IT 21 060

Examensarbete 15 hp
Juli 2021

Hello Framework! A heuristic

method for choosing front-end
JavaScript frameworks

Mitra Bauer

Institutionen för informationsteknologi

Department of Information Technology
 Abstract
Hello Framework! A heuristic method for choosing
front-end JavaScript frameworks
Mitra Bauer

Teknisk- naturvetenskaplig fakultet

UTH-enheten This thesis investigates how to decide what JavaScript framework to use
when the front-end world is constantly changing and new technologies are
Besöksadress: added. Choosing the right framework can save a developer or company
Ångströmlaboratoriet
Lägerhyddsvägen 1 time, money and reduce complexity. An interview study with developers
Hus 4, Plan 0 was done to extract what is important in a framework. The interview
study concluded in six themes regarding what to investigate when
Postadress: choosing a framework along with a checklist and in-depth questions.
Box 536
751 21 Uppsala Based on the results from the interview study together with a test
application design, four web applications were implemented in the
Telefon: JavaScript frameworks React, Angular, Vue and Svelte. This was followed
018 – 471 30 03 by a qualitative evaluation to see if this investigative approach
Telefax: provided meaningful insights.
018 – 471 30 00
The test application results show that the combination of in-depth
Hemsida: questions and design specification gave a basic technical understanding
http://www.teknat.uu.se/student
of the strengths and weaknesses of the frameworks together with an
understanding of their history and current development. The goal of this
thesis is to provide more ways of comparing frameworks in the front-end
communities and academic world.

Handledare: Mikael Laaksoharju, Mattias Hising

Ämnesgranskare: Mikael Laaksoharju
Examinator: Johannes Borgström
IT 21 060
Tryckt av: Reprocentralen ITC
Contents
1 Introduction 1
1.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Delimitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Background 3
2.1 Concepts within JavaScript development . . . . . . . . . . . . . . . 3
2.2 Front-end frameworks . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.1 React . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.2 Angular . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2.3 Vue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2.4 Svelte . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Related work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

3 Interview study 7
3.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.1.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.1.2 Analysis method . . . . . . . . . . . . . . . . . . . . . . . . 7
3.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

4 Test application design 13

4.1 Suggested application specification . . . . . . . . . . . . . . . . . . 14

5 Evaluation 16
5.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.2.1 React . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
5.2.2 Angular . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.2.3 Svelte . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.2.4 Vue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
5.3 Framework differences . . . . . . . . . . . . . . . . . . . . . . . . . 25

6 Discussion 27
6.1 Interview study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
6.2 Test application design and Evaluation . . . . . . . . . . . . . . . . 27
6.3 Choosing a framework . . . . . . . . . . . . . . . . . . . . . . . . . 29

7 Conclusion 30
7.1 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

Appendices 33

ii
A Interview Questions 33

B Checklist 34

C In-depth Questions 35

D Test Application Specification 36

i
1 Introduction
In the world of JavaScript there has been many new frameworks created for the
purpose of helping developers create better applications. [1] Each new framework
comes with their own background, set of rules and structures. How do we know
which framework to choose in this vast array of frameworks? As a developer, what
are the different criteria we need to be aware of when making this decision? Why
are there new frameworks constantly on the rise and what makes a framework die
or live?
While some projects are still using pure JavaScript, it might not always work
for the given user requirements in a project. Advanced topics such as state man-
agement, routing, data handling and manipulation are some of the problems being
solved for developers by using frameworks. Many of today’s frameworks are adopt-
ing single-page application principles [2] which enables users to navigate easily
between different web pages without waiting for them to reload whole new pages.
There has been quite a bit of work done in benchmarking the performance of
JavaScript frameworks [3,4] but not as much investigation into what other aspects
of a framework is interesting to look at from a developer’s perspective. This thesis
is about how to think when choosing a front-end framework for a project and is set
up in two parts. Part one is an interview study done with experienced developers
to extract what their experiences with different frameworks are and what qualities
are important. The interviews are followed up by an analysis resulting in listed
criteria and in-depth questions. In the second part a test application was built in
the top four most used JavaScript frameworks [5] to see how well the frameworks
answers some of the criteria and in-depth questions, as well as understanding their
similarities and differences. The test application and frameworks are evaluated
and followed by discussion about the applications, frameworks and the interview
study. The work will result in more ways to evaluate a framework and a checklist,
in-depth questions and specification are available in the appendices.

1.1 Purpose
The goal of this thesis is to give insight into what is sought after in a JavaScript
front-end framework, mainly from a developer’s perspective, and to suggest which
framework of today’s top 4 most used frameworks matches the developers’ expec-
tations. The test application together with the interviews resulted in a checklist,
list of in-depth questions and test application specification that is provided by
the end. Another purpose is to add a hypothetical way to inspect and choose
frameworks, contributing to the front-end communities and other academic works
related to this field.

1
1.2 Delimitations
Delimitations for this thesis is that the tested frameworks are JavaScript focused
and the test applications were developed for the web platform. The versions of the
frameworks were the latest stable releases as of May 2021.

2
2 Background
Front-end development is a fast-paced field of work where all types of helpful li-
braries are constantly being developed. Many of the frameworks today are written
for JavaScript, today’s most used scripting or programming language that develop-
ers use to create interactive web pages and applications. [6] Different frameworks
and libraries have different purposes. Some are for structuring the styling in a
more efficient way. Some are just to help developers structure their JavaScript
code in a set of rules with functions abstracted and working in the background.
For small to mid-sized projects one might not have to use any libraries but they
turn out useful when there are user requirements involved and a more complex
architecture planned out. There are also options for cross-platform development
meaning one code base will work on multiple platforms and operating systems.

2.1 Concepts within JavaScript development

Component architecture. Components provide a way to organize an applica-
tion. Different frameworks may come with different ways to set up a component’s
logic, HTML and styling.

