drhuffman12/ci_matrix CouchDB (2.x? and 3.x?)

[drhuffman12]

CI 'matrix' - CouchDB

Expand the CI to run the tests against a 'matrix' of version combo's made up of the:

• latest 2 versions of CouchDB (2.x? and 3.x?)
  ... possibly via CircleCI or in GitHub Workflow/Action or otherwise

Later/separate PR's!:

• [tbd] latest 2 versions of Crystal (stable? and nightly?)
• [tbd] multiple OS's (Linux/Mac/Windows)

---

• latest 2 versions of CouchDB (2.x? and 3.x?)
  In the test section, I am outputting the CouchDB versions; e.g.:

  • v2 (as of https://app.circleci.com/pipelines/github/drhuffman12/couchdb.cr/12/workflows/b4c835e3-639a-4cd2-8168-9de4924c430b/jobs/14):

      #!/bin/bash -eo pipefail
      crystal spec --junit_output=test_results
      ..info.is_v2? # => true
      .................
      
      Finished in 177.25 milliseconds
      19 examples, 0 failures, 0 errors, 0 pending
    

  • v2 (as of https://app.circleci.com/pipelines/github/drhuffman12/couchdb.cr/12/workflows/b4c835e3-639a-4cd2-8168-9de4924c430b/jobs/13):

      #!/bin/bash -eo pipefail
      crystal spec --junit_output=test_results
      ..info.is_v3? # => true
      .................
      
      Finished in 348.04 milliseconds
      19 examples, 0 failures, 0 errors, 0 pending
    

• Check into what is causing the big difference in milliseconds times and what to do about it (if needed).

  • Why/Root Cause?
    • Maybe this is due to some sort of queue delay?
    • Maybe this is due to multiple 'builds' the 'matrix'?
    • Do we need to be concerned?
    • Is there a 'simple' fix?
  • Looks like just a CircleCI queue/job issue. The next commit basically swapped the times.):
    • https://app.circleci.com/pipelines/github/drhuffman12/couchdb.cr/13/workflows/40988907-c51c-4de2-a006-4fbb9252aee8/jobs/16
    • https://app.circleci.com/pipelines/github/drhuffman12/couchdb.cr/13/workflows/40988907-c51c-4de2-a006-4fbb9252aee8/jobs/15

• Why is the CircleCI 'job' for Container couchdb:X (where X is 2 or 3) erroring out?
  (Luckily, the CI build succeeds and the tests pass.)

  • Why/Root Cause?
    • Maybe this is due to some sort of queue delay?
    • Maybe this is due to multiple 'builds' the 'matrix'?
    • Do we need to be concerned?
    • Is there a 'simple' fix?
    • Ok for now?

See also: TechMagister#14