Category Archives: TDD

Update on Continuous Deployment a few months in

On average, I’m doing just under four totally automated zero-downtime production deployments per day since I started doing continuous deployment on this project in early August 2011.

Obviously, that’s an average, so some days there are more, and some days there are fewer.

How many times during that span has improperly tested code broken things in a way that impacted users? Just once, and it was fixed in just a few minutes by adding back the code that I over-zealously removed.

Also in that time, we went from 0% automated test coverage to 18% automated test coverage, refactored a lot of duplicate code, and added new features.

Continuous Deployment for Existing Software – Do it Now

I recently wrote about doing continuous deployment from day one with a software project that had pretty good test automation and (if I do say so myself) a somewhat modern and decent architecture.

I’ve recently transitioned to working on a fairly ambitious overhaul of an existing project that has been around for a few years. The software has already reached a (mostly) working steady state and is used every day by real customers.

The very first thing I did, ahead of making any functional changes, was automate the deployment system using essentially the same kind of test gauntlet and approach for zero-downtime deployments I was using at Victors United. I added some very basic test automation, starting from the top of the test automation food pyramid with a simple “is the web server able to execute code, connect to the database server, and return an ‘OK’ rsponse?” test. From there, I’ve working my way down, just getting around to writing my first “pure” unit test just today.

Only after I was confident that I could do zero-effort and zero-downtime deployments that

  1. Wouldn’t completely destroy the system (I had a test for that) and
  2. Could be rolled back very easily if something went screwy

did I make my first functional change to the software.

And then, I made the smallest functional changes that could work. I tested them locally, added some test automation around them, and then let the automatic deployment system do its magic.

A few years ago, I was interested in doing continuous deployment, but I felt that the level of test automation wasn’t good enough, I wanted to be at least at around 90%.  Now I think that the less test automation you have, the more important it is to start doing continuous deployment of tiny incremental changes right away.

If you are working on software that’s actually in use and are interested in doing continuous deployment someday, there’s no better someday than today. Seriously. Do it now. It will start making your life better almost immediately.


Always Be Shipping – Real-World Continuous Deployment

Have you ever edited code directly on a production server? I’ll admit that I have, years ago, before I knew better. It’s easy and fast and gets emergency fixes out there as quickly as possible. You get to know what works and what doesn’t because you make tiny changes. If you’re working with a system that anyone cares about, it’s also dangerous and stupid.

I’ve also worked with development organizations that take the opposite extreme. Even the most trivial server updates needed to be scheduled weeks in advance. In order to push the updated code, you had to notify everyone, take all servers down in the middle of the night, run through a long series of manual steps, and then do a lot of manual testing. When things in production aren’t quite how they are in the test environment, this is followed by a hasty rollback or panic-induced middle-of-the-night attempts at bug fixing.

Both of these extreme approaches are (maybe) appropriate in some environments. But neither are appropriate for a new web startup that’s trying to move quickly and still provide a reliable trustworthy experience to their customers.

At the previous company I worked for, we often talked about IMVU-Style Continuous Deployment as our ideal process, but it was always something “in the future”. We were hesitant (some of us more than others) to do automatic deployment without at least a little manual intervention. We always wanted to have more test automation, or a smoother deployment system, or whatever.

Since it seemed to be hard (for me anyway) to move to an existing development organization to a continuous deployment system, I started to wonder what would happen if you do it that way from day one? I got a chance to answer that question when I co-founded a startup last year. One of the very first things I did, before we had anyone using the site, was to create an solid automated test & deployment system that was as fast and easy as possible without being dangerous and stupid.

Here’s the basic workflow that happens in our office multiple times every day.

Step 0. We make changes on our local dev envirnments, with a bias toward making the smallest possible change that adds value. That could be a bug fix, correcting a typo, a stylistic tweak, a stubbed-out new feature, whatever. Once I’m confident in my local (manual and automated) testing that the change is good (not perfect, not feature-complete, but just better), I push that to my github repository.

From there, the continuous integration server pulls down the new code and does the following:

Step 1. Does the code still compile. If not, the build fails and everything stops.

