For a PR to manipulation-solutions that uses squash and merge, the single commit that is commited to master will have a brand new sha value (it won't be any of the sha's in the PRs commit history). Therefore, the workflow in README.md in the solutions repo is wrong (one has to first merge to manipulation-solutions, not the other way around as the README suggests).
A current solution is:
- Merge the solutions to
master in the manipulation-solutions repo
- In the PR in the
manipulation repo: Run solutions/install.py to generate a solutions_sha.txt with the correct commit sha
- CI/tests will now run in the PR in the
manipulation repo for the newly commited solutions.
The problem with this is that if the tests fail, then one has to go back and update the manipulation-solutions repo again. But I don't really know if there is a way to avoid this.
For a PR to
manipulation-solutionsthat uses squash and merge, the single commit that is commited tomasterwill have a brand new sha value (it won't be any of the sha's in the PRs commit history). Therefore, the workflow in README.md in the solutions repo is wrong (one has to first merge tomanipulation-solutions, not the other way around as the README suggests).A current solution is:
masterin themanipulation-solutionsrepomanipulationrepo: Runsolutions/install.pyto generate asolutions_sha.txtwith the correct commit shamanipulationrepo for the newly commited solutions.The problem with this is that if the tests fail, then one has to go back and update the
manipulation-solutionsrepo again. But I don't really know if there is a way to avoid this.