release/v0.10.2

[fivetran-joemarkiewicz]

PR Overview

This PR will address the following Issue/Feature: Issue #32

This PR will result in the following new package version: v0.10.2

This will not be a breaking change since this will not impact the end results of the models. This will simply choose to partition by source_relation if that feature is being used or not.

Please provide the finalized CHANGELOG entry which details the relevant changes included in this PR:

Bug Fixes

• Updated the is_most_recent_record window function in the following models to exclude the source_relation field from the partition statement when pinterest_ads_union_schemas or pinterest_ads_union_databases variables are empty. Also, modified it to skip the window function if the upstream table is empty, using the new result_if_table_exists() and is_table_empty() macros. This change addresses Redshift's issue with partitioning by constant expressions.
  • stg_pinterest_ads__ad_group_history
  • stg_pinterest_ads__advertiser_history
  • stg_pinterest_ads__campaign_history
  • stg_pinterest_ads__keyword_history
  • stg_pinterest_ads__pin_promotion_history

Contributors

• @JessicaKMarkiewicz (PR #35)

PR Checklist

Basic Validation

Please acknowledge that you have successfully performed the following commands locally:

• dbt run –full-refresh && dbt test
• [n/a] dbt run (if incremental models are present) && dbt test

Before marking this PR as "ready for review" the following have been applied:

• The appropriate issue has been linked, tagged, and properly assigned
• All necessary documentation and version upgrades have been applied
• docs were regenerated (unless this PR does not include any code or yml updates)
• BuildKite integration tests are passing
• Detailed validation steps have been provided below

Detailed Validation

Please share any and all of your validation steps:

In addition to relying on the standard buildkite integration tests, I was able to confirm the compilation state based off these changes for the following scenarios. Please note, this is only validating the ad_group_history model. However, since the code changes are consistent across the other models this should be appropriate coverage.

• ✅ When running on an empty schema (the results should just be 1=1 as is_most_recent_record).
  • 
• ✅ When running on a single populated schema. We do not partition by source_relation since that field is a constant expression when not using the union data feature.
  • 
• ✅ When using the union data feature and running on two populated schemas the window function appropriately partitions by the source_relation since that is no longer a constant expression and should be included in the window function to differentiate sources.
  • 
• ✅ When using the union data feature and running on two empty schemas we do not want to use the window function at all in order to avoid a similar constant expression issue. Therefore, the result is appropriately a simple 1=1 as is_most_recent_version.
  • 

Additionally, I was able to test that these changes work as expected and do not cause any issues when using this version in downstream dependencies:

• fivetran/dbt_pinterest@main...test/upstream-pinterest-source-update
• fivetran/dbt_ad_reporting@main...test/upstream-pinterest-source-update

If you had to summarize this PR in an emoji, which would it be?

🥧

[fivetran-reneeli]

Looks good!