Merge train

[rudymatela]

Closes #77.

This is built on top of #134, ..., #155.

• handle new commit on PR on the train
  • add automated test
• remove getTrain?
• handle closing of PRs with pending builds (restart remaining train)
  • add automated test
• handle overriding remaining train failures with NotIntegrated
  • add automated test
• handle tagging throughout the train with two tags
  • add automated test
• handle rebase failures
  • add automated test (base build success)
  • add automated test (base build failure)
• consider reverting b697ef1 nah!
• thoroughly test all possible scenarios with 2 PRs (all: one can think of)
• test some relevant scenarios with 3 PRs.
• sweep TODOs and refactor ugly code
• support showing BuildSucceeded in the web interface somehow
• test in a git-sandbox repository, paste some screenshots here.
• one more full self code review
• send this PR for code review
• merge and tag

Test Case 1, success, success, success

Web before approval

Web right after approval

Web final

PR 120

PR 121

PR 122

[rudymatela]

@OpsBotPrime merge on friday (just a test, should be a no-op)

[rudymatela]

For reference, this is the old hash before rebasing: ed07d6d

I should probably squash everything at some point to restart.

[rudymatela]

For reference, this is the old hash before rebasing: ed07d6d

I should probably squash everything at some point to restart.

Everything is squashed in 23bbe70 now.

[rudymatela]

Note-to-self:

On Tuesday, carry on from 8e440b0 by running nix run -c stack test --ta '--match "(FOCUS)"' locally to fix the "behind" comment gathering and rendering.

[rudymatela]

There are conflicts again after the latest change.

I will squash everything then rebase on top of the new master.

Since 5139125 is a working version, I temporarily created branch junk/merge-train/integration/1 to serve as a reference.

[rudymatela]

Everything was squashed in 4c7c4ac.

I will rebase this single commit at some point.

[rudymatela]

Everything was squashed in 4c7c4ac.

I will rebase this single commit at some point.

Rebased as 13df8f0. There weren't as many conflicts due to the squash. :-)

[rudymatela]

Note-to-self: for adding automated test here:

• handle overriding remaining train failures with NotIntegrated
  • add automated test

It seems the branch is created on the origin, but not the local somehow does not detect it:

$ git -C /run/user/1000/testsuite-74fda068-fdd1-4bba-97b1-ac6062cd7af2/repo-origin branch -a
  ahead
  alternative
  fixup
  intro
  master
* unused

$ git -C /run/user/1000/testsuite-74fda068-fdd1-4bba-97b1-ac6062cd7af2/repo-origin branch -a
  ahead
  alternative
  fixup
  integration/6
  intro
  master
* unused

$ git -C /run/user/1000/testsuite-74fda068-fdd1-4bba-97b1-ac6062cd7af2/repo-local branch -a
* (HEAD detached from 3b2ca25)
  master
  remotes/origin/master

🤔 💭 This seems to work "for real" but not on the automated tests... Perhaps I should double-check how these repositories are created...

[rudymatela]

The culprit line is here:

https://github.com/channable/hoff/blob/master/tests/EventLoopSpec.hs#L166-L169

It seems --single-branch does more than cloning just a single branch but rather changes the way git handles the repository so that it avoids keeping track of other branches altogether!

This was hard to find, I threadDelayed before and after the push to check the status of the repositories manually...

[rudymatela]

Before rebase on top of #155: junk/merge-train/integration/2

[rudymatela]

@ReinierMaas Hey, thanks for the review again. I think I have addressed your suggestions with my last five commits: 4b21410, 44e03dd, a699fd2, 75d9176 and 6b60bd7.

Can you please take a look again?

[ReinierMaas]

LGTM! Thanks for implementing. Now that I have reviewed it I want to see it in action.

[rlycx]

Nice work! Really hyped to see this in action :D

I've added some comments and suggestions, some of them we already discussed while going over the code in person (but I'm adding them here for completeness' sake anyway).

[rudymatela]

Nice work! Really hyped to see this in action :D

I've added some comments and suggestions, some of them we already discussed while going over the code in person (but I'm adding them here for completeness' sake anyway).

@RileyApeldoorn Thanks for the review. I think I have addressed your comments now.

[rlycx]

Nice work!!

[alter2000]

I love this! The code LGTM and matches the logic described on #77 but I'm not sure about the tests as I haven't had enough time to look at them. Thank you Rudy 🥳 !

[rudymatela]

@ReinierMaas, @RileyApeldoorn, @alex-mckenna and @alter2000 Thank you all for the detailed reviews. The PR was quite big so I think it was after all useful getting many eyes 👀 to look at this.

I will try to deploy this tomorrow early morning.

@OpsBotPrime merge

[OpsBotPrime]

Pull request approved for merge by @rudymatela, rebasing now.

[OpsBotPrime]

Rebased as 9497f8d, waiting for CI …

[OpsBotPrime]

CI job started.