Skip to content

500 Conclusion

Fasust edited this page Nov 25, 2019 · 33 revisions

Page Table of Contents

Introduction

So, you’ve made it. This is the final chapter of the guide. I will use this chapter to reflect on the guide, evaluate its strengths and weaknesses, and highlight how I and this project will move foreword from here on out. I will also take this opportunity to share my personal opinion of the current state of the Flutter Framework [1]. That being said, the guide has already been quite opinionated up until now.

Evaluation of the Guide

I am pretty happy with how the guide turned out. I managed to find a diverse range of sources for most of the topics I wanted to cover. I kept close to the informal, “from developer for developer”-style of the original Angular guide [10] and I think it reads pretty well. The guide did end up being a bit longer then I planned (~50 A4 pages without figures and references). If I would write it again, I would choose a more narrow and clear scope for what the guide should include. For example, I might have excluded or shortened chapter 1 The Flutter Framework and instead focused even more on chapter 2 Architecting a Flutter App. I would have also liked to include more scientific sources on Flutter. But as it is still such a new Framework, next to no scientific research has so far been conducted on it. As of the writing of this guide, there has not been a single publication on the Flutter Framework through IEEE [95] or ACM [96].

All in all, I think I fulfilled the goal I set myself when starting this guide. To bridge the gap between the basics of Flutter and clean, structured Flutter development. The result is a resource I would have liked to have when bridging that gap myself.

Future of the Guide and Next Steps

This guide was commissioned by Capgemini Cologne [97] as part of their DevonFw [9] open-source initaive. It is supposed to aid Capgemini’s mobile developers when building large-scale application using the Flutter Framework. It will also be the basis for the Bachelor Thesis I am starting this November. In my Thesis, I will create and document a large scale Flutter application for DevonFw. It is supposed to be an example project that other developers can use as a guideline for creating their own large scale Flutter applications. More specifically, DevonFw has used the “My Thai Star” app [98] as a reference project for Angular [69]. I will create a Flutter version of it.

Suggestion for Future Works

One thing I and peers in the mobile development community [99] would like to see is more scientific research on the Flutter Framework: Lagre scale benchmarking, usability tests, scalability tests, architecture evaluation and so on.

My Thoughts on Flutter

I am guessing you could already pick up on my position regarding Flutter in the previous chapters. I am a Fan. The applications it produces run remarkably smooth and performance is usually one of the main drawbacks of cross-platform frameworks [100], [101]. The trade-off used to be less development cost for less performance. With Flutter this is no longer the case.

I was originally not a fan of how State Management [12] in Flutter is handled. I am coming from native Android [16] development and the switch from imperative to declarative thinking took a little while for me. I now understand that the design of Flutter has one central goal: Maximize performance [102]. And the way that State is handled is a direct consequence of that goal. State Management also becomes a lot less of an issue once you start following a clearly defined architectural style. I had to build a few iterations of Wisgen [11] to figure this out for myself. But once I implemented it with BLoC [7], I began to understand how Flutter could work in a large scale application.

Another great thing I noticed during the writing of this guide is how open the Flutter Team is towards the community. The Framework is completely open-source and many features have been implemented by developers outside of Google. The Flutter team is also very active on social media, with its own weekly podcast and Youtube channel [103].

One thing I don’t like about Flutter is how easy it is to write ugly code. I already touched on this topic in the last chapter. Having deeply nested code is, to a degree, unavoidable in Flutter. Splitting Widgets into smaller parts helps but adds a lot of boilerplate. I hope there will be a better solution to mitigate this problem in the future.

Another thing I am not a fan of is dependency injection in Flutter. At the moment the only solution is to inject dependencies from the UI into the BLoC. This created an ugly dependency between UI- and Data-Layer. But Google is currently working on a solution for this, so the problem is known and will be addressed [104].

All that being said, I won’t be going back to native app development any time soon.

References >

Back to Top