Document Object Model (DOM). The DOM is the data representation of

the objects that comprise the structure and content of a document on the web. It
represents the document as nodes and object in a way so programming languages
can connect to the page, serving as a programming interface for HTML and XML
documents. Programs can change the document structure, style and content. [7]

Single-page application (SPA). SPA works in the way that parts of a web
application are being rerendered on user input instead of reloading entire pages,
which results in faster transitions and a more smooth application experience. [2]

State management. An application that allows users to show, add and remove
content needs to track and update the underlying data. This underlying data is
known as state. State can for example live locally in a component or kept in some
kind of global state that components or pages can access in certain ways. [8]

Reactivity. Every time state or data changes the user interface (UI) needs to
be updated and match the data. Different frameworks have different syntax and
methods to make the UI reactive. [8]

3
Reusability. Code reuse has many benefits and allows you to write less code.
Other benefits are better readability of code, better structure of code and reduced
testing effort. Ways of reusing code could be extracting classes and methods when
opportunities arise, resulting in less errors or inefficiencies. The avoidance of du-
plication through extracting methods makes code more maintainable as any future
modifications only need to be applied in one place and prevents inconsistency when
there are multiple sections of code that needs to synchronize changes. [9]

Routing. There are several ways of navigating in a web application. Server-side

routing lets your browser communicate with a server and fetch new content to
display, which makes the URL in the address bar change. The browser remembers
the navigation history and allows for navigation back and forth.
Modern web applications or SPA load a single HTML shell and continually
update the DOM without navigating users to new addresses on the web. When an
SPA is complex and renders enough unique pseudo-webpage views, it’s important
to bring routing functionality. When routing is handled in the client application,
it’s called client-side routing. [8]

2.2 Front-end frameworks

While you can still write JavaScript code without any helping frameworks, many
developers today are working with frameworks as a tool to help them manage code
structure, performance and reusability. From State of JS surveys [5], a survey
answered by 23,765 developers the top four used JavaScript frameworks are React,
Angular, Vue and Svelte. They are all adopting single-page application (SPA)
principles.

2.2.1 React
The first version of React was released in 2013 and is owned by the social media
company Facebook. [10] It is an open-source library and is maintained by both
Facebook and its community of developers and companies. React includes state
management and rendering state to DOM but does not include routing by default
as well as certain client-side functionalities. It is the top most used framework for
building JavaScript user interfaces in 2020 and ranked second place in satisfaction
among users. [5]
React’s key feature is the composition of components and that components
written by different people should work well together. Other design principles are
common abstraction meaning they avoid to add features that can be implemented
by users hence less useless library code. Stability, interoperability and good devel-
oper experience is also at the core of the React framework development. [11]

4
2.2.2 Angular
AngularJS started out as a side project of a Google employee [12] and was the
second most used JavaScript framework in 2020 and ranked eighth place in sat-
isfaction among users [5]. It was first released 2010 but was scrapped and com-
pletely rewritten in 2014-2015. The latter version was renamed to just Angular
and is in fact a TypeScript based open-source library. TypeScript is a superset of
JavaScript.
Angular suggests applying the single responsibility principle to all components,
services and other symbols. One example is defining one service or component per
file and making sure to define small functions. Angular has its suggested style
guide to follow and their design principles can be understood by implementing the
style guide. [13]

2.2.3 Vue
Vue.js or Vue is an open-source library made to be adoptable and is able to scale
between a library and a framework depending on different use cases. The initial
release was in 2013 and is maintained by the creator and a team of core members.
[14] It was the third most used framework for building JavaScript user interfaces
in 2020 and also ranked third in satisfaction among users. [5]

2.2.4 Svelte
One of the most recent JavaScript frameworks is Svelte which was first released
in 2016 and rewritten in TypeScript 2019. [15] Svelte converts an application into
JavaScript at build time rather than interpreting it at run time. In the State of
JS 2020 survey it quickly got the highest satisfaction score, ranking number one
over React, Angular and Vue. It is still on the rise, ranking as the fourth most
used framework for building JavaScript user interfaces. [5]

2.3 Related work

There has been a few studies done on the performance of different JavaScript
frameworks but so far not many that are including Svelte. A study on React,
Angular and Vue had the research question Which JavaScript framework is most
suitable in terms of performance, modularity and usability for a single-page appli-
cation? and came to the conclusion that React performed best due to its startup
time but was very similar in terms of performance. React also stood out in modu-
larity, but Vue was the suggested framework when it came to documentation and
learning curve. [4]

5
 Yet another study from 2019 was investigating React, Angular and Vue. In
terms of popularity and community React ranked highest. Vue took first place
when it came to performance and learning curve. Angular was suggested as the
most fitting framework when it comes to building large scale applications. [16]
Going back one more year to 2018, another study looking at React, Angular and
Vue suggested that the three do not stand far apart from each other. Angular was
raised having a high TypeScript appreciation, emphasising guidance and structure
and being object-oriented. Vue had a shallow initial learning process, emphasis on
new popular technology and separation of concerns in one file. React was suggested
to have a strong focus on using JavaScript, a large ecosystem and high importance
of flexibility. [17]

6
3 Interview study
There is not much research done on how to choose frameworks holistically. De-
pending on what kind of project it is or what competencies there are in a team,
choosing technologies can be tough - sometimes simply the more familiar frame-
work is chosen because of initial efficiency and ease of use. There might not be
time to look into a certain framework or spend time on learning it. But if there is
time and interest, what approach can one take and what should one look for?
The goal of the interviews was to find some general questions to answer and
topics to check when researching frameworks for both personal and company use.

3.1 Methodology
The interviews were qualitative semi-structured 30 minutes long sessions. Semi-
structured interviews allow for more flexibility where the researcher can follow
topics of interest during the interview without sticking to a structured set of ques-
tions, rather using the prepared questions as guidance to keep the interview on
track. [18] The participants were interviewed one at a time totaling in four in-
terviewees, held at the interviewees’ workplace. Most questions were prepared
in advance but there was also space for general discussions and following up on
statements. Questions can be found in appendix A.

3.1.1 Participants
The interview study was done with four developers between the age of 24 and 42
with several years of programming experience. All worked at the same company,
two of which had previous experience from other companies. All were front-end
development oriented. Gender is not relevant for these in-depth questions and
hence is not reported.

3.1.2 Analysis method

The interview results were analysed using Thematic Analysis which emphasizes
identifying, analysing and interpreting themes within qualitative data. [19] There
are six phases of Thematic Analysis:

1. Familiarising yourself with your data: Transcribing, reading and noting down
initial ideas.

2. Generating initial codes: Systematic coding of interesting features across the

entire data set.

7
 3. Searching for themes: Collating codes into potential themes and group the
data.

4. Reviewing themes: Checking if the themes work and generating a thematic

map.

5. Defining and naming themes: Refine specifics of themes and overall analysis.
Clear definitions and names.

6. Producing the report: Extracting examples, analysing them and relating

back to research questions.

