Get proper UX for runtime errors

[rvsia]

When runtime error happens, there is no proper UX.

Current behavior

cc @terezanovotna

[nlcwong]

@rvsia Can you explain the steps that was taken to get this error? Is it reproducible?

[rvsia]

@nlcwong I had to make a mistake in the code to get this error. It is something what users should never see, when everything works correctly. But never say never...

[KendraMar]

@rvsia Sorry for the delay in getting back to you!
@nlcwong
@MariSvirik

• Can you tell us more about this error? When might it happen/what causes it?
• is this something the customer can take action on? if so, what do you want the customer to do? Should the customer just refresh and the error will likely clear and they'll see content? or do they need to file a support ticket?

FYI, Maria Svirikova and Natalie Wong are gathering as many error conditions that interrupt user activity as they can. Once they understand the nature of these errors, they'll develop patterns to show them in the UI.

[rvsia]

@KendraMar

1. it is caused by errors in the code or by some data inconsistency. (i.e.: the API returns data in wrong format which cannot be handled by the UI) What we know about these errors: simple description (it tells why it failed, however, this info is mostly useful only for developers, as you can see in the screenshot), stack (place where it happens in the code), SentryID (with help of this we can track errors) edit: it is possible to collect all types of errors: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Error 😃 however, not sure how much it is useful for users and for the design of the app.

2. the whole app (Sources, not the menu, header, etc.) failed, so user can only refresh it. However, we can show him a button or refresh/redirect it automatically. (I think we need to show only the SentryID, as users can send it to the customer support.) And yes, the error should dissapear after the refresh, if not, then there is some huge problem in the code itself. However, I am pretty confident that this is not going to happen, because we have strong testing.

In ManageIQ, there is a simple “Something went wrong” page with some description what can be wrong (error on server etc.)