V0.19.0

[dataders]

should wait until 0.19.0 is actually released, but the CI tests will still be helpful

[dataders]

Weird error on the Invalid column name 'id'. for the test_dbt_snapshot_strategy_timestamp test.

2021-01-11 17:02:59.100974 (MainThread): Completed with 2 errors and 0 warnings:
2021-01-11 17:02:59.101065 (MainThread): 
2021-01-11 17:02:59.101149 (MainThread): Database Error in test failing (test/failing.sql)
2021-01-11 17:02:59.101219 (MainThread):   ('42S22', "[42S22] [Microsoft][ODBC Driver 17 for SQL Server][SQL Server]Invalid column name 'id'. (207) (SQLExecDirectW)")
2021-01-11 17:02:59.101282 (MainThread):   compiled SQL at target/compiled/dbt_test_project/test/failing.sql
2021-01-11 17:02:59.101349 (MainThread): 
2021-01-11 17:02:59.101423 (MainThread): Database Error in test passing (test/passing.sql)
2021-01-11 17:02:59.101518 (MainThread):   ('42S22', "[42S22] [Microsoft][ODBC Driver 17 for SQL Server][SQL Server]Invalid column name 'id'. (207) (SQLExecDirectW)")
2021-01-11 17:02:59.101581 (MainThread):   compiled SQL at target/compiled/dbt_test_project/test/passing.sql

[mikaelene]

There are ci tests on the branch? https://app.circleci.com/pipelines/github/dbt-msft/dbt-sqlserver?branch=v0.19.0&utm_campaign=workflow-failed&utm_medium=email&utm_source=notification

[mikaelene]

I don't get that error when testing local?

[dataders]

@mikaelene maybe it's because I'm installing this way which should take the latest commit from dev/kiyoshi-kuromiya because it is the default branch. Maybe this is because you were installing pre-release cut on PyPI? feel free to change this to whatever you like.

[dataders]

@mikaelene I made this PR so now every change that is pushed to the v0.19.0 branch will now get tested. Should be helpful to have in addition to your local testing, but I might have goofed something w/ CircleCi...

[mikaelene]

I noticed some weird behaviour that the latest version, even installed from git does not include the latest commit. Can't get my mind around why it is happening. I'll try to hardcode the latest commit.

[mikaelene]

I don't get why it uses the old version. Incorrect syntax near '!'. is what is supposed to be fixed. Maybe we have to wait?!?... Thanks for trying @swanderz !

[dataders]

@jtcohen6 sorry to pull in into yet another dbt-msft PR, but any idea why CircleCI isn't pulling the latest commit from dev/kiyoshi-kuromiya?

[mikaelene]

I think the package are cached by circleCI. Since it is the same version on the new changes it will still take the cached version of the package.

[dataders]

I think the package are cached by circleCI. Since it is the same version on the new changes it will still take the cached version of the package.

I think you're onto something. However, I'm trying to read about clearing the cache and my head is spinning...

[mikaelene]

I think I have solved it. Pushing in a minute

[mikaelene]

AAAHHHHHH! I am getting crazy! :-)

[jtcohen6]

@swanderz @mikaelene I'm not enough of a Circle pro for this one. Does it work when you run these tests locally?

In all likelihood, we'll be coming out with a second RC this week, given the (tiny) changes we've merged since the first one.

[mikaelene]

It works locally and it fails in CircleCI on the thing we fixed in the latest commits. But I am fine waiting a couple of days.

[dataders]

@swanderz @mikaelene I'm not enough of a Circle pro for this one. Does it work when you run these tests locally?

In all likelihood, we'll be coming out with a second RC this week, given the (tiny) changes we've merged since the first one.

yeah i found an option for circleci python orb's install-packages orb, pypi-cache: false but no luck...

I'm still going to make a release cut of this so I can keep testing w/ Synapse

[dataders]

I'm still going to make a release cut of this so I can keep testing w/ Synapse

update: somehow test_dbt_snapshot_strategy_timestamp is passing over in microsoft/dbt-synapse#39

_________________________ usecase: test_dbt_data_test __________________________
usecase execution failed
   spec failed: ("Invalid result, missing required key 'unique_id'",)
   no further details known at this point.

[mikaelene]

It is now on PyPi

[mikaelene]

I want to merge this to master but can't male the tests pass locally. Actually I don't see how they can work in CircleCI either. Or why they worked before. From my understanding the data test executes
with dbt__CTE__INTERNAL_test as ( select 1 as id where id = 1 )select count(*) from dbt__CTE__INTERNAL_test

and select 1 as id where id = 1 can't be executed on SQL Server?!

I get Invalid column name 'id' which makes sense as I get that when executing the SQL in my Editor on SQL Server. It works on other DBs.

Anyone have a clue why this works in the CircleCI env?

[dataders]

Anyone have a clue why this works in the CircleCI env?

I have no idea! and I've been banging my head on this for a week now.... I'm starting to learn now via Stack Overflow...

[mikaelene]

It actually passes when run with tox. I don't get this tox thing but I'll do the merge.