Step 2. The build agent runs the unit tests (where “unit tests” are defined as tests that run with no external dependencies, these take just a few seconds). For anything that does require external (generally slow) dependencies (network API, databases, filesystem, whatever) we use test doubles (fakes, mocks, stubs, whatever).

This first feedback loop is about catching and preventing errors in core business logic and is generally measured in seconds, not minutes.

Step 3. The build agent runs a set of tests that rebuild the database from a reference schema and exercises all of the repository layer code.

Step 4. The build agent runs another set of tests that test our dependencies on external APIs (twitter, geolocation services, etc.)

These two sets of tests run in a few minutes, so the feedback loop isn’t quite as tight, but it’s still pretty darn fast. Basically, they make sure that the assumptions that we make with our test doubles in our unit tests aren’t totally wrong.

I’ve written about these sorts of automated test distinction a couple of years ago, in a post about the Automated Testing Food Pyramid.

Step 5. Provided that the entire gauntlet of tests has passed so far, the code gets automatically deployed to a staging server.

Step 6. There’s an additional set of tests that run against the staging web server. These tests can find configuration problems and code that just does the wrong thing in a web context. These tests are pretty shallow. They hit all of the user-facing pages/JSON endpoints and fail if anything is totally broken.

Step 7. The build artifacts are copied from TeamCity to a new folder on our productionvserver, and then the web server is reconfigured to serve from that folder instead of the folder it had been serving from.

At this point, we’ve verified that the core business (game, in this case) is OK, verified that the persistence stack works as expected, that our integration with external APIs works as expected, and that the code doesn’t completely break in a web context. We’ve done a zero-downtime deploy to the production web server.

That’s cool, but we’re not quite done yet. There’s two more steps.

Step 8. Run a set of tests against the production web site to make sure that all of the pages that worked a few moments ago still work.

Step 9.  Have external monitoring systems in place, so if your changes make things slow or unresponsive. You’ll know.  We use pingdom.

Yikes! There’s a bunch of distinct steps here, and it seems really complicated (because it is). But it’s all totally automated. All I need to do ?

git push origin master

Because there’s zero deployment effort on my part, I do this all the time.  I find it very energizing to know that I can just do stuff in minutes instead of hours or days or (heaven forbid) months.

If (when) something goes wrong, I’ll know immediately. If a bad bit of code manages to roll through the test gauntlet, I can roll back easily (just reconfiguring the web server to use the last known good set of code). I’ve only had to roll back a couple of times over the course of several months and 326 deployments.

Just like when the folks at IMVU wrote about this process, I’m sure that some people in the audience convinced that I’m a crazy person advocating a crazy way of working. Here are the objections I’ve heard before.

Yeah, but this doesn’t give your QA team any time to test the code before it goes out!
We’re a small startup. We don’t have a QA team. Problem solved.

Yeah, but isn’t that incredibly dangerous?
No. The safest change you can make to a stable production system is the smallest change possible. Also, we design the individual parts of the system to be as encapsulated as possible, so we don’t tend to have crazy side-effects that ripple through and create unintended bugs.

When we make a change or add a new feature, we can manually test the hell out of that one thing in isolation (before checking in) instead of feeling like we need to spend a lot of time and effort manually testing everything in order to ship anything.

Yeah, but what about schema changes?
For schema changes that are backwards compatible with the code that’s out there (e.g. new tables, whatever). We have a simple system that executes the appropriate DML on application startup.

For non-compatible schema changes and things like server migrations, we have to take down the site and do everything manually. Fortunately, we’ve only had to do that twice now.

Yeah, but you have to spend all of that time writing tests. What a waste!
The time we spend writing tests is like the time a surgeon spends washing their hands before they cut you open. It’s a proven way to prevent bugs that can kill you.

Also, we get that time back, and then some, by not having to spend nearly as much time with manual testing and manual deployments.

Yeah, but what about big new features you can’t implement in just one sitting?
Traditional software development models (both waterfall and agile methods like Scrum) are organized around the idea of “multiple features per release (or iteration)“. Continuous deployment is organized around the idea of “multiple releases (or iterations) per feature“. As a result, we end up pushing a lot of code for features that aren’t done yet. For the most part, these simply unavailable through the UI or only exposed to users who have particular half-built features associated with their accounts. I credit Flickr with this general approach.

