork, before asking a developer to work on them.
3. Its hard to change the code. You’ve got to really dig into multiple areas and ask many questions about how things are working or are intended to work before you can make any changes.
This is the most likely case. Actually usually a combination of 2 and 3. And they both share a common problem—the code and system do not have a design or have departed from that design.
I find time and time again with most software systems experiencing a slow down in feature development turnaround that the problem is the code itself and the system has lost touch with its original design.
You only find this problem in successful companies though, because…
Sometimes you need to run with your shoelaces untied
I’ve consulted for several startups that eventually failed. There was one thing in common with those startups and many other startups in general—they had a well maintained and cared for codebase.
I’ve seen the best designs and best code in failed startups.
This seems a bit contradictory, I know, but let me explain.
The problem is that often these startups with pristine and well maintained code don’t make it to market fast enough. They are basically making sure their shoes laces are nicely tied as they stroll down the block carefully judging each step before it is taken.
What happens is they have the best designed and most maintainable product, but it either doesn’t get out there fast enough and the competition comes in with some VB6 app that two caffeine fueled not-really-programmers-but-I-learned-a-bit-of-code developers wrote overnight or they don’t actually build what the customer wants, because they don’t iterate quick enough.
Now am I saying that you need to write crap code with no design and ship it or you will fail?
Am I saying that you can’t start a company with good software development practices and a clean well maintainable codebase and succeed?
No, but what I am saying is that a majority of companies that are successful are the ones that put the focus on customers and getting the product out there first and software second.
In other words if you look at 10 successful companies over 5 years old and look at their codebase, 9 of them might have some pretty crappy or non-existent architecture and a system that departed pretty far from the original design.
Didn’t you say something about pulling up your pants?
Ok, so where am I driving at with all this?
Time for an analogy.
So these companies that are winning and surviving past year 5, they are usually running. They are running fast, but in the process of running their shoelaces come untied.
They might not even notice the shoelaces are untied until the first few times they step on one and trip. Regardless they keep running. And to some degree, this is good, this is what makes them succeed when some of their failed competitors do take the time to tie their shoelaces, but those competitors end up getting far behind in the race.
The problem comes pretty close to after that 5 year mark, when they want to take things to the next level. All this time they have been running with those shoelaces untied and they have learned to do this kind of wobble run where they occasionally trip on a shoe lace, but they try to keep their legs far enough apart to not actually step on a shoelace.
It slows them down a bit, but they are still running. Still adding those features fast and furious.
After some time though, their pants start to fall down. They don’t really have time to stop running and pull up those pants, so as they are running those pants slip further down.
Now they are really running funny. At this point they are putting forth the effort of running, but the shoelaces and pants are just too much, they are moving quite slow. An old woman with ankle weights power walks past them, but they can’t stop now to tie the shoelaces and pull up those pants, because they have to make up for the time they lost earlier when the pants first fell down.
At this point they start looking for ways to fix the problem without slowing down and pulling up the pants. At this point they try running different ways. They try skipping. Someone gets the idea that they need more legs.
I think you get the idea.
What they really need to do at this point though is…
Stop running, tie your shoes and pull up your pants!
Hopefully you’ve figured out that this analogy is what happens to a mature system’s code base and overall architecture.
Over time when you are running so fast, your system ends up getting its shoelaces undone, which slows you down a little. Soon, your system’s pants start to fall down and then you really start to slow down.
It gets worse and worse until you are moving so slow you are actually moving backwards.
Unfortunately, I don’t have a magic answer. If you’ve gotten the artificial speed boost you can gain from neglecting overall system design and architecture, you have to pay the piper and redesign that system and refactor it back into an architecture.
This might be a complete rewrite, it might be a concerted effort to get things back on track. But, regardless it is going to require you to stop running. (Have you ever tried to tie your shoelaces while running?)
Don’t feel bad, you didn’t do anything wrong. You survived where others who were too careful failed. Just don’t ignore the fact that your pants are at your ankles and you are tripping over every step, do something about it!
I worked on an interesting problem this week that might have looked like I was running around in circles if you just looked at my SVN commits.
The problem, and the eventual solution, reminded me of an important part of software development—of building anything really.
