Is anything of this section CrateDB specific? Looks like it's mostly explaining SQLAlchemy, in which case we could refer to the upstream docs instead and test the functionality in python tests

Hi. While the SQLAlchemy API methods per se are not specific to CrateDB, I found this section worth to keep, in order to display different kinds of inspection and reflection methods of SQLAlchemy together with their slightly CrateDB-specific outcomes, also because they are not mentioned elsewhere in our documentation (modulo has_table), and also because the corresponding spots are a bit obfuscated within the SQLAlchemy documentation.

In other words, while bringing those snippets together from different parts of our documentation, I have made an extra effort to discriminate between the different kinds of introspection/reflection features offered by SQLAlchemy.

• SQLAlchemy inspector
• Schema-supported reflection
• CrateDialect-specific reflection methods

Some examples from the upstream documentation at runtime inspection API, SQLAlchemy inspector, and Schema-supported reflection, have been paired with CrateDB-specific responses.

The result is a reasonably concise overview about them

>>> inspector.get_schema_names()
['blob', 'doc', 'information_schema', 'pg_catalog', 'sys']

>>> inspector.default_schema_name
'doc'

>>> dialect.has_schema(connection, 'doc')
True

Other than this, some parts of the documentation section within SQLAlchemy: Create, retrieve, update, and delete are also just vanilla SQLAlchemy operations, but I think the trio of getting-started.rst, crud.rst, and inspection-reflection.rst still make sense in the context of the CrateDB driver, both as documentation, as well as a doctest scenario. Now that they have been refactored, I think they are in a better shape than before.

This sounds similar to the crud document. Would it make sense to merge the two into one? Or instead clarify the difference and link from one to the other.

Thanks for your attention. You are right. I've tried to improve this now with 5dc0e04.

The previous version was a bit of a potpourri of different topics. From that mix, two new documents have been created. One is about "advanced querying", including exercises about aggregations and full-text capabilities, and the other one exercises operations on the CrateDB container and geospatial special types concisely, now also consistently using the Object and ObjectArray type aliases, instead of the funny-sounding Craty type.

I will be happy about any kinds of suggestions on improving the wording/tone, like your previous excellent patches.

• SQLAlchemy: Advanced querying
• SQLAlchemy: Working with special CrateDB types

Is this still the case?

Good question. I also wanted to highlight this note while reading it, but missed to do so. Which suggestion do you have?
a) Investigate the matter.
b) Just remove this sentence.

a) + create an issue in the crate repo to support it if there are variants which it can't handle

I will do. However, to move on with this, and other subsequent patches, I've deferred it to GH-489. I hope this is fine.

Maybe move this elsewhere - the previous section is "Introduction to fulltext indexes" and the next section is "Fulltext search with MATCH predicate" - having an intermediate aggregation/count/group-by breaks the flow

I agree and will try to find a different spot.

The section has been moved one slot to the bottom with fe90300.