Yeah, but that might work for a solo developer, but it can’t work for a team.
There are actually three developers on the team.

Yeah, but I’m sure this only works with very experienced developers
One of the guys on the team has only been programming for the last year or so and hasn’t ever worked on a web project before. Tight feedback loops help everyone.

Yeah, but what about code you want to share with other that you don’t want released?
We use github, so creating additional branches and sharing with them is trivial. We also have a dedicated “preview” branch that triggers a parallel test/deploy gauntlet that sends code to a staging server instead of the production servers.

Yeah, but this will never work at my organization because…
OK. That’s cool. Don’t try it if you feel that it won’t work for you. You’re probably right. You’re not going to hurt my feelings either way. I found something that’s working really well for me, and I want share my experience to show other people that it’s possible.

What this really means

Half of what makes this process work is that we’re honest with ourselves that we’re human and will make mistakes. If we have multiple tight feedback loops between when we’ve broken something and when we know we’ve broken it, it’s faster and easier and cheaper to fix those mistakes and prevent similar mistakes from happening again.

The other half is the idea that if you design, implement, test, and release exactly one thing at a time, you know with certainty which change introduced a problem instead of having to ask the question “which of the dozen or so changes that we rolled out this month/sprint/whatever are causing this problem”.

About the site

Victors United is an online turn-based strategic conquest game. You can play asynchronously or in real time. You can play against robots or humans. If you’re playing against humans, you can play against your friends or against strangers. Unlike some other popular web based social games that I don’t like to mention, this is a real competitive game where strategy and gameplay matter.

About the tech 

The tech here is kind of beside the point. This general approach would work just as well with different technology stacks.

The front end is HTML5 + JavaScript + jQuery. The backend is IIS/ASP.NET MVC2/SQL Server/Entity Framework 4. Our servers are hosted in the SoftLayer cloud. External monitoring is provided by pingdom.

The test gauntlet is a series of distinct nUnit assemblies, executed by TeamCity when we push new code to GitHub. There’s a single custom PowerShell script that pulls down the build artifacts and tells IIS to change what directory it serves code from.

Oh no, not more of the same TDD discussion Tedium.

Every once in a while, otherwise reasonable people get together to argue about TDD with religious zeal. In the most recent flare-up, I’ve been disappointed that on all sides, as nobody is saying anything new.


At the risk of adding yet more noise, I did have a two nuanced thoughts that are at least new to me and I thought I would share them, along with a recent personal anecdote.  If this is obvious or old-hat to you, then I’m sorry.  If you think I’m  too stupid for words and that I’m drinking and/or selling snake-oil enriched kool-aid, I look forward to what will undoubtedly be informed and insightful feedback.

1. s/driven/aware/

The hardcore position of “never write a line of production code without a failing test for it” probably does more harm than good. Different kinds of code require varying degrees of effort (cost) to write and maintaintests for. Different kinds of code give varying degrees of benefit from test automation. Without always realizing it, developers make cost-benefit decisions all the time, and good development organizations empower their developers to act on those decisions.

That said, the cost-benefit decisions developers make must at least be informed decisions. A professional developer who hasn’t taken the time to learn how to use the appropriate test frameworks for their language/environment (jUnit, nUnit, whatever) is just plain negligent in 2009.  Test automation is just one tool in a competent developer’s toolbox, but a critical one.  I wouldn’t trust a carpenter who didn’t know what a hammer was, or a cardiologist who hadn’t bothered to learn about this newfangled angioplasty business.

Test-driven may not be appropriate for every context, but everyone needs to be at least test-aware.

2. s/first/concurrently/

Test-first is a really helpful approach, but it doesn’t work with the way everyone thinks, and mandating that everyone must always think in exactly the same way is the worst sort of micro-management.  The other extreme, writing test automation for a large system after it’s complete, is often prohibitively difficult and (frankly) boring as hell.

