Editor component possible race condition

[nvartolomei]

I was wondering if there could be a race condition.

Suppose I try to edit an article, Editor component si being mounted and EDITOR_PAGE_LOADED event is dispatched.

The request is sent to server, but the server didn't respond yet. (1)

Now I go back to homepage, editor component is unmounted, EDITOR_PAGE_UNLOADED is dispatched, the state related to editor component is cleared.

Now I try to edit another article, Editor component is mounted, EDITOR_PAGE_LOADED is dispatched for another article.

The request is sent to server, and server responds instantly. (2)

Reducer for EDITOR_PAGE_LOADED action triggered by second request is called. Everything cool still.

Now, the first request (1) is full-filled, and reducer for EDITOR_PAGE_LOADED action is called, with information from first request, and in the editor I will see the contents for first article I tried to edit, instead of the second one.

Hm?

[vkarpov15]

Fair point. Will put in a fix tomorrow

[vkarpov15]

@nvartolomei what do you think about #8?

[nvartolomei]

Not sure if I like this one, page change makes all requests made through promise middleware obsolete. This is too much, imho. Let's say I have a header that stays the same on all pages, an action from header will be killed in middleware if editor is unloaded?

[vkarpov15]

You're right, hadn't quite thought of that. Have any other ideas? I'll do some research and see if I can come up with something better today.

[vkarpov15]

@nvartolomei see what you think of #9 . Slightly different approach based on explicitly cancelling all tracked promises when a route unloads.

[nvartolomei]

@vkarpov15 looks good to me.