3.2 Results
The thematic analysis resulted in six themes. They are reported here in the term
of what was said and questioned about each theme. Most of the questions are also
applicable when researching other technologies.

Community. Whether you are stuck or having difficulties solving a problem

or just want to learn more about a certain matter, it is helpful to have a well-
established community - this is one of the most important factors. If you are
looking to learn the framework properly it might be worth to look for good tutorials
or online courses.

• How much support can someone find, both from official documentation and
developers’ experiences?

• If the framework is available on GitHub, what does the activity and history
look like? What can you see in the comments?

• What has been said on Stack Overflow [20] or other articles?

Quotes from interview:

... I would say the most important is community. To a certain extent

also who is creating the framework, there’s a difference to take a small
company’s framework or a smart person’s idea compared to Facebook’s
framework.

... if there are many users you could kind of assume that the commu-
nity is big enough and that’s a sign of that it (the framework) is being
maintained.

8
Maturity. To check how developed a framework is might be a good idea - choos-
ing a framework that is new or not entirely stable could end up being more work
than expected. Some new frameworks do not have big communities, so it is up to
you whether you like a challenge or want to have as much support as possible.

• Is the framework mature?

• Is it still being developed?

• Are there existing bugs?

• How many are working on the framework and are they developing it in the
same direction?

• What is the code’s philosophy?

• How is the documentation and GitHub history or activity?

Quote from interview:

I don’t invest a lot of time in something that hasn’t been around for a
while, with React you understand it’s groundbreaking and here to stay.

Technical aspects. Writing down your technical requirements is a good way to

summarize what qualities you are looking for. In the interviews multiple people
expressed that a good framework should save time, remove complexity and help
them get things done.

• What are the functions and features of the framework?

• How is the code built?

• What platforms does it support?

• How are the debugging options?

• What is the size of the framework and does it matter?

• What plugins are there and is the framework expandable?

Quotes from interview:

It’s more important to have the opportunity to do anything rather than

using a certain framework, if it has restrictions you lose a lot.

9
 A framework that is good at everything isn’t the best at something.
Then it might be better to use a few smaller libraries and create your
own kind of framework.

A good framework is a small one with big possibilities. Efficient. It is

flexible, easy to create own things instead of just following a track and
if you don’t it all breaks.

A powerful framework is used by people and companies I can relate to.

It solves a problem I’ve had for a long time in my career and that I
never understood how easy it was to do. It is powerful and very few
succeed in doing so, so it’s like "I’ll bring a new concept to do these
things and it will save you time and make your life easier".

Practical use. Hands-on experience with a framework might be the deciding

factor as a developer gets to see how it actually works. Maybe it suits them
really well or they end up being more confused. A way to test it is to just try
implementing something small. Sometimes you do not have a lot of time to spend
on learning a framework and if a basic task feels too complicated, it might not be
worth looking into the framework any more depending on what you need it for.

• Is it easy to work with and understand?

• Is it fast to work with?

• What disadvantages and advantages can you find?

• Is it supported by your current development environment?

Quotes from interview:

... after comparing the two frameworks we chose the one we were used
to.

Use whatever is easiest to use. The technology is secondary to the user

experience if the frameworks aren’t completely useless, but when they
are equally good, it feels like the most important is to choose whatever
you can work fast with.

Is it easy to understand? I read the documentation in two hours, now

I can start building. I find those kind of frameworks really great, those
that are easy to understand and gets things done.

10
 I usually just take whatever seems coolest or most interesting after read-
ing the documentation, but if you don’t understand the documentation
on the first read it’s like - forget it, I will never test this. ... There can’t
be magical stuff and if the most important parts lack documentation,
then you know, this will be troublesome.

The most important part (in a framework) for me is how much time
I win, in experienced time or real time, in a so called developer iter-
ation. ... All tools that remove loads of troublesome difficulties like
reloading, including new files, everything really that removes complex-
ity and makes things go faster - that’s the tools I have adapted and
worked with and makes me a more efficient developer. It’s the most
important.

Project requirements. It is quite important to prepare requirements before-

hand to see the size of the project and what resources it needs. You may end
up finding that you need to do more than expected. If a change is needed after
starting the project implementation, you might need to be flexible and change
technology or requirements as needed.

• What will the project be used for?

• How will the framework help and what problems will it solve?
• Does it support your current technologies?
• If you are contemplating adding a framework to an existing project, is it easy
to add and apply?

Quote from interview:

I tend to argue for those that are, instead of these huge frameworks,
smaller libraries. That are easier to switch, easy to add, because often
you don’t start from scratch and things already exist. You have some
code, or a system that already exists and if you add new functionality -
you can’t just rewrite everything. However it easy to add a small part
in a system.

Company questions. Choosing the right technologies can improve the product
delivery time and make the product easier to maintain. A project often involves
more than one person, which makes choosing technology not only a technical
question but a personnel question. It might also be important to check whether
the framework is open-source and what license it has to avoid future complications.

11
 • How many in the team know the framework and how many would need to
learn it?

• Can one work productively with the framework and will it affect production
speed?

• What will it cost to use this technology?

• Will the code be reusable or shareable between projects?

• What current problems will it solve and does it support the current tech-
nologies that are being used?

Summarized checklist of topics to investigate

 Community

 Documentation

 GitHub history

 GitHub activity

 Functions/features of the framework

 Support for multiple platforms

 Plugins

 Tutorials

 Existing bugs

 Size of framework

 Possibility to expand framework

 Stack Overflow questions/activity

 Articles

 Support for your development environment, tools, IDE

 Debugging options

12
4 Test application design
To further look into the Technical Aspects, Practical Use and Community part of
the interview study results a test application was built. The test application and
other research resulted in a final checklist, in-depth questions and specification to
use when investigating frameworks, available in appendices B, C and D.
The inspiration for the test application design is based on a combination of the
results from the interview study and today’s most used social media applications
which normally consists of some feed page and lists of data. The application design
needs to address some in-depth questions from the interview study but also let the
developer get enough experience with the framework to feel confident in making a
technical decision. For this application the scope will be:

• Handling of libraries - How do we manage project libraries?

• Reactivity - In what way does this framework support reactivity?

• Reusability - What components are reusable, in what way?

• Component architecture - Is there a component architecture setup or recom-

mended?

• Data handling - How do we fetch and handle data in this framework? How
is data passed?

• State management - Is state management included and how does it work?

