Reputation: 482
I have a mathematical algorithm that I want to keep separated from React. React would be a view to the state within that algorithm, and should not define the way of how the logic is flowing within the algorithm. Also, since it is separated, it's much easier to unit test the algorithm. I have implemented it using class components (simplified):
class ShortestPathRenderer extends React.Component {
ShortestPath shortestPath;
public constructor(props) {
this.shortestPath = new ShortestPath(props.spAlgorithm);
this.state = { version: this.shortestPath.getVersion() };
}
render() {
... // Render waypoints from shortestPath
}
onComponentDidUpdate(prevProps) {
if (prevProps.spAlgorithm !== this.props.spAlgorithm) {
this.shortestPath.updateAlgorithm(this.props.spAlgorithm);
}
}
onComponentWillUnmount() {
this.shortestPath = undefined;
}
onAddWayPoint(x) {
this.shortestPath.addWayPoint(x);
// Check if we need to rerender
this.setState({ version: this.shortestPath.getVersion() });
}
}
How would I go about this using React hooks? I was thinking about using the useReducer method. However, the shortestPath variable would then be a free variable outside the reducer and the reducer is no longer pure, which I find dirty. So in this case the whole state of the algorithm must be copied with every update on the internal state of the algorithm and a new instance must be returned, which is not efficient (and forces the logic of the algorithm to be the React-way):
class ShortestPath {
...
addWayPoint(x) {
// Do something
return ShortestPath.clone(this);
}
}
const shortestPathReducer = (state, action) => {
switch (action.type) {
case ADD_WAYPOINT:
return action.state.shortestPath.addWayPoint(action.x);
}
}
const shortestPathRenderer = (props) => {
const [shortestPath, dispatch] = useReducer(shortestPathReducer, new ShortestPath(props.spAlgorithm));
return ...
}
Upvotes: 1
Views: 1529
Reputation: 11600
I'd go with something like this:
const ShortestPathRenderer = (props) => {
const shortestPath = useMemo(() => new ShortestPath(props.spAlgorithm), []);
const [version, setVersion] = useState(shortestPath.getVersion());
useEffect(() => {
shortestPath.updateAlgorithm(spAlgorithm);
}, [spAlgorithm]);
const onAddWayPoint = (x) => {
shortestPath.addWayPoint(x);
// Check if we need to rerender
setVersion(shortestPath.getVersion());
}
return (
... // Render waypoints from shortestPath
)
}
you can even decouple logic further and create useShortestPath
hook:
reusable stateful logic:
const useShortestPath = (spAlgorithm) => {
const shortestPath = useMemo(() => new ShortestPath(spAlgorithm), []);
const [version, setVersion] = useState(shortestPath.getVersion());
useEffect(() => {
shortestPath.updateAlgorithm(spAlgorithm);
}, [spAlgorithm]);
const onAddWayPoint = (x) => {
shortestPath.addWayPoint(x);
// Check if we need to rerender
setVersion(shortestPath.getVersion());
}
return [onAddWayPoint, version]
}
presentational part:
const ShortestPathRenderer = ({spAlgorithm }) => {
const [onAddWayPoint, version] = useShortestPath(spAlgorithm);
return (
... // Render waypoints from shortestPath
)
}
Upvotes: 1
Reputation: 311
You can switch class-based in example in functional analog just using useState hook
function ShortestPathRenderer({ spAlgorithm }) {
const [shortestPath] = useRef(new ShortestPath(spAlgorithm)); // use ref to store ShortestPath instance
const [version, setVersion] = useState(shortestPath.current.getVersion()); // state
const onAddWayPoint = x => {
shortestPath.current.addWayPoint(x);
setVersion(shortestPath.current.getVersion());
}
useEffect(() => {
shortestPath.current.updateAlgorithm(spAlgorithm);
}, [spAlgorithm]);
// ...
}
Upvotes: 1