Components

[estrattonbailey]

So mount() and unmount() are much trickier than they would appear at first glance. I think that's the main barrier to having "components" as you would think about them in something like React.

As filed here, on-load correctly fires onunload when replaceChild(next, prev) is fired, because for a short period of time (I assume one tick, but a setTimeout(() => {}, 0) does not work) the DOM node is actually not present in the DOM. on-load is doing its job correctly. Since there is no reference to the node for a split second during that time, I'm having trouble finding a good way to tell if the node removal was in the act of morphing the DOM, or actually a valid unmounting, like from a route change.

Other issues

A good outline (from Choo) is here. It also references a few issues that I'll also need to ensure are addressed and tested and added to docs:

• onload performance shama/on-load#13
• Morphing between two instances of a component leaks proxy nodes choojs/nanocomponent#29
• WIP: Use node id as a hint of which node to compare to choojs/nanomorph#61

Additionally:

• https://github.com/jongacnik/byo#why-does-this-exist - nested components, "we can't quite yet pull off with nanocomponent"

Other ideas

• this is dope https://github.com/aaaristo/rbel
  • related: choojs/nanohtml@master...timwis:custom-elements
• whole bunch here Discussion: choo/component choojs/choo#593 (comment)

Questions

1. On global state updates, they're currently morphing on a local level, and returning the morphed ref. How does this affect the top-level morph operation? Does it essentially morph twice (I don't think so) or does it result in the same node? (this should be what happens)
2. How can I better utilize .isSameNode() or expose it for users?

[estrattonbailey]

No longer needed after #5 🤓