My advice is to always at least think about how you would write tests for your code before writing it. That will help keep you from painting yourself into untestable corners. Also, interlacing test writing immediately after you get a small subset of your system done is going to be much easier than testing the whole thing after the fact.  Personally, I move back and forth between writing the tests first and writing the code first. The key for me is that I’m working in short code-test-code-test cycles, using persistent (that is, I don’t throw it away when I’m done) test code as the primary mechanism for executing the code I’m writing as I’m writing it.  I don’t think of the process as being test-first, I think about testing concurrently with coding.

Recent Anecdote

Sure, anecdotes aren’t data, and they can’t prove anything, so take from it what you will

I just finished a pretty big refactoring project (that is a “pure” refactoring, the external interfaces and behavior of the existing stay the same but the underlying implementation was improved) of a system that had some decent test automation. Every time I got an edge case behavior wrong, introduced a side-effect, or removed a necessary side-effect (yuck), a test would go from green to red. This saved me at least a few days of development and testing time, and reduced the chances that I would release bugs to our QA guy (bad) or our production system (even worse).

Automated Test Distinctions: The Food Pyramid

This is something I came up with a while ago when I was working on a large integration/orchestration project for a now-defunct MVNO (mobile virtual network operator). It was a pretty complex project, which had many external collaborators and end-points into the system. The project was having some issues with being able to reliably deliver incremental functionality on time. I had a directive to use my judgment to essentially turn the project around.

I had just come off a project which had excellent pure unit test coverage, but not very good overall test coverage with actual collaborators. The integration points were much more difficult and painful than they needed to be. On the other extreme, I inherited a bunch of existing non-unit tests as part of the MVNO project. These were brittle, took a long-time to run, and failed for a bunch of different reasons. As a result, they weren’t run very often, became neglected, and eventually worthless.  There was no continuous integration in place, so the feedback loop between breaking something and finding what you broke was very loose.

To make matters just a little more dicey, I was working with another test-focused developer who had a strong preference for writing epic end-to-end tests, while I was trying to make the code more able to be tested in isolation. I resented the fact that “his” tests were keeping me from running the tests as part of the cruise control build.

I came up with what I called the “Food Pyramid” as a way of balancing these different forces and points of view. I broke the one massive test project into three distinct assemblies with their own goals.  The differences between the projects are best explained in a series of pictures:

So, did it work? Yes and no. We eventually had much smoother and more confident deployments thanks to the end-to-end tests. I also managed to refactor and add new code more quickly with fewer bugs thanks to my unit-test safety net. But, just as I was starting to get a handle on the project, the client filed for bankruptcy and stopped paying their vendors (which unfortunately included me). It seems that having a viable business model trumps all.

I have, however, used these and similar distinctions successfully on subsequent projects, and I know that at least one development organization I’ve explained the idea to has adopted the pyramid as part of their overall testing strategy.

Testing Abstract Classes In Isolation

In my post about guilt as a code smell, a comrade pointed out that it’s perfectly possible to test abstract classes in isolation, you just make a concrete derived class as part of the test (thanks Craig!). Having a distinct concrete subtype for testing is something I’m already doing a lot with endo-testing, so it’s not even totally without precedent.

It does still bother me, though, and I’m not 100% sure why. Some thoughts:

Let’s say your abstract class is using the Template Method pattern, where it has a public method which just delegates “downwards” to abstract methods with different overridden implementations. This is a perfectly good use of an abstract class, yet it seems kind of pointless to test in isolation, as you’re going to be testing each implementation and will be testing those base classes anyway. 

The scenario that I had in the other post had a different kind of abstract class, with no public methods, just protected ones. I’m just achieving zero redundancy by moving code logic upwards in the tree. Testing here becomes trickier, as there’s no publicly exposed design surface. Should I just make one up? The right solution, for me, for that time, was not to test the abstract class, but to move that logic into a distinct service class so I could work with it more directly.  It’s textbook “favor composition over inheritance” design patterns stuff.

In any case, I stand corrected. It’s absolutely possible to test an abstract class in isolation. It is impossible, however to test a base class in isolation without testing its abstract class. Which could make testing the abstract class in isolation kind of pointless.