Sometimes you must tear it down!
No really, sometimes you build a structure only to tear it down the very next day.
It’s not a mistake. It is intentional and productive and if you are not doing it, you very well might be making a real mistake.
That is highly illogical
Imagine for a moment that you are tasked with the job of repairing the outside walls of a 2 story building.
There are of course many ways you could go about doing something like this.
Use a ladder and just reach the part of the wall the ladder allows you to. Then move that ladder, repeating the process as needed to complete the repair to the entire wall.
Try to lower yourself down from different windows to reach as much of the wall as possible.
Tear down the entire building and rebuild the building and walls.
I am sure there are plenty of other methods besides what I listed here.
Yet, a very simple approach would be to build a scaffolding.
A scaffolding is basically a temporary construction used just to help repair or initially build a building which is built for the very purpose of eventually being torn down.
Software is different, we don’t contend with physics!
You are absolutely right!
We contend with a much more powerful force…
Conceptually anything you can create in software could be created without any kind of software scaffolding. Unfortunately though, the complexities of the logic of a system and our abilities as humans to only contain so much of it in our brains impose a very real limitation on our ability to even see what the final structure of a metaphysical thing like software should be.
So what am I saying here then?
I’m just saying that it sometimes helps to remember that you can temporarily change some code or design in a way that you know will not be permanent to get you to a place in the codebase where your head is above the clouds and you can look down and see the problem a little better.
Just like there are physical ways to repair a wall on a 2 story building that don’t involve wasting any materials by building something that will be torn down, there are ways to do build software without ever building scaffoldings, but in either case it is not the most efficient use of time or effort.
Don’t be afraid to take a blind hop of faith
Source control is awesome!
Source control lets us make changes to a code base, track those changes and revert them if needed.
Working on a code base without source control is like crawling into a small crevice in an underground cave that has just enough room to fit your shoulders—you are pretty sure you can go forward, but you don’t know if you can crawl back out.
But, with source control, it is like walking around a wide open cave charting your trail on a map as you go. You can get back to where you want to any time.
So as long as you have source control on your side, and especially if you are using a distributed source control where you can make local commits all day long, you don’t have to be afraid to step out on a software ledge and see where things go.
Oftentimes I will encounter some problem in code that I know needs to be fixed and I know what is wrong with it, but I just don’t know how to fix it.
When I encounter a problem like this, I will often have an idea of at least one step I could take that should take me in the direction I believe the code needs to go.
A real example
To give you a real example, I was working recently on some code that involved creating a lot for a product. (Think about the lot number you might see on a box of Aspirin that identifies what batch of ingredients it came from and what machine produced it when.)
Up to the point in which I had been working with my teammate on refactoring this code, there had only been only one way to produce the lots.
We were adding some code that would add an additional way to create the lots and the labels for those lots. We also knew that there would be more additional ways in the future that would need to be added.
Well, we knew we wanted to have a generic way of specifying how lots should be created and labeled, but we didn’t know what the code that would do that would look like.
We could have rewritten the entire lot handling code and made it generic, but it was quite complex code and that would be equivalent to tearing down a building to fix a wall.
It was apparent there were some parts of the existing code that were specific to the way that the current lots were being produced, so we knew those parts of the code had to be changed.
So what did we do?
We started with what we knew needed to be changed and what we thought would move us in the direction we wanted to go, and we built some structures to allow us to refactor those parts of the code, knowing that we would probably end up deleting those structures.
The scaffolds that we created allowed us to have a midway point in the long journey across the Pacific ocean of complexity in our refactor trip.
The point here is that we had to take a bit of hop blindly in a direction we thought things needed to go and part of that hop involved creating some scaffolding type structures to get that code to a place where it still worked and we could examine it again for the next refinement.
The final refinements ended up deleting those scaffolds and replacing them with a more elegant solution, but it is doubtful we would have been able to see the path to the elegant solution without building them in the first place.
Show me the code!?
You may wonder why I am talking about code in such an abstract way instead of just showing you a nice little “scaffold pattern” or a real code example.
I’m not going to show the code, because it isn’t relevant to my point and it is so situational that it would detract from what I am trying to say here.
