Warns are not summarized at the end of dbt test

[MartinGuindon]

Issue

When using the new warn severity for tests, any warn results are not summarized at the end of the dbt test output, hence they could be easy to miss in larger projects.

Results

The current ouput looks like everything is fine and dandy, even if there are warnings:

15:23:56 | Finished running 282 tests in 19.27s.

Completed successfully

Done. PASS=282 ERROR=0 SKIP=0 TOTAL=282

I would instead expect an output similars to errors, summarizing how many warnings there are and which test(s) are returning warnings:

Completed successfully with 1 warnings:

Warning in test employees_overlapping_dates (tests/employees_overlapping_dates.sql)
  Got 23 results, expected 0.

  compiled SQL at target/compiled/absobi_dbt_operations/data_test/employees_overlapping_dates.sql

Done. PASS=281 ERROR=0 WARN=1 SKIP=0 TOTAL=282

[drewbanin]

Thanks @MartinGuindon - really good idea! Would love to add this for the 0.14.0 bugfix/patch release

[emilieschario]

Digging into this a bit, here is what I'm finding:

• we'll need to parse the warn errors and then putting that into here https:/fishtown-analytics/dbt/blob/399b33822a85493a8790df787eed59e951aecf6c/core/dbt/ui/printer.py#L330
• and update this function to reflect those changes https:/fishtown-analytics/dbt/blob/399b33822a85493a8790df787eed59e951aecf6c/core/dbt/ui/printer.py#L317

From looking at the run task, it should trickle through if that is the update.
https:/fishtown-analytics/dbt/blob/399b33822a85493a8790df787eed59e951aecf6c/core/dbt/task/run.py#L14

I will have to dig deeper into how the results are parsed for errors to see how/if warns needs to be different

Am I thinking about this the right way?

[beckjake]

This is the key method: https:/fishtown-analytics/dbt/blob/399b33822a85493a8790df787eed59e951aecf6c/core/dbt/ui/printer.py#L266

I think what I would do is:

• change TestRunner.after_execute() here to call a new function you make in dbt.ui.printer
• have that new function be a lot like the current print_run_end_messages, except have it call a new function instead of print_run_status_line (print_test_status_line?)
• have print_test_status_line handle the warn results accordingly

[emilieschario]

Got it! Thanks for that additional color @beckjake. Could you assign this to me to indicate it's being worked on? I'd like to take a stab at it.

[drewbanin]

Hey @emilieschario - is this in progress? We're hoping to cut an RC of 0.14.1 early next week, then publish the final 0.14.1 release later in the week. Let me know if there's anything we can help out with!

[emilieschario]

Sorry @drewbanin I won't have the bandwidth to get to this quickly. No worries if you'd rather reassign this!

In the future, I'll make sure I have time carved out before asking for it to be assigned!

[drewbanin]

No problem at all @emilieschario! Thanks for the update :)

[drewbanin]

Closed by #1654