Fix FRBC example crash

[wcoenraads]

Now that S2Message is defined in message.py, the FRBC example in the repository no longer works. We need to fix this (which should be straightforward), and ideally add some kind of check in the pipelines that the examples still work.

I think we may have broken semver with the move of S2Message, as that's a breaking change and we did a patch version update (0.3.0 to 0.3.1).

[Flix6x]

Thanks for reporting this. I recommend these follow-up actions before closing this issue:

• To fix the issue:
  • 1) Update the examples with a PR to the main branch: either simply renaming S2Message to S2MessageComponent as S2Message in the imports, or (preferably) changing to from s2python.message import S2Message should do the trick (probably related issue: S2MessageComponent to S2Message #80).
  • 2) Open an issue to discuss covering the examples in the test suite. Add a test that runs example Resource Manager. #90
• To fix the broken semantic versioning [UPDATE: wasn't broken to begin with]:
  • 3) Open up a branch called 0.3.x and on that branch revert add S2 Message Union type #37.
  • 4) Release v0.3.2 from the 0.3.x branch (might need an edit to .github/workflows/ci.yml, because it only looks for tags on main right now).
  • 5) Yank v0.3.1 (may require PyPI access).

I suggest Seita takes up 1, 3 and 4. Could you take up 2 and (after 3 and 4 are done) 5?

[wcoenraads]

Sounds good! I'll look into PyPI access, I don't have it myself - but I assume at least one of my coworkers has access.

[jorritn]

Status update:

• The fix has been applied to frbc_example in master.
• I see that version 0.4.0 and 0.4.1 have been released with the S2Message change applied (but still with a not working example).
• I opened Add a test that runs example Resource Manager. #90 to address the test case

@Flix6x could you revert #37 on a new branch and release 0.3.2? Then we'll will yank 0.3.1 on PyPi.

[Flix6x]

I branched off the v0.3.1 release and reverted #37 only to discover that the renaming had not actually happened yet on v0.3.1. Instead, I found that it happened on v0.4.0. Therefore, there is no need to revert #37, release v0.3.2 and yank v0.3.1.

It looks like semantic versioning wasn't broken.