You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Our CI configuration displays test results from each package on a separate line, making it a lot easier to see which test is failing:
To accomplish this, our .github/workflows/test-all-packages.yml has a list of steps, each of which runs a single package's tests:
- name: yarn test (acorn-eventual-send)run: cd packages/acorn-eventual-send && yarn test
- name: yarn test (agoric-cli)run: cd packages/agoric-cli && yarn test
- name: yarn test (assert)run: cd packages/assert && yarn test
- name: yarn test (autoswap-frontend)run: cd packages/autoswap-frontend && yarn test...
The actual set of packages is defined by our top-level package.json, in the workspaces: key:
Having this defined in two separate places is a maintenance hassle and a footgun (it's easy to add a new package and forget to add it to the CI list, which means tests don't get run).
This ticket is about creating a script that parses both files and complains if the lists don't contain the same items. We can probably allow the order to differ: we currently keep workspaces: in dependency order (so that yarn build constructs things in the correct order), whereas package.json is currently alphabetical. (using lerna might allow us to keep both in alphabetical order). But the set of entries from one must equal the set from the other. The script should display the difference. Bonus points for a script that can generate a new test-all-packages.yml from the package.json and a template of some sort.
The text was updated successfully, but these errors were encountered:
warner
changed the title
build GitHub Action test list from package.json
build GitHub Action test list from package.json, or at least compare them automatically
Feb 19, 2020
Our CI configuration displays test results from each package on a separate line, making it a lot easier to see which test is failing:
To accomplish this, our
.github/workflows/test-all-packages.yml
has a list of steps, each of which runs a single package's tests:The actual set of packages is defined by our top-level
package.json
, in theworkspaces:
key:Having this defined in two separate places is a maintenance hassle and a footgun (it's easy to add a new package and forget to add it to the CI list, which means tests don't get run).
This ticket is about creating a script that parses both files and complains if the lists don't contain the same items. We can probably allow the order to differ: we currently keep
workspaces:
in dependency order (so thatyarn build
constructs things in the correct order), whereaspackage.json
is currently alphabetical. (usinglerna
might allow us to keep both in alphabetical order). But the set of entries from one must equal the set from the other. The script should display the difference. Bonus points for a script that can generate a newtest-all-packages.yml
from thepackage.json
and a template of some sort.The text was updated successfully, but these errors were encountered: