[Regression] GENERIC_CHART_AXES: numeric x-axis treated as categorical values

[rumbin]

When using a numeric column as the generic x-axis, the x-axis values aren't linearly distributed along the x-axis but they are evenly distributed, just as if they were categorical values

How to reproduce the bug

0. Have the GENERIC_CHART_AXES feature flag enabled
1. Create a dataset or a SQLLab query with numerical columns, e.g.

select
  *
from (values
  (1,1)
  , (2,2)
  , (5,5)
  , (10,10)
) as v(x,y)

2. Visualize the query/dataset as bar chart (or line chart)
3. Use x as x-axis and, e.g., max(y) as y-axis
4. See that the x-axis values are shown with equal spacing instead of the expected linear distribution of the x values

Expected results

This is what it should look like and what it did correctly in Superset 2.0.1:

Actual results

X-axis values are seemingly treated as categorical values:

Any distribution/scatter/histogram plots based on this functionality are actually unusable due to this bug.

Environment

(please complete the following information):

• browser type and version:
• superset version: 3.0.2rc1
• python version: python --version
• node.js version: node -v
• any feature flags active:

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.

[michael-s-molina]

@villebro @zhaoyongjie Could you take a look at this one?

[villebro]

I will take a look, thanks for reporting @rumbin

[WojtekWaga]

Hey @villebro I checked the backend response and it's identical in this respect. I fed the json from Superset 2.0.1 backend into Superset 3.0.2rc1 frontend and it also misinterprets XAxis type. This made me convinced it's a FE issue. It looks like the getAxisTime defaults to AxisType.category in this case whereas it should return AxisType.time.

[villebro]

I was able to repro, investigating now

[rumbin]

@villebro, we found a fix on our side. @WojtekWaga will report here tomorrow and/or provide a PR.

[villebro]

@rumbin @WojtekWaga oops, I already opened a fix 🙁 Do you mind reviewing the PR #26087 to see if it's in line with the changes you were proposing?

[michael-s-molina]

Thank you all!