The point is not what our problem or solution was, but how we got there.
There isn’t a “scaffolding” pattern that you can just apply to your code, rather it is a tool that you can use and shouldn’t be afraid to use in order to move your code forward to a better design. (To its happy place.)
Are you really sure you want to create the cancel button for your application?
You know, I might click it.
Not only might I click it, I might click it at the most inopportune time.
I might click it right in the middle of that large file that you are copying, right after you spun up that 2nd thread.
Cancel is a commitment
The next time you consider creating a cancel button, I suggest you think of it as a commitment.
In my world the cancel button has two promises.
- Stop what is currently happening in a non-destructive way.
- Stop RIGHT the F NOW!
I’ve been encountering a good deal of cancel button which don’t obey the basic laws of cancelling that any user would expect.
I have actually been randomly clicking “cancel” whenever I see it, much to my coworker’s dismay.
I started doing this, because I wanted to see how widespread the cancel cancer is.
And it is pretty widespread. A majority of cancel buttons I have been testing neither stop right away or prevent destruction.
I found instead that clicking the “cancel” button in most programs is sure to hang that program for an extended period of time, ultimately causing you to kill the process, and it tends to leave processes half done without rolling them back.
Clearing things up
Let me be a bit clear here. I am not talking about cancel buttons you see in an “Ok / Cancel” dialog. Most of the time those cancel buttons actually work, because they are really operating as “back” buttons, they aren’t actually cancelling a process that is happening live.
I am talking mainly about cancel buttons that cancel an active ongoing activity. For example, cancel buttons in an installer or during an SVN update.
We could call these kinds of cancel buttons “asynchronous cancel buttons.”
But, I need to provide a way for the user to cancel
Good, but don’t lie about it.
There are certain things that just can’t be cancelled.
When I get on a plane, I can cancel my trip when I am sitting there waiting for the door to close. I can even cancel my trip while the plane is taxing to the runway, if I yell “I have been RAPED by a BOMB that is on FIRE!”
But, I can’t cancel my trip, once the plane has lifted off the ground. If I try to cancel it then… well, bad things would happen. Very bad things.
So how come when I am installing your software, or your software is updating its database, I have this shiny little cancel button I can click at any time during that process?
Surely I cannot cancel just any time!
Surely there are parts of the process that cancelling would be fatal or it is too late to rollback.
My point is, if the user truly can’t cancel, don’t present a button that says they can. More specifically, if you can’t obey the laws of cancel 1 and 2 from above, don’t even show a button. Just say “sorry, you can’t cancel this process at this time.”
I don’t even need to know why I can’t cancel. I mean, it will make me feel better if you tell me that the Unicorn Glitter Engine is in a critical state and any disruptions could end life as we know it, but I’ll settle for you just greying out that cancel button or not displaying it at all.
Putting back on the developer hat
I’m guilty of it myself. I know I have created cancel buttons in the past that have caused pain and anguish.
But what can we do about it as developers?
First off, we should be thinking carefully about breaking a long running process into steps. At each step of the way we should consider if it is conceivable to cancel that step without destroying data and hanging the application.
In any long running process, we should be able to identify certain parts which are cancellable and those which do not make sense to cancel.
It is your job as the developer to ensure that if you decide to allow for cancelling that the cancel undoes the existing process and work immediately.
I cannot stress this enough!
This is the behavior that most users expect and the very meaning of the word cancel.
To do this might take extra work. You might have to think a bit more about the threading situation of your application. You might have to think about rollback situations. But, if you don’t do it, your cancel button will become just like the boy who cried wolf and no one will believe them.
And if you’re not willing to go through this effort, at the very least, be kind enough to your users to just remove the cancel button, because you can bet I’ll be clicking it!
If you’ve made it this far with your BAT implementation, you have finally arrived.
Not to say that you’ll ever be done expanding your automation framework and building and improving your BATs, but you are at the point of having a mature usable BAT framework and you can be proud of that accomplishment.
If you truly want to see where the rabbit hole goes though… Read on
Taking it to 11
If you have designed a good automation framework for writing your BATs, you will most likely have created an internal domain specific language (DSL).
