Can't make a timeserie if date is TIMESTAMP WITHOUT TIME ZONE (postgres)

Make sure these boxes are checked before submitting your issue - thank you!

• I have checked the superset logs for python stacktraces and included it here as text if 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

Superset version

0.28 RC6 from 2018-09-12

Expected results

A timeseris graph is displayed

Actual results

• Unexepeted Error is shown in the UI
• Inspecting the asnwer from the corresponding https request, it shows a stacktrace finishing with : File "/home/work/incubator-superset/superset/utils.py", line 366, in datetime_to_epoch return (dttm - epoch_with_tz).total_seconds() * 1000 File "pandas/_libs/tslibs/timestamps.pyx", line 320, in pandas._libs.tslibs.timestamps._Timestamp.__sub__ TypeError: Timestamp subtraction must have the same timezones or no timezones

Steps to reproduce

• Create source from a table having a date column of type TIMESTAMP WITHOUT TIME ZONE (postgres 9.3)
• Try creating a timeseries graph using it

Workaround

• According to my colleagues (I could not test), the same database allowed making TS graphs with a version of superset he installed a month ago from an “image for easyer installation he found on github”
• Adding a record having value my_date WITH TIME ZONE 'MST', typed TIMESTAMP and using it as the time for the graph worked

[apimastery]

Here is how I fixed this when running a clone of the repo’s master branch from 2018-11-01 and the Docker install (after few “adjustments” to make it work). I hope this will help others until it is permanently fixed…

For the “World’s Bank Data” dashboard:

• Select Sources --> Tables from the top-level menu.
• Click on the “Edit” icon for the wb_health_population table.
• Click on the “List Columns” tab.
• Scroll down to the “year” column.
• Click on the “Edit” icon for the “year” column.
• In the “Expression” box, enter “year ::timestamptz” (without the double quotes).
• Click on the “Save” button.
• Access/Refresh the dashboard.

I had another issue with the “World’s Bank Data” dashboard - some charts were erroring out with ‘column “xxxx” does not exist’. For example, ‘column “sp_pop_totl” does not exist’. To fix:

• Select Sources --> Tables from the top-level menu.
• Click on the “Edit” icon for the wb_health_population table.
• Click on the List Metrics tab.
• Find the metric for the offending column. For example, sum__SP_POP_TOTL and click on its “Edit” icon.
• Surround the column name in the “SQL Expression” field with double quotes. For example, change it from SUM(SP_POP_TOTL) to SUM(“SP_POP_TOTL”)

HTH

[kingo55]

@davidgreenshtein - Sometimes you can work around it by creating a new calculated column and coercing to ::timestamptz

It doesn’t work if you just update the database expression though.