• Styling - How does styling in terms of scoping and conditionals work?

The base application was implemented in React, Vue, Angular and Svelte. All
variants were built for the web platform in mind. The first application was imple-
mented in React as a template app for the other frameworks. I chose to start with
React because of previous work experience. As for guidelines and best practices,
as much as possible was followed from official documentation and otherwise from
other community resources such as Stack Overflow or GitHub threads. Setting up
the projects was also done by official recommendations and documentation. The
coding paradigm was functional if applicable. Since some frameworks were new in
terms of my own experience I also drew inspiration from community websites.
Another way to test the test application design is to give a simple specification
to a developer and see how well they manage to implement it. As for the applica-
tion built in Vue, the specification was handed to another experienced developer
to see how the proposed design was built and afterwards I gathered feedback for
more input on the specification.

13
 Finally, these are the questions from the interview study that should be an-
swered by implementing the specification:
• Is the framework mature?

• Is it still being developed?

• Are there existing bugs?

• How is the documentation and GitHub history or activity?

• What are the functions and features of the framework?

• What platforms does it support?

• How are the debugging options?

• What plugins are there and is the framework expandable?

• Is it easy to work with and understand?

• Is it fast to work with?

• What disadvantages and advantages can you find?

• Is it supported by your current development environment?

• What will it cost to use this technology?

4.1 Suggested application specification

• Navigation bar:

– Links to Home and Feed page where Home is shown at root path (/)
– Theme toggle with theme value in a global state, available themes
should be default and dark theme

• Home Page:

– Title
– Logo of some sort
– Link to framework homepage

• Feed Page:

– Feed component consisting of image, title and id or likes

14
 – Simple error handling, showing a text if data fetching request fails
– Data fetching from some open-source REST API should be paginated
and kept in a local state
– List of feed items with infinite scrolling and a loading spinner at the
bottom, spinner should be from a third-party library
– Routing/navigation library if not built-in

Other guidelines:

• Coding style as functional as possible

• Follow project setup from official documentation

• Feel free to use inspiration from community websites

15
5 Evaluation
This section includes the evaluation of the test applications built in React, Angular,
Vue and Svelte. Both similarities and differences will be presented and the test
application design will be scrutinized. The way that this evaluation is done could
be one way for developers to investigate frameworks in the future, which is one of
the purposes of this thesis.

5.1 Methodology
A systematic evaluation will be done on the four test applications by looking at
the application criteria from Subsection 4 one framework at a time. That includes
handling of libraries, reactivity, reusability, component architecture, data handling,
state management, styling and finally the interview study in-depth questions.
Since the Vue implementation was outsourced to another developer some ques-
tions can not be answered in the same way as the other three that I implemented
myself. However outsourcing the implementation is important to answer questions
about how clear the specification is for someone who has not yet implemented
it and how well one gets to know a certain framework, evaluating whether the
specification is valid enough for testing a framework.

5.2 Results
Before diving into the separate framework evaluations I will share what was com-
mon or the same in all the frameworks given the application criterias. This is to
avoid repetition.

• Data handling: Used JavaScript’s Fetch API in all applications for fetching
data from the open-source API and it worked fine. The API provided images,
title and id.
• Handling of libraries: All frameworks used Node package manager (npm) for
managing libraries.
• Building the theme toggle I used community resources for inspiration and
all frameworks had articles on how to handle themes.
• Following the official documentations on setting up the application, each
framework project came with a landing page from which I was able to reuse
title, logo and link to framework website.
• As for debugging, I did not make use of any framework specific ways to
debug. I only used console logging and a network tab in the browser.

16
 • All frameworks supported having an error state variable that could condi-
tionally show an error text instead of the infinite scrolling list when a data
request failed.
• All frameworks are still being developed and have existing bugs but are still
mature enough to handle the requirements of the test application.
• All frameworks are free to use and under MIT license.
• All frameworks only support the web platform by default.

5.2.1 React
Following the official documentation it was an easy project setup with Create React
App, an officially supported way to create single-page React applications which
offers a build setup with no configuration. I added a third-party library called
react-router-dom for routing and navigation. For the infinite scroll I took inspi-
ration from community resources and found a third-party library for the loading
spinner using npm. Altogether the React application was a successful implemen-
tation since all specification requirements were met. After finishing the implemen-
tation the built bundle size was 235 KB and took 4.0 seconds to build. For a visual
presentation of the application, see figure 1 and 2.

Figure 1: Home page, default theme Figure 2: Feed page, dark theme

Reactivity. Using React’s state and prop variables allows components to update
whenever some state has changed. In this test application states were held locally
at page level and passed down as props to the smaller components.

Reusability. Components were reusable and mapped to data in the feed page.
A separate file was created for the list item component and imported in the page
file.

17
Component architecture. From own experience I decided to do a simple com-
ponent architecture with a folder for pages and one for components. This was not
set up from the start and had to be done manually.

Data handling. The data flow is unidirectional. The data was fetched from an
open-source API, put in a local state for the Feed page and updated when new
data was fetched since it was a paginated request.

State management. Local states were held in Feed page for the fetched data,
loading and error handling. Global state for the theme was done with React’s
Context API.

Styling. Create React App came with some default styling, I left some of it and
added some new styles in a global CSS file that came with the project setup.

In-depth Questions

• How is the documentation and GitHub history or activity?

The documentation and GitHub history is vast and activities like pull re-
quests are regularly showing up. There was also a lot of community resources
for the theme handling I researched. Currently there are 542 GitHub open
issues and 192 open pull requests. In Stack Overflow, searching for questions
with a react tag shows 313 294 results.

• What are the functions and features of the framework?

In the test application I got to make use of React’s state handling, lifecycle
methods, Context API for creating a theme provider and custom hooks.

• What plugins are there and is the framework expandable?

Yes, there are a lot of third-party libraries and I added one for the loading
spinner and one for routing. Searching for react in the npm package library
shows 162 771 results.

• Is it easy to work with and understand?

Yes but I have previous experience.

• Is it fast to work with?

Yes because of previous experience.

• What disadvantages and advantages can you find?

Advantages: Easy to split up components and creating custom hooks and
providers to separate features.

18
 Disadvantages: Could be that there is not a lot of suggestions on project
architecture.

• Is it supported by your current development environment?

Yes, I already had everything set up for React projects to work.