What is an internal DSL?
Basically it is a set of APIs and rules built in another language that function as their own language.
You might have some code that looks like this:
Pages.Login.Goto(); Pages.Login.LoginWithDefaultUser(); Pages.Customers.Goto(); Pages.Customers.AddNewCustomer();
You can see how this code is really looking like a language itself. It is hard to tell if this code is even C# or Java.
This is a good example of an internal DSL.
If we were to take this same section of a BAT test and write it as an external DSL, it might look something like this:
LOGIN WITH DEFAULT-USER ADD-CUSTOMER WITH DEFAULT-CUSTOMER
With a true DSL we can simplify even further to get right down to the bare minimum set of commands we need to execute what we want in our application.
How to implement a DSL?
There are a number of tools and techniques that can be used to create a DSL. I’ll hit on some of those a bit later on, but first we need to design the DSL itself.
Designing a DSL can be challenging, but I have found the best approach is to use a similar approach to what we used to design our automation framework.
It is best to work backwards from the syntax that would make your BAT the easiest to write and the most clear. Once you have done that you can worry later about implementing the DSL to support the syntax you have come up with.
You should focus on creating a limited vocabulary for your DSL so that it will be easy for someone to learn and know what their options are.
The more restrictions we put into a DSL the easier that DSL becomes for doing a specific purpose.
We want to give the user of a DSL no more options than required for what they want to do and no less.
This isn’t always easy to do. It often requires some careful thought and refinement.
Your DSL should also use terminology and phrases that are familiar to the business domain of your application. By doing this you will make it so someone understanding your application will understand BATs written in that DSL.
I also always try to make a BAT DSL read as much as plain English as possible.
Once you have designed the DSL, you will need a way to implement it.
Microsoft also has a DSL SDK you can use in VS2010, although I haven’t used it myself.
The basic idea behind implementing a DSL is that you have to take some sort of grammar you design for your DSL, parse it into a syntax tree, then use that tree to generate code that will actually perform the actions.
It is by no means a trivial task, but using a tool like ANTLR, makes it much simpler than trying to completely generate a parser and execution engine yourself.
Why even bother?
Creating a DSL for your BATs is completely optional, but there are some significant benefits to doing so.
The biggest one I have found is that you are much more likely to get business people to write BATs if you have a simple DSL they can use to create them. Once you get everyone using BATs to define behavior and requirements, communication becomes much more simple.
Another great thing about creating a DSL is that you can create a runner that can interpret the DSL BAT scripts which can be run on any machine. Typically, to run BATs created in Java or C#, you would need a C# or Java compiler and a test runner, etc. By creating a DSL and a stand alone executor, you can very easily allow anyone to run BATs with little to no setup.
Writing your BATs in a DSL will also make writing the BATs and understanding them much easier. A good DSL can get rid of all the little language and API nuances that don’t really add information to the specific domain you are working in. A good DSL strips information down to the minimum amount needed and is able to do so, because it assumes the context is your application’s business domain.
You will also find that you are better able to separate framework from tests when you have a DSL in place. One issue that many teams face when creating a BAT framework is that some people come along and write lower level code in the higher level BATs. This creates a problem because it is hard to force someone to use the higher level API when the framework and the BAT code are the same language.
Wrapping it up
So that completes my series on Getting up to BAT.
Hopefully you have learned enough to be able to implement your own BAT framework and get BATs running for your application.
I tried to cover most of the important steps here without getting too deep into the details.
If you have any questions or are interested in getting more in depth training or consulting on getting BATs setup for your environment, send me an email or leave a comment.
It can be a little difficult to get started and get a BAT framework up and running, but the rewards are well worth the effort!
As always, you can subscribe to this RSS feed to follow my posts on Making the Complex Simple. Feel free to check out ElegantCode.com where I post about the topic of writing elegant code about once a week. Also, you can follow me on twitter here.
When we last left off we had just gotten our BATs as part of the acceptance criteria for any new backlogs that are worked on. This puts us at a point where we could really say that we have successfully implemented BAT testing.
You don’t want to get too comfortable just yet though, because the next hurdle you will most likely face will be the problem of not having enough time to execute all your tests.
