Feature: on_mount() component lifecycle method

[hgzimmerman]

Description

**I'm submitting a Feature Request

I want to have one of my components send a request to an agent as soon as it is created and have its reply message assist in setting up the component.

What I currently do is send the message off in the create() method and hope that the component mounts and is ready to receive the response from the agent in time. In practice, it always does, but it doesn't seem correct to send off requests before the newly-created component is mounted and ready to handle replies.

I propose that an overridable on_mount() method be added to the Component trait. By default, it would not contain any code, and would be called after the component is attached to the vdom(?).

I admit that I'm coming from a point of ignorance and I accept that some queuing mechanism may already exist to avoid reply messages being dropped when their target component isn't ready. If that is the case, I'll happily close this issue and consider the sending messages to agents in create() acceptable.

[jstarry]

@hgzimmerman Yeah, this proposal sounds great! Feel free to get started and ping on gitter or here if you have any questions

[hgzimmerman]

I can open a PR whenever, I'd just like to check some assumptions first:

If I understand correctly, the component lifecycle is as follows:

• Empty
• Ready
• Created
• Destroyed
  with Processing as a standin during updates/transitions.

I've stuck the on_mount() call (embedded in mounted() for convenience) in the transition between Ready and Created, like this:

 ComponentState::Created(state.create().update().mounted())

The way a component is mounted appears to rely on CreatedState::update() calling VNode::apply() calling VComp::apply(), which in turn calls Unmounted::mount() and then sets the VComp's state to Mounted(mounted). As far as I can tell, the first call to the aforementioned update() is the first time the component will be attached to the vdom tree.

If the above assumption is correct, then I think I stuck the on_mount() call in the correct location.

[jstarry]

Yeah sounds right to me!