Fix DB tests

[eighttrigrams]

Problem

When we enabled the GitHub actions for Forkana, many tests failed, most likely due to changes already made on our side after the fork. Specifically the database test. We opted to comment those out for the time being and see what we can recover.

Solution

• Find which tests work on the basis of the PSQL tests.
• Skip the failing ones.
• Enable the PSQL tests in CI
• Enable sqllite tests
• Enable or discard the MYSQL tests in CI - Decision has been made to discard MYSQL
• Enable or dicard the MSSQL tests in CI - Decision has been made to discard MSSQL

Old notes from after the initial investigation

Currently those block merging the linting changes (of #56) which have value on their own. We are confronted with 1) linting issues 2) automated security checks issues 3) testing issues at the same time because we just turned them on. Separating these issues helps us get on track in an orderly fashion.

See also

• (Re-)enable security checks #57
• Get Linters working on PRs #49

[eighttrigrams]

https://github.com/okTurtles/forkana/pull/56/files#r2528803095

See instances of if: false # needs.files-changed.outputs.backend == 'true' || needs.files-changed.outputs.actions == 'true' in pull-db-tests.yml.

[eighttrigrams]

This here #68 in partial fulfillment.

[eighttrigrams]

This here #73 in partial fulfillment (mysql, msql).

[taoeffect]

Alright, before we get to the PR, let's figure out first if this is something we want to do at all.

When we enabled the GitHub actions for Forkana, many tests failed, most likely due to changes already made on our side after the fork. Specifically the database test. We opted to comment those out for the time being and see what we can recover.

Yes, that would be a problem, but in your Solutions section I see this:

• Enable the PSQL tests in CI

So it sounds like we've already enabled one set of DB tests, and in fact the tests for the database that we'll most likely be using (PostgreSQL).

It's my understanding that Github gives us a fixed amount of "free credits" to run the tests each month. I'm not sure if this is per-project or across the whole org. If it's across the whole org, we definitely want to be conservative.

Tests also take a while to complete:

If they run in parallel, I suppose that's not that big of a deal (if we're relying solely on Github Actions), but it does make testing locally more difficult to enable more tests too.

So I'm hesitant to enable tests that we don't necessarily need. I see downsides but not much of an upside to enabling tests for databases we won't be using. Is this really something we need to do?

[eighttrigrams]

Follow up issue created for discontinuing support for MYSQL and MSSQL: #74