Activity journal - progress update

Progress update about the Activity journal project.

Going full on TDD

In the end, I decided to do an experiment and to start with almost no libraries and no architecture at all. I would like to discover a need for dependencies and architecture during TDD cycles. This is the hard way and not really a production-time way, at least when development for Android is considered. But it’s an interesting exercise.

I am still biased by my ideas about the technologies and architecture I want to use therefore I may steer my way into them a bit. I also decided to consider only a single activity application (with navigation using fragments) for this project, even though I am not convinced that it actually is the correct approach.

How it goes?

Here is a Github repository. Keep in mind, it’s heavy work in progress.

What I did so far:

• Wrote the first test and then generated first activity, because TDD!
• Prepared a setup for Circle CI, a code coverage report generation and a Codecov integration. It took more time than I anticipated, but it’s a one-time investment. Working Circle CI setup without any functionality is available here.
• Did a few sessions and managed to implement some basic functionality for timeline and a simplified add activity flow.

The application is now in a very strange state.

There is no persistence yet. I started just with an in-memory List for ActivityRecords. The idea is that as I implement various requirements and write tests for them, I will get to adding a persistence.

There is almost no architecture yet. But the need for a better structure is already showing itself. I introduced a repository for ActivityRecords when I was implementing sorting and there is a small backing object that helps to manage state during the add activity flow. I want to add a little bit more functionality and than I will try to “refactor some architecture out”.

So far I was writing mostly espresso tests. It’s not as bad as I expected, but it does really make sense to move as much testing as you can to unit tests.

I was able to keep my quite strict TDD approach for most of the time. I recall only one situation where I had to take a bigger leap because I wasn’t able to figure out a small step.

But I must admit this approach is definitely slower than throwing in MVVM and Android architecture components and starting from there. I believe that it’s a valid approach (considering TDD) to use your knowledge of the platform/domain and start with some architecture and libraries. You should just see them as a temporary scaffolding that may change.

In the end, I must also say that it’s fun and a really interesting exercise.

Ehm, streaming, really?

Since this is my training and exercise project I want to exercise as much as I can. And I had this silly idea I might do some live coding, streaming on twitch.tv. The idea is, that during streaming there are 2 additional aspects I am training:

• I am forcing myself to overcome some fears about coding when watched. Similar emotion can influence pair programming or coding challenges during interviews.
• I am practising my active English when I am trying to explain what am I doing and why.

I still feel quite embarrassed about it. I am quite nervous for most of the time and I feel that I struggle with things I would be normally able to solve quicker. Also, I feel this strange guilt every time I run into something I don’t know and need to figure out (Google, StackOverflow) and somebody is watching. But those are exactly the emotions I want to overcome. And slowly I am getting used to it.

Feel free to subscribe and there are also recordings available. But huge disclaimer: I am awful at it, it’s boring, most of the time I just struggle with Espresso or something. You should know better ways how to spend your time.