The dashboard provides information on the outbreak and prevalence of COVID-19 in The Netherlands. It combines measured and modelled data from various sources to give a broad perspective on the subject.
If you want to contact the dashboard team, feel free to open an issue for technical questions, bug reports, or security findings. If you have a generic question or remark about the corona policy of the Dutch government, please consult the frequently asked questions or contact page on the dashboard.
The core team works directly from this open-source repository. If you plan to propose changes, we recommend opening an issue beforehand where we can discuss your planned changes. This increases the chance that we might be able to use your contribution (or it avoids doing work if there are reasons why we wouldn't be able to use it).
The project is set up as a monorepo and therefore the code is organized in multiple packages.
app
: The main application that contains the front-end part of the dashboard. READMEcli
: Command-line tools for things like data validation.cms
: Configurations for the Sanity content management system. READMEcommon
: Commonly shared code that multiple packages are using, like types and utils.icons
: A React icon component library, used by the CMS and by app. README
You can run these commands to quickly get started. We advise you to read what's happening behind the scenes by reading the app README
$ yarn
$ yarn bootstrap
$ yarn dev
In this project, we use Yarn instead of NPM, so the documentation assumes you
have the yarn
executable installed on your system.
If you would like to run the code on your local machine check out the readme documentation of the app and (optionally the) cms packages.
Without describing in detail all the rules we tend to follow here are some worth noting:
-
All filenames are written in kebab-case.
-
We use named exports where possible. They improve typing and help with refactoring.
-
We aim to stop using barrel files (using an index file in a folder to bundle exports for the consuming code). Barrel files require manual maintenance and are therefore prone to neglect if forgotten. Also, imports are auto-generated and collapsable by the IDE thus barrel files give us no advantage.
-
When writing complex components, we like them to have their own folder with sub-folders for
logic
andcomponents
containing code which is only used internally by the component. In the case of logic it can also be a filelogic.ts
if there is not a lot of business logic. When a component and sub-components share some local types they are often put in a separatetypes.ts
file to avoid circular dependencies. -
We leverage interfaces to define component property structures rather than types. The name of the interface should follow the name of the component followed by a
Props
suffix. An example would be as below.interface RowProps { ... } const Row = ({ myProp }: RowProps) => { ... };
-
Booleans are prefixed with is/has/should etc.
-
Data schema properties and locale keys for the CMS are all snake_cased. These could be viewed as external data sources / APIs.
-
Event props follow a pattern of
onEventName
for the component props API. For handling the event we aim to use a name that describes what the function does as opposed to usinghandleEventNameSubject
. Specifically, if the function is not specifically created to handle an event or if the function doings can comfortably be described in a function name. -
We prefer to use function expressions over named function declarations. This means
const doSomething = () => {}
instead offunction doSomething(){}
. -
Short functions, especially lambda's, are okay to write on a single line.
-
Short if-statements are okay to put on a single line, especially if it only calls a single command:
if (isGoingToHappening()) doSomething();
-
We avoid using
boolean && doSomething();
inside the component's JavaScript logic, but do use it inside the component's JSX ({boolean && ( ... )}
) to conditionally render (parts of) the component. -
We avoid unnecessary short-hand variable names like
arr
for arrays, ori
for index, oracc
for areduce
accumulator. -
Completely separate Javascript logic from HTML/JSX. This also means removing maps from the JSX. Additionally, if you have nested maps extract them into components passing the required data to map to the component.
-
We prefer early returns. If statements should be on multiple lines, so no single line if statements.
- We write Styled Components using its OOTB tagged template literal functions instead of using an additional layer of the Styled System's
css()
method. This method improves readability, makes code easier to understand, and sticks to the fundamentals of CSS. This method still allows for usage of Styled System's theme definitions, yet removes a dependency on the actual package. - We avoid using magic numbers in code, be it logic, JSX, or styles. Magic numbers are often derived from the theme defined by Styled System and resolve to properties such as spacing and font-sizes, but are unclear on their own. Instead, we import the desired property and refer to the index in that properties array. An example would be
padding: 3
(undesired) vspadding: space[3]
(desired).
- We do not have a hard preference or requirement for using
git rebase
orgit merge
. Developers should follow what works best for them, but it should be noted that both methods are allowed and actively used in this project. - We do not have a hard preference or requirement for squashing a large number of git commits, but it can be useful to apply this when creating a pull request. This action should be used on an 'as-needed basis': if a pull request grows large due to a large amount of commits, it might improve reviewability when multiple commits are squashed. It should be noted that pull requests are squashed when merged, aside from pull requests to
master
. This is to keep a clear view of features and fixes that were merged as part of a release. - Continuing on the above: we should write a comprehensive commit message when squash merging a pull request. This message should be a (filtered) summary of the commits on the branch.
- It sometimes happens that features, or more so epics, need to be released as a whole because of dependencies within the different chunks of that feature or epic. In this case, it is wise to create an
epic/
branch which serves as a base for the different features to be implemented as part of the epic. Branching strategies for this branch are the same asdevelop
: features (feature/
) are branched off of and merged back to theepic/
branch. Theepic/
branch, as a whole, is merged intodevelop
once it is ready. If the implementation of an epic spans multiple releases, individual releases should also be merged into theepic/
branch in order to keep it up to date. In other words: it also helps to regularly syncdevelop
with theepic/
branch as to prevent conflicts in the end. - We use the following branch names:
epic/COR-XXXX-descriptive-name-of-epic-branch
for epicsfeature/COR-XXX-descriptive-name-of-ticket-branch
for featuresbugfix/COR-XXX-descriptive-name-of-ticket-branch
for bug fixeshotfix/COR-XXX-descriptive-name-of-ticket-branch
for hotfixestask/COR-XXX-descriptive-name-of-ticket-branch
for bigger features that are best reviewed in smaller chunks
- We use commit messages according to: https://www.conventionalcommits.org/en/v1.0.0/
feat(optional-scope): commit description example
for featuresfix(optional-scope): commit description example
for fixeschore(optional-scope): commit description example
for cleanups
For developers actively working on the platform we recommend reading the documentation here.