5.2.2 Angular
The official documentation for Angular proposed to use the Angular CLI (command-
line interface) to setup a project. When creating the project one gets prompted
if they want to use routing and what stylesheet format they would prefer out of
CSS, SCSS, LESS and Stylus. I chose to go with the built-in routing and CSS.
The Angular CLI generates a lot of boilerplate code when creating components,
such as recommended architecture, base tests and CSS files. I used the ng generate
command for creating modules, components and services.
Due to limited implementation time and difficulty to find community resources
on how to build a infinite scroll list I decided to use the third-party library ngx-
infinite-scroll. As for the loader spinner I could not find a library that supported
Angular 12 which was the version setup by the Angular CLI. I found that the
community resources were sometimes limited to a certain Angular version. In my
experience there was not as many resources as for React and Vue. Since Angular
was new to me I used a lot of community resources to get an understanding for
best practices and common design patterns. After finishing the implementation
the built bundle size was 709 KB and took 19.4 seconds to build. For a visual
presentation of the application, see figure 3 and 4.

Figure 3: Home page, default theme Figure 4: Feed page, dark theme

Reactivity. The Angular CLI set up the project with RxJS by default, a li-
brary for reactive programming using Observables that makes it easier to compose

19
asynchronous or callback-based code. [21] Whenever new data was fetched the
components were updated.

Reusability. I could reuse and map components using the *ngFor keyword but
the file structure is different from React and Svelte in the sense that I created the
list item in the same template file as the feed page.

Component architecture. With the Angular CLI the whole architecture is

set up for you and when creating new components or modules it organizes the
whole structure automatically. The component is separated with logics in a Type-
Script file, view in a HTML template file and styling can be separate or inside the
template file.

Data handling. Angular has a two-way data binding syntax, data can flow in
both directions. I did not make use of the two-way data binding syntax in the test
application. Data for the Feed page was stored in the component TypeScript file
and mapped in the HTML template file.

State management. Local states such as data, loading and errors states were
handled in the Feed TypeScript file and inserted in the HTML template file.

Styling. The styling was set up to use separate CSS files but there is also the
option to keep it directly in the template file. The styling is encapsulated into the
component’s view and does not affect the rest.

In-depth Questions

• How is the documentation and GitHub history or activity?

The documentation and GitHub history is vast and activities like pull re-
quests are regularly showing up. You just have to make sure you are looking
at the right Angular version. Currently there are 2075 GitHub open issues
and 172 open pull requests. In Stack Overflow, searching for questions with
a angular tag shows 255 558 results.

• What are the functions and features of the framework?

A few of the functions and features I got use in the test application were
Angular’s CLI, lifecycle methods, directives, dependency injections and An-
gular’s first party library for routing.

• What plugins are there and is the framework expandable?

Yes, there are a lot of third-party libraries but a lot of them are bound to

20
 a certain Angular version. Unfortunately I could not find a loading spinner
for Angular 12, but I did add a library for the infinite scroll list that worked
well. Searching for angular in the npm package library shows 50 845 results.

• Is it easy to work with and understand?

Yes after some time but it took some rereading to grasp certain concepts.
To me the learning curve felt steep. I had to go back to the documentation
multiple times to grasp certain concepts, check syntax and how to structure
my code.

• Is it fast to work with?

Quite fast to work with when you get used to the rules and using the Angular
CLI.

• What disadvantages and advantages can you find?

Advantages: Angular CLI sets everything up for you and creates boilerplate
code when adding components. Clear directions on code structure and how
to split logic and view.
Disadvantages: If you do not use the Angular CLI it will be a lot of boilerplate
code to write. There is also some extra work to add custom elements.

• Is it supported by your current development environment?

Yes, I already had everything set up for Angular projects to work.

5.2.3 Svelte
Following the official documentation and setting up a new project was easy. For
Svelte I had to install a plugin for my editor VS Code to syntax highlight the
code. To me the syntax was the most different among the four frameworks. I
added a third-party library for routing called svelte-routing. Even though Svelte
is the newest of the four there is many packages available and it was easy to find
one for the loading spinner. After finishing the implementation the built bundle
size was 156 KB and took 1.7 seconds to build. For a visual presentation of the
application, see figure 5 and 6.

Reactivity. Svelte is using the dollar sign ($) for reactive declarations to indicate
dynamic values that will rerender the component when state has changed.

Reusability. Components were easy to create, import and use in other files. For
me it was a similar mindset to React.

21
 Figure 5: Home page, default theme Figure 6: Feed page, dark theme

Component architecture. The Svelte documentation had some recommenda-

tions - each component file can have a script tag containing the variables, lifecycle
methods and other logic. This is followed by HTML code and CSS styling in a
style tag, kept in the same file. Some logical functions such as contexts and stores
are separated into their own files. This structure seems similar to Vue, but in
Svelte you do not have to use the template tag to write HTML in the Svelte files
as you do in Vue.

Data handling. As a general rule, data flow in Svelte is top-down - a parent

component can set props on a child component, and a component can set attributes
on an element, but not the other way around.

State management. State management is included in the framework by using

stores and contexts. Contexts and stores seem similar but they differ in that stores
are available to any part of an app, while a context is only available to a component
and its descendants.

Styling. CSS rules are scoped to the component. However in my experience

some computed styling did not change with the global tag used for creating or
overriding global styles.

In-depth Questions

• How is the documentation and GitHub history or activity?

The documentation is simplistic but covers most functions and features
through a thorough tutorial. In GitHub there are pull requests regularly
showing up but compared to Vue, Angular and React there was less activity.

22
 Currently there are 628 GitHub open issues and 141 open pull requests. In
Stack Overflow, searching for questions with a svelte tag shows 1870 results.
• What are the functions and features of the framework?
In the test application I got to make use of Svelte’s lifecycle methods, reactive
declarations, syntax for conditional rendering, stores and contexts.
• What plugins are there and is the framework expandable?
Yes, there are a lot of third-party libraries and I added one for the loading
spinner and one for routing. Searching for svelte in the npm package library
shows 2929 results.
• Is it easy to work with and understand?
Even though Svelte was quite different from my previous experiences the
official tutorial made it easy to understand and learn. Learning curve was
gentle. I found it easy to get the project running and implementing the
specification.
• Is it fast to work with?
Yes. Developer experience was good but the website crashed from time to
time when reloading the page so I needed to rebuild the project.
• What disadvantages and advantages can you find?
Advantages: Easy and fast to learn. Small bundle size. Good examples and
tutorial on their official website. Simplistic and clear coding style.
Disadvantages: Still quite new and less resources than the other frameworks.
• Is it supported by your current development environment?
I had to install a plugin for syntax highlighting but other than that everything
worked fine.