You want to think about this ahead of time
Nothing worse than getting everything going and then not being able to execute the entire test suite, because you didn’t plan ahead.
You don’t want to get to the point where confidence in your BAT suite it lost because you are not able to get all the tests executed in a reasonable amount of time.
The more frequently your tests are run the more value they have.
By reducing cycle time from the time a breaking change is made in the system and the time it is discovered, you greatly reduce the risk it imposes on your software and you decrease the scope of the code changes in which it could have occurred.
To put it simply, the faster you can find out you broke something, the more likely you can fix it before it does damage, and the more likely you will be to know what thing you did caused the breakage.
How can we reduce cycle time?
There are a few different strategies we can employ and we can mix and match some of these strategies.
Straight forward parallelization
The best and most effective thing you can do is to take a given test run, split up the execution of those tests on multiple machines and execute them in parallel.
This approach is going to give you the best bang for your buck. You should really try to get some amount of parallelization going before attempting any other solution, since it is going to make a huge impact on the total execution time for your tests without making any sacrifices.
There are even many ways you can mix and match to do parallelization:
- Use multiple physical machines
- Run many virtual machines on a single host
- Put an executor program on every team member’s machines that will execute tests when that machine is idle or put into test execution mode (perhaps at night)
- Use a cloud based computing platform to execute your tests
- Run multiple browser instances on a single machine
With this approach, you would preload some of the test data that your tests might generate by manually clicking through screens.
This is fairly hard to explain, so let me give you an example:
Suppose you had a set of tests that all involved creating customers, but each customer you create in the system takes about 3 minutes to create by clicking through the screens to get them into the system.
We don’t need to have 500 tests all executing the same exact logic in the system 500 times for 3 minutes just to generate all the customer data that will be used in the tests.
Instead, we can leave a few tests that are exercising the customer creation functionality, and we can execute a SQL script to push all the other customer data into the test database for the other tests to use.
Using this technique we might be able to reduce our total execution time by 3 minutes * each test, or about 25 hours for 500 tests.
This can be a huge savings in time, and it doesn’t come at that high of a cost. The sanctity of our tests is slightly compromised, but we are taking a calculated risk here knowing that we already have covered the area of execution which we are preloading data for.
Consider this technique when you notice certain things in the system taking a very long time to do.
Test runs by area
With this technique, we can reduce the total execution time in a given period by splitting up test areas and running them either at different times or in response to changes in certain areas.
You have to be very careful with this approach, because if you don’t do it correctly, you can start to erode some of the safety your BAT tests are providing you.
I would only do something like this as a last resort, because it is so easy to virtualize today, and hardware is so cheap.
I’d much rather run too many tests than too few.
With test randomization, we are going to take our total desired execution time, and divide it by the average time for running a test. We then can use that number of tests to run to randomize the execution of our tests each time we run them and only run the number or tests that will fit in the desired execution time.
This choice is also a compromise that I typically don’t like to take.
It can be very useful though, combined with other techniques when you still don’t have enough time to execute your entire test suite.
The basic idea here is that you are going to randomly run tests each time to fill up the time you have to run tests.
This one seems fairly obvious, but can be very helpful.
Often I will see teams starting out with automation, trying to write way too many BAT tests for a given feature. Sure, with automated tests it is possible for run a test for every single possible combination of values in your 5 drop downs, but will it really benefit you?
In many cases you have to think about what you are trying to protect against with your BATs. Sometimes running every combination of data selection choices is going to be important, but other times you are only going to need to write tests to test a few of the happy path scenarios.
It is important to find a balance between test coverage and test volume and not just for execution time. There is a logistical overhead to having a large volume of mostly redundant tests.
So even though this technique might seem dangerous and counter-productive, I will almost always employ it to some degree.
Here are some of the things you might want to watch out for as you are scaling out and streamlining your execution of BATs:
- Parallelization issues. If you are using shared state, you can run into big trouble when your tests are executing in parallel. There are many manifestations of this. You could have issues at the database level or at the local machine memory level. The best way to avoid this kind of problem is to use separate virtual machines for each test execution, and not reuse data setup between test cases.
