Exploring Arrow, SQlite

Installation

In addition to the packages in the Pipfile, you need to locally install sqlite and sqldiff.

Goals

• can we represent the Spine datastructure in Arrow?
  • e.g. expressing "relationships"
  • build against the entity-only branch; see: @spine-tools/Spine-Database-API#215
• look at the cache implementation

ADBC Tasks

1. figure out how to set the PostgreSQL URI env variable correctly ✅

   $ docker run --rm -p 5432:5432 --name postgres-adbc-test -e POSTGRES_PASSWORD=mysecretpassword -d postgres
   $ export ADBC_POSTGRESQL_TEST_URI="postgresql://postgres:mysecretpassword@localhost:5432/postgres"
   

2. figure out how setup & teardown for tests work
3. a. replicate table setup for FOREIGN KEY using their test framework (steal from PRIMARY KEY) linking to a PRIMARY KEY
4. b. replicate table setup for FOREIGN KEY using their test framework (steal from PRIMARY KEY) not linking to a PRIMARY KEY
5. research about how to test FOREIGN KEY elements
6. create test cases

Build hints

Setting export CMAKE_EXPORT_COMPILE_COMMANDS=ON creates a compile_commands.json in the build directory. clangd can use this to correctly recognise names and integrate with LSP. When using the Ninja generator, you have to explicitly call ninja -t compdb > compile_commands.json, for Makefiles it is generated by just calling make.