5.2.4 Vue
This application was outsourced to another experienced developer. The Vue CLI
was used to initiate the project and you choose whether to use state management,
routing, stylesheet framework, linting, compiler and more. Folder structure was
also created by the CLI. The specification requirements were met, although the
styling looked a little different from my implementations the functionality was the
same. The built bundle size was 180 KB and took 1.2 seconds to build. For a
visual presentation of the application, see figure 7 and 8.
The feedback from the developer was that they felt it was a clear enough
specification to get a feel for the framework. It would be good to avoid unnecessary
styling and the instructions for the error handling felt a bit unclear. The time it
took to build the application was a bit longer than expected.

23
 Figure 7: Home page, default theme Figure 8: Feed page, dark theme

Reactivity. Using Vue’s state and prop variables allows components to update
whenever some state has changed. In this test application states were held locally
at page level and passed down as props to the smaller components.

Reusability. A few components were made and easily imported and reused in
the page files.

Component architecture. The component architecture is similar to Svelte in

the sense that logic, view and styling live in the same file. However the component
setup looks quite different in the way methods, data, components and lifecycle
methods are instantiated or injected into the template.

Data handling. Two-way data binding is supported for input components. Data
was fetched in Feed page and kept in a local state.

State management. A store containing the states for the theme toggle was
created in this project.

Styling. Vue has a scoped attribute one can use to scope the CSS styling to the
file.

In-depth Questions
• How is the documentation and GitHub history or activity?
The documentation and GitHub history is vast and activities like pull re-
quests are regularly showing up. However since there is a new Vue version
rolling out (3.0) the version used in the application (2.0) does not have the
same frequency of pull requests opened. Currently there are 327 GitHub

24
 open issues and 203 open pull requests. In Stack Overflow, searching for
questions with a vue tag shows 79 642 results.

• What are the functions and features of the framework?

State management, routing, lifecycle hooks were used in this project.

• What plugins are there and is the framework expandable?

There are many plugins and libraries for Vue, for this project vue-spinner
library was added. A routing library was already included in the ecosystem.
Searching for vue in the npm package library shows 50 381 results.

• Is it easy to work with and understand?

Not possible to assess due to outsourcing the implementation.

• Is it fast to work with?

Not possible to assess due to outsourcing the implementation.

• What disadvantages and advantages can you find?

Advantages: Allows partial adoption of functionality. Possible to build pure
HTML and CSS pages.
Disadvantages: Vanilla JS can be used but may affect reactivity. Using
global state complicates the use of two-way data binding.

• Is it supported by your current development environment?

I had to install a plugin for syntax highlighting.

5.3 Framework differences

Some of the differences between the frameworks were that Svelte and Vue has a
similar component architecture in the sense that logic, HTML and styling live in
the same file. An Angular component is split up into three different files, one
for HTML, one for styling and one for logics. React gives you the freedom to
choose whatever component architecture you want but in this application logics
and HTML were in the same file and CSS was separated.
Svelte had scoped CSS styling in each file by default, while Vue has a scoped
attribute one can use to scope the CSS styling to the file. React does not have
scoped styling by default and the application was just using one global CSS file
for all components and pages. Angular CLI set up the components with separate
component-scoped CSS files.
Two-way data binding syntax were supported in Vue and Angular while Svelte
has a bind directive for it.
Out of the two new frameworks I got to learn, Angular had a rather steep
learning curve in comparison with Svelte that had a gentle one.

25
 React and Angular projects were already supported by my editor while I had
to install syntax highlighting plugins for Vue and Svelte.
Svelte had the smallest online community and React had the largest out of
the four frameworks. Svelte also had the fewest available libraries and packages
through npm, while React had the most. The Angular application was the only
one that did not use a third-party library for the loading spinner because of version
incompatibility.

26
6 Discussion
In these subsections I will discuss the results from the Interview study, Test ap-
plication design and Evaluation and finally some discourse on which framework
could be the better choice.

6.1 Interview study

First of all I would like to highlight some points that could affect the results. There
might have been some bias in the responses since the interviewees all worked to-
gether at the same company, which could result in similar answers. Doing semi-
structured interviews could sometimes result in discussions that had leading ques-
tions, which could also affect the answers and maybe something would be brought
up in just one interview. In those cases it can be difficult to validate answers as
something of heavy weight. Another bias could be the interviewees’ tendency to
raise React as an example because of their strong experience with it. This could
mean their responses and preferences already leaned towards something similar to
the features and mindset of React.
I found the common theme to be that a framework should improve the de-
veloper experience with environments and tooling, in addition to the main point
which is improving code quality and adapting functions and code patterns that
feel easy and efficient to use. There was a big emphasis on community and having
good documentation along with other community resources. It was interesting
that also the philosophy and founder of frameworks were brought up, an ethical
aspect of using technology. It could be a good idea to check who or what is behind
a framework to know what we support by using it.
Outside of the technical aspects project requirements and company questions
are important in choosing a framework. This might not matter too much if the
project is personal but in a company setting choosing something that might not
live up to expectations can cost time and money.

6.2 Test application design and Evaluation

Since I already had experience with React it was quite easy to implement the first
application. I thought it would be mostly copying and pasting the code to create
the other applications but each framework had their specific syntax and structures
to follow so I had to spend more time learning about them than expected. React
was quick in building the production build but had the second largest production
build size. The production build sizes for Vue (180 KB) and Svelte (156 KB)
were quite close though. The one that stood out the most was Angular with a
production build size of 709 KB, even after optimizing the build command with

