Fix invalid assumptions about Project ID's in BQ table

[KJTsanaktsidis]

The bigquery table resource tries to infer the project and dataset from the table's full name. However, it does so incorrectly as it assumes that the project ID cannot have : or . characters in it, yet both are valid.

This is leading to errors like this for me when I run terraform plan:

* module.bigquery.google_bigquery_table.connect_product_analytics_table_user_search_filters: 1 error(s) occurred:

* module.bigquery.google_bigquery_table.connect_product_analytics_table_user_search_filters: google_bigquery_table.connect_product_analytics_table_user_search_filters: Invalid BigQuery table specifier. Expecting {project}:{dataset-id}.{table-id}, got outbound.io:outbound:connect_product_analytics_experiment.user_search_filters

As you can see, our project ID is outbound.io:outbound, so the string-splitting algorithm here doesn't work.

When using this branch, the plan works correctly.

Also - I'm targeting the PR against master, but we're actually still using the 1.20 version of the provider. Is there any way this fix can get backported to the 1.x series as well?

[KJTsanaktsidis]

Any chance of getting some 👀 on this soon? The fact that this resource is in our statefile breaks planning for our entire team (unless they're running a fork of the provider)

[rileykarson]

Thanks @KJTsanaktsidis! Sorry about the delay. I'll duplicate this up to our code generator, Magic Modules, so it hits google-beta as well and then merge this PR.

I don't believe we're able to backport the change unfortunately. Out of curiosity, what's blocking you from advancing to a 2.X series release?

[KJTsanaktsidis]

Hey, thanks for this. The main thing blocking us upgrading to a 2.x release is we are using the bigtable resources in a deprecated way that’ll need to be fixed (drawing a proper distinction between instances and clusters). I need to spend some time to have a good look at how yo upgrade it without accidentally deleting our cluster :)

[rileykarson]

Fair enough! It's also always an option to use multiple provider instances at different versions, although I personally wouldn't recommend that with more than 1-2 resources ahead or behind the rest.

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. If you feel I made an error 🤖 🙉 , please reach out to my human friends 👉 hashibot-feedback@hashicorp.com. Thanks!