- Ineffective error reporting. If you run a huge volume of tests in parallel, you better have a good way to sort through the results. It is much harder to figure out why things failed when they are run across multiple machines.
- Test order dependencies. Make sure tests don’t rely on being run in a certain order or you will have lots of pain when you disrupt that order.
- Environment setup. Make sure all your test execution environments are exactly the same unless you are specifically testing different environments for execution. You don’t want tests failing on one machine but passing on another.
As always, you can subscribe to this RSS feed to follow my posts on Making the Complex Simple. Feel free to check out ElegantCode.com where I post about the topic of writing elegant code about once a week. Also, you can follow me on twitter here.
What if I told you that there is a program so complicated that is the source responsible for all other programs ever created?
Give me your most complex program and I will trump it by presenting you with the program that was responsible for creating the human who created that program.
If you haven’t figured it out by now that program is human DNA and that masterful programmer is the Creator Himself.
Doesn’t matter what religion you are
It is pretty difficult to fail to recognize that a program as complex as a human being’s genetic code was created, not randomly assimilated.
As programmers we have a unique perspective in life. We are able to truly understand God’s masterwork in the creation of human life. We are able to grasp loosely the ungraspable magnificence of a program so complicated, yet executed so simply.
We also have the unique position of fundamentally understanding that complex programs could never be created by random chance, regardless of time. We can look at a piece of code or functioning computer program and we can understand that its creation required an intelligent creator. Other people and professionals do not have this insight that so many of us are fortunate to have.
Qualities of the code
We’ve been treating data like code recently in C# with the introduction of Lambda expressions and other languages have been doing it for much longer, but the ultimate example of this practice is right there in our own DNA.
DNA is data, a message, and code all together at once. Just like a data structure in the memory of a computer executing a program, it lacks meaning without the context of the cell that it is in. Amazingly though, this data is recombined in different ways that are dictated by two separate and different instances of itself to produce a third separate instance of itself that still makes sense.
Just to give you an idea of how insanely complex that idea is (despite the simple execution of it), it would be like if you wrote a webserver that could spawn new instances of itself by contacting another, but different copy of your webserver and getting 1/2 the source code payload from itself and the contacted server, then actually combining that code to create a new copy that is different, yet still passes all of your unit and integration tests.
One of the qualities we often look at for our code is the size of the code in relation to what it can accomplish.
A decent estimate is that the human genome is about .35 gigabytes of data. That data is stored on such a small space that we can’t even see it with the naked eye. There is no more compact form of data in the universe that we know of.
Not only is this data compact, but consider how much functionality is built into this small .35 gigabyte package? If you had to write acceptance tests for the human body, you might spend several lifetimes doing it. If you had to write acceptance tests for just a single cell in your body, that feat itself could take years and still not be complete. Imagine trying to write a program that took up .35 gigabytes of data, yet contained millions of acceptance tests worth of functionality.
We certainly could go on and on about the qualities of the human code, but I want to point out what perhaps is one of the most unique and amazing… the ability for the code to create its own hardware.
In actuality the code and that hardware are much the same. DNA replicates by a chemical reaction that causes the entire molecule to unwind and split. RNA inside a cell is created from specific portions of the DNA in order to create different protein chains which are used to construct little machines that will replicate the DNA and cells and in turn create machines capable of doing the same.
The process is very complex and not wholly understood, but in effect the end result is that DNA contains the instructions inside of a cell to create the machinery to host another copy of itself and execute that code.
Let’s not forget the little Easter eggs that have been thrown in by the designer of our codebase.
How about a hiccup or a tear?
Sadness or laughter?
There are some “evolutionary theories” as to why these things exist, but in reality they serve no specific purpose for either our survival or well being.
In my mind these are clearly Easter eggs left by the designer to show His handiwork.
Why am I writing this post now?
I didn’t want to go completely off subject and boast about my wonderful daughter that was born on April 17th 2011 at 2:33 PM, since this is a professional blog.
But, there you have it.
The amazing birth of my first child reminded my of the incredibly complex code that keeps us all breathing, procreating and living life that we seem to take for granted each and every day.
Welcome to the world Sophia Grace!