27
options to reduce the size. In terms of production build speed Vue was the fastest
with 1.2 seconds.
With Angular I experienced it was more difficult to adapt their principles
and also finding the right information, I wonder if that is because of my previous
experiences being used to finding information related to React. I did also encounter
a few community resource solutions that would not work in Angular because of
versioning. Overall, when you got used to their style guide it got easier but still
felt a little clumsy in comparison to React and Svelte. It was slow in building the
production build (19.4 seconds) and had the largest production build size. Angular
CLI solved a lot of boilerplate code which was quite nice though, it seems like it
could scale well when a project is growing.
Svelte was very easy to learn and fast to work with, it will be interesting to
see how it grows in the developer community. Even though Svelte is rather new
and does not have a big community like the other three frameworks I found all
the resources I needed to complete the test application design. Svelte was fast in
building the production build and was the build size was smaller than the other
three frameworks. For Svelte it would have been interesting to try SvelteKit, their
latest framework for building web applications, but it is in beta stage. Maybe
using SvelteKit would have resolved some of the reloading issues I encountered.
Since I did not implement the Vue application it was quite difficult to evaluate it
more than saying that it met the requirements. I did not get to know the functions
and features of the framework, but, I got to see the specification requirements met
and that could be a sign of the test application design being successful in the end.
For the specification, two-way data binding was not compatible with the theme
toggle input that was dependent on a global state. It could be a good idea to add
something to the specification that makes use of two-way data binding, perhaps a
form.
I chose not to add any code examples in the evaluation because of two major
reasons. The first reason is that I felt like the actual code is not the focus of the
thesis, rather more of a high-level evaluation of certain topics. The second reason
is that with front-end development there usually are many ways to solve the same
problems and with framework standards constantly changing it could mean my
solutions would be obsolete in the near future.
To me the test application design was enough to get a basic understanding of
the frameworks but there was definitely functions and features that I did not get to
implement. In combination with the in-depth questions from the interview study
I got a good understanding of their features, history and development activity.
For a bigger study, a more elaborate specification that implements more of the
framework features would be interesting.

28
6.3 Choosing a framework
Short summaries of the frameworks:
• React - Has the largest online community and strong backing. There are
loads of libraries, plugins and a lot of GitHub and Stack Overflow activity.
It ranked second place in satisfaction and first place in usage according to
the State of JS survey.
• Svelte - Has the smallest community. The framework is new and fast, pro-
duces a small bundle size and tutorials are thorough. Learning curve is
stable. Good for smaller projects but how will it scale in larger applications?
It ranked first place in satisfaction and fourth place in usage according to
the State of JS survey.
• Angular - Has a large community but ranked ninth place in satisfaction and
second place in usage according to the State of JS survey. It has a strict set
of rules and out of the four frameworks the only framework that enforces the
use of TypeScript. Angular has the most open GitHub issues.
• Vue - Has a medium sized community. It ranked third place in satisfaction
and usage according to the State of JS survey. Vue has the fewest open issues
on GitHub out of the four but the most open pull requests which could be a
sign of mature framework in development.
Most of the developers in the interviews wanted the freedom to have a smaller
framework with the opportunity to expand it with other plugins or libraries. Con-
sidering that, maybe Angular would not be the best choice as it comes with a lot
of strict rules. We can also see that it has lost popularity in the State of JS 2020
survey. Vue and Angular both have similar directives and component architecture,
maybe Vue could be the more up-to-date choice between the two. React wins in
the race of community support and available resources and could be the safest
choice ranking high on usage and satisfaction. Svelte is a new contender, maybe
more appropriate for personal or smaller projects since the community and online
resources still are rather small. Open issues on GitHub is rather high compared
to React and Vue that have been around for a longer time. The high satisfaction
score could indicate that it will grow larger and more stable in the coming years.
Comparing the results with related work in section 2.3 they seem similar. Ac-
cording to the studies, React was best in the performance but Vue suggested as
better when it came to documentation and learning curve. Another study claimed
Vue had a better performance and raised React for being more modular. In com-
parison Angular did not stand out in any of these areas which supports the drop
in satisfaction rankings. It will be interesting to see if Svelte will be included in
more JavaScript framework comparisons in the future.

29
7 Conclusion
This thesis has provided a set of questions, checklist and specification to investigate
frameworks. The suggested way to evaluate a framework would be to implement
the specification in appendix D together with researching topics in the checklist
and in-depth questions in appendices B and C.
Building the test application in the four frameworks React, Angular, Vue and
Svelte gave knowledge about the functions, features and developer experience. The
checklist and in-depth questions resulting from the interview study added to the
understanding of non-technical parts such as community resources and activity.
React seems to be a good and safe choice in regards to its community and online
resources size, in combination with the high ranking in developer usage and sat-
isfaction. Svelte has a gentle learning curve and is on the rise ranking at the top
of the satisfaction score but still needs more community support to grow in usage
and resources. Angular is still widely used but losing popularity in satisfaction
rankings and has a lot of open GitHub issues. Vue has some elements similar to
Angular but is ranking high in satisfaction and usage which could indicate that it
is a more appropriate choice if you want to use a similar framework with less strict
rules.
In the end, choosing a framework is about preference, experience and require-
ments. Each project needs to be evaluated on an individual basis. Hopefully this
thesis can aid in the choosing of frameworks.

7.1 Future work

What could be further tested or researched?

• More frameworks could be included in the test application evaluation, maybe

even testing cross-platform or native app frameworks.

• Investigate if there are other features that are important to test in addition
to the ones in this thesis.

• Create an even more complex specification to be able to answer more in-depth

questions and result in a more in-depth comparison of differences between
frameworks.

• Outsource coding to more developers and collect more feedback about the
test application specification.

• Extend the size of the interview study and get more data from developers
with different experiences.

30
References
[1] Wanyoike M. "History of front-end frameworks";. 2018. Accessed on: 2021-06-
10. [Online]. Available: https://web.archive.org/web/20210611092309/
https://blog.logrocket.com/history-of-frontend-frameworks/.

[2] "SPA (Single-page application) - MDN Web Docs Glossary: Definitions

of Web-related terms | MDN"; 2021. Accessed on: 2021-06-11. [On-
line]. Available: https://web.archive.org/web/20210616121411/https:
//developer.mozilla.org/en-US/docs/Glossary/SPA.

[3] Mariano CL. Benchmarking JavaScript Frameworks. Masters disserta-

tion. Technological University Dublin; 2017. Available: https://arrow.
tudublin.ie/scschcomdis/94/.

[4] Olsson N, Ockelberg N. Performance, Modularity and Usability, a Comparison

of JavaScript Frameworks. Bachelor’s dissertation. KTH, School of Electrical
Engineering and Computer Science (EECS); 2020. Available: http://urn.
kb.se/resolve?urn=urn:nbn:se:kth:diva-272109.

[5] State of JS 2020;. 2020. Accessed on: 2021-06-11. [Online]. Avail-

able: https://web.archive.org/web/20210411021418if_/https://2020.
stateofjs.com/en-US/technologies/front-end-frameworks/.

