reference schema name when creating index

[kgeis]

fixes #2

[brylie]

Please update the README to document this feature.

[kgeis]

This is a bug fix, not a feature. I don't understand how it would fit in the README.

[brylie]

@drewbanin can we please get a review/approval for this bug fix?

[drewbanin]

hey @jtcohen6 - do you think you could take a look at this one?

[brylie]

It's also worth considering the impact native dbt Postgres index creation support will have on this package.

dbt-labs/dbt-core#804

[jtcohen6]

Right on — I just merged dbt-labs/dbt-core#3106, which adds native support for Postgres indexes in dbt-postgres. The tact we took there was, rather than namespacing indexes to specific schemas, creating a globally unique (hashed) name for each index based on the relation, properties, and creation timestamp.

If either of you think that's a mistake—that indexes should be namespaced to the same schema as the relation they're defined on—we can still change that before releasing v0.20.0.

In any case, I'm happy to approve and merge this PR as is, while also acknowledging that we should announce the deprecation of this package.

[brylie]

I think the approach you took is fine, as long as the index names will still be human-readable and relatively concise. Agreed on deprecating this package unless it may serve as a foundation for other Postgres-specific functionality.

[jtcohen6]

@brylie The approach over in dbt-labs/dbt-core#3106 prioritizes global uniqueness and conciseness over human readability. (The index name is just a MD5 hash.) Could you say a bit more about your desire for human readability? It makes sense intuitively, and I want to make sure we're taking a balanced approach to these trade-offs.

[kgeis]

It looks like you have a different approach, but I still updated my PR for correctness. It was not appropriate for me to reference a schema in the index name.