Implement printSchemaAndCapabilities, fix scalar type dependencies in schema response

[BenoitRanque]

This PR implements the printSchemaAndCapabilities cli command.

For this, we move the schema_response code into the common crate. We also move the configuration reading code into common

This PR also includes a bugfix for schema response generation:

We try to only return scalar types actually used by the tables in the configuration.

Scalar types may have aggregate functions returning other scalar types, for example sum(Int32) returns Int64, creating scalar dependencies.

These dependencies are now recursively accounted for.

[daniel-chambers]

Some general feedback on this PR, intended to help you make PRs that are more easily reviewable in the future 🙂

• The PR title is not descriptive and is basically useless. What is this PR doing? Please at least rename this PR before merging.

• This PR does not just do one thing. Instead it rolls up multiple different changes (some of which are mentioned in the description, some of which aren't):

  • printSchemaAndCapabilities
  • a fix to scalar types in schema
  • switching from manual Display instances for errors to using thiserror
  • create a release by bumping the version number

  You should try to make one change per PR. Not only does it help you keep track of changes when you go back into your history, but it also helps the reviewer by cleanly separating changes out into separate reviewable pieces. It also makes reverting changes easier in the unlikely case that is required.

I'm happy to approve this, but please do make the requested minor changes before merging.

[daniel-chambers]

What's this file? If this is just some file you used when testing locally, please delete it before merging.

[BenoitRanque]

Good catch. Removing

[daniel-chambers]

Can this be more descriptive so that end users understand what this means better?

[BenoitRanque]

Improved

[BenoitRanque]

Some general feedback on this PR, intended to help you make PRs that are more easily reviewable in the future 🙂

• The PR title is not descriptive and is basically useless. What is this PR doing? Please at least rename this PR before merging.

• This PR does not just do one thing. Instead it rolls up multiple different changes (some of which are mentioned in the description, some of which aren't):

  • printSchemaAndCapabilities
  • a fix to scalar types in schema
  • switching from manual Display instances for errors to using thiserror
  • create a release by bumping the version number

  You should try to make one change per PR. Not only does it help you keep track of changes when you go back into your history, but it also helps the reviewer by cleanly separating changes out into separate reviewable pieces. It also makes reverting changes easier in the unlikely case that is required.

I'm happy to approve this, but please do make the requested minor changes before merging.

RE: PR Title

That's 100% my bad. IDE picked up the last commit and used that for the name

Deleting that testing file, did not intend to commit it

This PR is a bit rushed because I needed to get that schema fix out ASAP, but duly noted on all points

Appreciate the feedback!

[daniel-chambers]

One thing I just noticed is that your changelog doesn't have dates on the versions, which means it technically isn't following "keep a changelog". Versions should look like this:

## [1.1.1] - 2023-03-05

Also useful given your changelog update says the "Note the DDN CLI does not yet implement this"... knowing when this was written would be useful to readers.