[6] Stack Overflow Developer Survey 2020;. 2020. Accessed on: 2021-06-11. [On-
line]. Available: https://web.archive.org/web/20210609125451/https:
//insights.stackoverflow.com/survey/2020.

[7] "Introduction to the DOM - Web APIs | MDN"; 2021. Accessed

on: 2021-06-15. [Online]. Available: https://web.archive.org/web/
20210616130639/https://developer.mozilla.org/en-US/docs/Web/
API/Document_Object_Model/Introduction.

[8] "Introduction to client-side frameworks - Learn web development

| MDN"; 2021. Accessed on: 2021-06-15. [Online]. Available:
https://web.archive.org/web/20210616131819/https://developer.
mozilla.org/en-US/docs/Learn/Tools_and_testing/Client-side_
JavaScript_frameworks/Introduction.

[9] Anthony RJ. Systems Programming: Designing and Developing Distributed

Applications, 1st ed. Morgan Kaufmann; 2016.

[10] "Release v0.3.0 · facebook/react"; 2013. Accessed on: 2021-06-11. [Online].

Available: https://github.com/facebook/react/releases/tag/v0.3.0.

31
[11] "Design Principles - React";. Accessed on: 2021-06-11. [Online]. Avail-
able: https://web.archive.org/web/20210611081105/https://reactjs.
org/docs/design-principles.html.

[12] "About Miško Hevery";. Accessed on: 2021-06-11. [Online]. Available:

https://web.archive.org/web/20210615143251/http://misko.hevery.
com/about/.

[13] "Angular coding style guide";. Accessed on: 2021-06-11. [Online]. Avail-
able: https://web.archive.org/web/20210611081240/https://angular.
io/guide/styleguide.

[14] "Releases · vuejs/vue";. Accessed on: 2021-06-11. [Online]. Available: https:

//github.com/vuejs/vue/releases/tag/0.6.0.

[15] "Releases · sveltejs/svelte";. Accessed on: 2021-06-11. [Online]. Available:

https://github.com/sveltejs/svelte/releases.

[16] Saks E. JavaScript Frameworks: Angular vs React vs Vue. Bachelor’s thesis.

Haaga-Helia University of Applied Sciences; 2019. Available: https://www.
theseus.fi/handle/10024/261970.

[17] Wohlgethan E. Supporting Web Development Decisions by Comparing Three

Major JavaScript Frameworks: Angular, React and Vue.js. Bachelor’s disser-
tation. Hamburg University of Applied Sciences; 2018. Available: https:
//reposit.haw-hamburg.de/handle/20.500.12738/8417.

[18] Wilson V. "Research methods: Interviews";. EBLIP [Internet]. 2012 Jun.12

[cited 2021 Jun.11];7(2):96-8. Available from: https://journals.library.
ualberta.ca/eblip/index.php/EBLIP/article/view/17196.

[19] Braun V, Clarke V. Using thematic analysis in psychology; 2006. Qualitative

Research in Psychology, 3 (2). pp. 77-101.

[20] "Stack Overflow - Where Developers Learn, Share, & Build Careers";. Ac-
cessed on: 2021-06-14. [Online]. Available: https://stackoverflow.com/.

[21] "RxJS";. Accessed on: 2021-06-12. [Online]. Available: https://rxjs.dev/.

32
Appendices
A Interview Questions
About you

• How old are you?

• How long have you worked or done programming?

• What type of programming do you like the most?

• How long have you been at Company X?

• Do you have development experience another company?

General questions

• How do you choose new framework?

• What are you looking for in a new framework? What areas and what do you
want to work for you?

• If you find two or more similar frameworks, how do you compare them?

• What is it that finally makes you choose a new framework?

What are you thinking about when you hear these words:

• Powerful framework

• Fidelity

• Developer experience

• Versatility

How important are these for you when you choose a technology from 1-5?

• Versatility

• Tools and IDE

• Debugging

• OS

33
 • Developer Experience

• App quality

• Possible users

• App production cost

• Security

• Platform compatibility

• Easy to update

• Time to market

• Third party plugins

• Performance

• Community

• Documentation

• Easy to implement

• Fast to implement

• Reusability

• Easy to implement in existing project

B Checklist
Based on the interview study and results from the test app, these are some of the
following things to check when deciding on what framework to use:

 Community

 Documentation

 GitHub history

 GitHub activity

 Functions/features of the framework

34
  Support for multiple platforms

 Plugins

 Tutorials

 Existing bugs

 Size of framework

 Possibility to expand framework

 Stack Overflow questions/activity

 Articles

 Support for your development environment, tools, IDE

 Debugging options

C In-depth Questions
• Is the framework mature? Are there no more major bugs?

• Is the framework still being developed? How many are working on it? What
kind of people are developing it? Are they close or dragging the framework
in different directions?

• What do people say about the framework?

• How is the code built? What is its philosophy?

• How is the documentation/GitHub history and activity?

• What are the functions and features of the specific framework?

• Is it easy to create own plugins? Is the framework limited somehow?

• Is the framework modular?

• What kind of project will the framework be used for?

• Will the code be shareable between projects?

• Is the framework easy to add and apply to an existing project?

35
 • Try implementing something, what works well? Is it fast to use? Easy to
work with?

• Is the code easy to understand?

• How much time will be saved using the framework?

• Is it open source? Licensed?

• Does it support your current technologies?

Company focused questions:

• How many (in the team) know the framework?

• How many would would need to learn the framework?

• Can one work productively with the framework?

• Will the code be reusable and can it be shared between projects?

• What will it cost to use this technology?

• How is the framework going to help? What problems will it solve?

• Does it support your current technologies/libraries/frameworks?

• Will it affect production speed?

D Test Application Specification

• Navigation bar:

– Links to Home and Feed page where Home is shown at root path (/)
– Theme toggle with theme value in a global state, available themes
should be default and dark theme

• Home Page:

– Title
– Logo of some sort
– Link to framework homepage

• Feed Page:

36
 – Feed component consisting of image, title and id or likes
– Simple error handling, showing a text if data fetching request fails
– Data fetching from some open-source REST API should be paginated
and kept in a local state
– List of feed items with infinite scrolling and a loading spinner at the
bottom, spinner should be from a third-party library
– Routing/navigation library if not built-in

Other guidelines:

• Coding style as functional as possible

• Follow project setup from official documentation

• Feel free to use inspiration from community websites

37