alighasami/xyflow
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

chore(hooks): tsdoc update

Browse Source
This commit is contained in:
moklick
2025-02-11 13:59:17 +01:00
parent ebec872318
commit d8ab3bf0bd
50 changed files with 895 additions and 271 deletions
Download Patch File Download Diff File Expand all files Collapse all files
packages/react/src/container/NodeRenderer/index.tsx
+25 -23
View File
@@ -44,29 +44,31 @@ function NodeRendererComponent<NodeType extends Node>(props: NodeRendererProps<N
<div className="react-flow__nodes" style={containerStyle}>
{nodeIds.map((nodeId) => {
return (
// The split of responsibilities between NodeRenderer and
// NodeComponentWrapper may appear weird. However, it’s designed to
// minimize the cost of updates when individual nodes change.
//
// For example, when you’re dragging a single node, that node gets
// updated multiple times per second. If `NodeRenderer` were to update
// every time, it would have to re-run the `nodes.map()` loop every
// time. This gets pricey with hundreds of nodes, especially if every
// loop cycle does more than just rendering a JSX element!
//
// As a result of this choice, we took the following implementation
// decisions:
// - NodeRenderer subscribes *only* to node IDs – and therefore
// rerender *only* when visible nodes are added or removed.
// - NodeRenderer performs all operations the result of which can be
// shared between nodes (such as creating the `ResizeObserver`
// instance, or subscribing to `selector`). This means extra prop
// drilling into `NodeComponentWrapper`, but it means we need to run
// these operations only once – instead of once per node.
// - Any operations that you’d normally write inside `nodes.map` are
// moved into `NodeComponentWrapper`. This ensures they are
// memorized – so if `NodeRenderer` *has* to rerender, it only
// needs to regenerate the list of nodes, nothing else.
/*
* The split of responsibilities between NodeRenderer and
* NodeComponentWrapper may appear weird. However, it’s designed to
* minimize the cost of updates when individual nodes change.
*
* For example, when you’re dragging a single node, that node gets
* updated multiple times per second. If `NodeRenderer` were to update
* every time, it would have to re-run the `nodes.map()` loop every
* time. This gets pricey with hundreds of nodes, especially if every
* loop cycle does more than just rendering a JSX element!
*
* As a result of this choice, we took the following implementation
* decisions:
* - NodeRenderer subscribes *only* to node IDs – and therefore
* rerender *only* when visible nodes are added or removed.
* - NodeRenderer performs all operations the result of which can be
* shared between nodes (such as creating the `ResizeObserver`
* instance, or subscribing to `selector`). This means extra prop
* drilling into `NodeComponentWrapper`, but it means we need to run
* these operations only once – instead of once per node.
* - Any operations that you’d normally write inside `nodes.map` are
* moved into `NodeComponentWrapper`. This ensures they are
* memorized – so if `NodeRenderer` *has* to rerender, it only
* needs to regenerate the list of nodes, nothing else.
*/
<NodeWrapper<NodeType>
key={nodeId}
id={nodeId}