Tim Sweña (Swast)

Found in ibis-project/ibis#2353

The ideal flow is that we add the operator to https://github.com/ibis-project/ibis/blob/master/ibis/expr/operations.py (plus add the pivot method to the table class(es). Then we register the implementation here: https://github.com/ibis-project/ibis-bigquery/blob/abe966b6ed8567fb26bb4f4c7f711f87921b9819/ibis_bigquery/compiler.py#L481-L484 (assuming this is...

For BigQuery this makes sense, especially now that we've done some pretty extensive testing of the SQLAlchemy connector for BigQuery. My concerns: - Maybe it'll be harder to share code...

I guess another concern is UDFs. I don't believe we have support for this in the SQLAlchemy connector (though it would be cool to have!)

Another thought: I'd still want some custom code to execute the query text once compiled & transform to pandas dataframe. SQLAlchemy -> pandas is possible with just the SQLAlchemy connector,...

Some news: we now have a 1.0 (General Availability) release of the official SQLAlchemy connector for BigQuery (https://github.com/googleapis/python-bigquery-sqlalchemy) After that and my failed attempt at migration in https://github.com/ibis-project/ibis-bigquery/pull/91, I'm thinking...

Related discussion: https://github.com/ibis-project/ibis/issues/8138#issuecomment-1919702297 which details problems with the current interval type.

I'm able to reproduce this issue after updating my local test environment on Python 3.9. It doesn't appear to affect fresh environments. My requirements.txt: ``` $ pip freeze aiohttp==3.8.5 aiosignal==1.3.1...

Since we are using numpy directly, we need to add it here: https://github.com/googleapis/python-bigquery-dataframes/blob/main/setup.py and also make sure that our minimum acceptable version is pinned here: https://github.com/googleapis/python-bigquery-dataframes/blob/main/testing/constraints-3.9.txt

Per https://numpy.org/neps/nep-0029-deprecation_policy.html, we should support at least numpy 1.24.x+