-
Notifications
You must be signed in to change notification settings - Fork 20
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Still broken on first build to new branch #25
Comments
can you provide your circleci config? it is working fine in |
My build file is large and complex but this is the relevant parts. I think the issue is because of a rebase where a job's commit is no longer a valid commit as it was rebased and dropped. Maybe a return code of 128 can be treated as a silent failure and allow the scan to occur.
My log outputs something like:
|
it could be related to this one #28 |
We are also still experiencing this issue. (Using 1.0.0 of the orb, though) If it helps debug, here is the relevant config: jobs:
build:
<<: *js_defaults
steps:
- checkout
- compare-url/reconstruct:
project-path: ~/app
- restore_cache: *restore-yarn-cache
- run: "yarn --frozen-lockfile --production=false --prefer-offline"
- save_cache: *save-yarn-cache
- run:
name: "Build all changed packages."
command: |
echo "Requires each package to have a build command in the root package.json in the format of build:PACKAGE_NAME. E.g. build:main"
CIRCLE_COMPARE_URL=$(cat CIRCLE_COMPARE_URL.txt)
COMMIT_RANGE=$(echo $CIRCLE_COMPARE_URL | sed 's:^.*/compare/::g')
echo "Commit range: $COMMIT_RANGE"
for package in $(cat packages/deployable-packages.txt packages/serverless-js.txt); do
if [ -n "$(git diff $COMMIT_RANGE --name-status -- packages/$package)" ]; then
echo "Building ${package}"
NODE_ENV="production" yarn build:$package
else
echo "${package} not modified; no need to build"
fi
done
echo "Pruning node_modules"
chmod +x ./.circleci/scripts/node-prune
./.circleci/scripts/node-prune
echo "Persisting to workspace:"
- persist_to_workspace:
root: ~/
paths:
- app with a spam of
and then
Unfortunately this is a private repo so you can't explore the commits, but this is a simple PR created by dependabot, which just changes our PR comparing from a new branch with one commit, to master. Doing a second commit would fix the issue. |
Is someone looking into this problem? |
I am encountering this problem as well. My The build: https://circleci.com/gh/mernst/randoop/92 The build fails with
Everything is open source, so I hope this will permit this bug to be investigated and resolved. Also, is there any workaround? I would prefer my build to proceed, with the environment variables not set, rather than the build to fail. |
We're running into this issue as well. I've submitted a PR for what I believe should fix the bug. The |
- reverted orb patch test attempt (submitted PR for iynere/compare-url#25) - was missing dependency for diff detection - looks like yaml alias can't be used inline so there will be 1 extra place where env file path is hard-coded - minor renaming/cleanup
* checkpoint commit of circleci diff detection logic - relying on shared workspace and list of variables to export into $BASH_ENV - tried using aliases where possible for config reuse. may add more later - will add documentation/notes in config to indicate the 2 places where project dependencies must be captured * preparing local patch for compare-url orb fix * revert test and fix misc bugs in circleci config - reverted orb patch test attempt (submitted PR for iynere/compare-url#25) - was missing dependency for diff detection - looks like yaml alias can't be used inline so there will be 1 extra place where env file path is hard-coded - minor renaming/cleanup * fix/refactor dependency logic in circleci - should have just done this in python to begin with... - went ahead and added back the temporary hack/patch for the compare-url orb * fixed bug in dependency calculation script - no change path wasn't initializing variable. this simplifies it a bit more too * ensure env file exists even in no-op flow * whitespace change to trigger ci diff logic * fix bug in export variable naming - also including minor readme change to trigger ci diff flow * comment the java AE project out in circleci again - also triggering C# project flow to check for any fixup bugs * aesthetic change in ci diff dep order
Seems the fix to #21 has caused another error:
fatal: Not a valid commit name 29201976786eae7f3a86ed1678b91ef5a62bfc00
unknown return code 128 from git merge-base with base commit 29201976786eae7f3a86ed1678b91ef5a62bfc00, from job 1302
The text was updated successfully, but these errors were encountered: