[Explore] Single metric selector is inoperable if the referenced metric no longer exists

[john-bodley]

If a metric is removed from a datasource the single metric widget is inoperable if saved slice or URL references the previously defined metric. Note this isn't a problem for visualization types which support multiple metrics.

Expected results

The metric widget to render correctly.

Actual results

The metric widget in inoperable (see attached screenshot).

Screenshots

How to reproduce the bug

1. Using the example data visit the "Number of Girls" chart.
2. Edit the datasource by deleting the sum_num metric and then clicking SAVE.
3. Reload the page.
4. Verify the METRIC widget is inoperable (see screenshot).

Environment

(please complete the following information):

• superset version: master
• python version: 3.7
• node.js version: 14

Checklist

Make sure to follow these steps before submitting your issue - thank you!

• I have checked the superset logs for python stacktraces and included it here as text if there are any.
• I have reproduced the issue with at least the latest released version of superset.
• I have checked the issue tracker for the same issue and I haven't found one similar.

Additional context

Add any other context about the problem here.

[Bonifacio2]

I'll take a look at this over the weekend.

[Bonifacio2]

I wasn't able to reproduce the error. After deleting the metric the save button is disabled and I get the following warning:

[john-bodley]

@Bonifacio2 I was missing a step (now updated). After deleting the metric you'll need to reload the page and then the metric selector will be inoperable.

[Bonifacio2]

Thank you for the update. I'll take a second look at it tomorrow.

[Bonifacio2]

Still unable to reproduce. :(
I'm guessing the change is only persisted to the database when you click the save button and not right when you remove the metric. After reloading the page it looks exactly like the first time it was loaded.
Are you using superset's latest code?

[john-bodley]

@Bonifacio2 this is reproducible using the latest code (cut from master on/around the 2nd of June) and was also reported by #14151.

[john-bodley]

@Bonifacio2 I was wondering whether there was any update on this? I rebased from master today and rebuilt my environment from scratch and was able to reproduce the issue as outlined in the description.

[junlincc]

@john-bodley did you have any feature flag enabled?

[Bonifacio2]

Hello, everyone. Sorry, but I no longer have the time to take care of this issue. I guess this it is up for grabs again.

[john-bodley]

@junlincc no—I was using the out-of-the-box configuration.

[rusackas]

@kgabryje said he'd be willing to do a little investigation here, so I'm (at least temporarily) sticking his name on it :)