DONE: Adding a cookiecutter template to make it easier to use exemplar as a template for new projects:
main ← enolan_cookiecutter1
opened 12:51AM - 10 Jun 25 UTC
This commit introduces a 'cookiecutter' directory that users can use to spin off… new projects using exemplar as a template. After invoking cookiecutter on the repository, users are prompted for a new project name, directory name, C++ version, paper, owner, and description, and these values are filled into the template.
It also adds two new checks into CI: that, when the template is filled in with the values for exemplar itself, the result is identical to exemplar; and, that if a name other than "exemplar" is specified for the project name, then the string "exemplar" does not appear in the output when the template is filled in.
These CI checks can be tested by maintainers of exemplar without
needing to update the pull request by running the script
check_cookiecutter.sh.
DONE: Completing a migration of the Docker infrastructure in infra to a new containers repository:
main ← enolan_containers1
opened 02:26PM - 23 Jun 25 UTC
The new repository, bemanproject/containers (https://github.com/bemanproject/con… tainers), is the new home for the part of the infra repository that handles building the CI test and devcontainer Docker images used by Beman repositories. The git history of these files has been transferred over as well, using a git filter-repo command.
DONE: Fixing a GitHub CI bug that caused CI to run twice on exemplar pull requests:
main ← enolan_citriggerfix1
opened 04:10PM - 22 Jun 25 UTC
When updating the branch associated with a pull request, previously, CI would tr… igger twice: once for the update to the branch, and another time for the update to the pull request itself.
This commit addresses the issue by restricting CI runs for updates to branches to those that update the main branch itself.
DONE: One line removal of a stray debug setting in the CI script to check the cookiecutter template:
main ← enolan_cookiecutterdbg1
opened 04:45PM - 23 Jun 25 UTC
The set +x here was needed for debugging but checking it in was accidental. This… commit removes it.
@ednolan - w.r.t to the duplicative PR – do we need to fix this in other repos? I feel like this isn’t happening in the scope repo – or I’m not sure how to tell.
It is happening with scope. For example, this update to pull request 34 triggered two CI runs:
I plan to address this as part of a broader effort to standardize how CI works across Beman repositories.
Maybe I don’t understand the change here – those we 2 different commits pushed at different times I think – are we saying we don’t want to rerun CI during the review?
I’ll make the change on the scope branch to see the behavior.
They appear to be two different commits but really they’re running CI for the same commit. One CI run is for the push to the feature branch, so uses the actual commit name, whereas the other run is for the update to the pull request, so it uses the name of the pull request itself (taken from the first commit in the chain), but is really also running for that commit.
