Add default mounted() method to Component trait

[hgzimmerman]

Fixes #572

I made the choice to have on_mount() not return a ShouldRender bool, because I don't think there is a situation where altering parts of a component state relevant to the view() should ever take place in on_mount() instead of create(). Let me know your thoughts, because its an easy enough change to make.

[hgzimmerman]

Actually, the thought just occurred to me that on_mount() could be used to grab a reference to an element introduced in the initial call to view(). One application of this could be to control Canvas elements, which you would likely want to re-render as soon as you initialize the canvas using stdweb.

Let me know if you agree that on_mount() should optionally cause a re-draw.

[jstarry]

@hgzimmerman yeah let's follow the lead of React and allow re-rendering on mount: https://reactjs.org/docs/react-component.html#componentdidmount

Their stated use-case of dialogs getting the width of the screen for example is something we should support. Also, like React, we will not update the browser screen until the mount is finished.

[hgzimmerman]

I think this accomplishes what you want.

The only point of ambiguity that I have is about your comment here:

Also, like React, we will not update the browser screen until the mount is finished.

Does calling update() exclusively handle vdom stuff and the steps for rendering to the screen are handled by another part of Yew's internals? Because update() can be called twice with what I have here, and I would like to avoid drawing to the screen twice if that is part of the functionality present within update().

[jstarry]

Yup, this is exactly what I had in mind. My understanding is that since the environment is single threaded, the browser won't update the screen until your code finishes.

[jstarry]

Great, thanks!!

[jstarry]

bors r+

[bors]

Build succeeded

• continuous-integration/travis-ci/push