Guilt as a Code Smell

One of the best things about working for Velocity Partners is that I get a chance to do presentations/brown bag seminars for their other clients/prosepective clients. I like to think that as I’m a hands-on developer who actually works with this stuff every day, I have different credibility from the full time trainers/presenters/coaches/pundits.  Note: I don’t want to disparage the full-time trainers that I know and respect, their credibility comes from the fact that they were hands-on coders for a long time, and have more time to do the reading/research/writing that someone working on deadlines may or may not have.

I’m currently putting together a presentation on Refactoring, and how that relates to the other agile tools and techniques (where “agile” simply means “modern development practices that work”). While looking for examples, I’ve been re-reading Martin Fowler’s great Refactoring book.  One thing that struck me was the focus on “getting the inheritance hierarchy right” which, after living in the design patterns (favor composition over inheritance) realm for the last few years, felt kind of odd to me.

Meanwhile, in my “day job” I’ve been working on a well encapsulated, generics-based .NET client for a family of REST-y XML services.  After making one major component, I found that I had to make another major component that does much the same thing. So, I created a new abstract base class and made both my existing component and the new component concrete derived classes of the base class. As I needed functionality for the new class, I generalized it and moved it up to the abstract class (using the protected modifier, of course) so I could use it in the other derived class.

It was working pretty well. I had very little code duplication and ReSharper is particularly good at the “Pull Members Up” refactoring, but I was starting to feel a little guilt about not doing things in a “pattern oriented way”. Sure, you could never use the two concrete components interchangably, so there was a violation of LSP, but I can be cool with that. The abstract class didn’t have any public methods on it, so there’s no danger of someone trying to couple to its publically-exposed interface.

After, I figured out where my guilt was coming from. It’s a coupling/testability problem. As the concrete types are tightly coupled to their base types, I couldn’t ever substitute a different type to handle that functionality. This is particularly important as the base type was all about making external (HTTP) service calls, so it was impossible for me to test any of the types in isolation.

The solution: it’s pretty obvious, but I just moved the functionality of the abstract base class into a service class with an interface. Now my two components (formerly derived classes) just have an instance typed to that interface. I can test all three parts in isolation (Q: How do you test an abstract class in isolation? A: You can’t) and I can re-use the same functionality across additional components in the project, further reducing duplication.

So, what I’m saying is: take your guilt seriously. If you have a bad gut feel about a design, it might very well be bad.  It’s just like any of the code smells in the original Refactoring book. It doesn’t necessarily mean that there’s a problem, but it’s worth looking at.

But, at the same time, I’m not going to beat myself up over my interim design. It was a good, easy stepping stone to a more optimal design. This is the sort of thing that Scott Bain’s new book Emergent Design: The Evolutionary Nature of Professional Software Development is about. I’ve only skimmed it so far (I’m working on a new presentation, after all) but I know Scott, and I know his approach to the subject. From what I’ve read so far, it seems to be the right text for the professional programmer who wants to move beyond “just getting it working” to the level of “getting it working well”.  I wish that I could have read it years ago.

More C# Partial Class Testing Strategies

I can’t take credit for this approach, and even if I could, I probably wouldn’t, because it makes me feel kind of icky.

Anyway, I recently heard about a legacy code testing strategy where you mark your class as “partial”, and in another file, you add whatever public properties/methods you need for your tests. You make the contents of the second (testable file) conditionally compiled (the classic #IF DEBUG) so the encapsulation is still there for any release builds.

It’s kind of like endo-testing, but you’re extending the class “sideways” instead of “downwards”.

Basically, it’s breaking encapsulation in a controlled way, and for the most part, I think it’s a bad idea if you’re working with a new design. If, however, you’re trying to get some meaningful coverage for your legacy code (which wasn’t designed for testability) it can be a good stop-gap in dealing with the legacy code refactoring catch-22: where you don’t want to make changes without tests, but you can’t make tests without making changes. 

Any strategy, such as this one, which allows you to get the first layer of tests down before further refactoring, should be embraced as a good thing.  If you find that you need to do this for any new/original classes, my guess is that your class is too big and needs to be decomposed further into more cohesive and testable classes.