Repository navigation
Deploy DAGs on dags-only runs with no diff, and reject image-name where it can't apply - #165
Merged
Merged
Conversation
jeremybeard
added this pull request to stack #169
October 15, 2026 17:25
jeremybeard
force-pushed
the
dags-only-fallbacks
branch
from
October 15, 2026 19:03
a51bfa1 to
8c72ca9
Compare
jeremybeard
force-pushed
the
dags-only-fallbacks
branch
from
October 15, 2026 19:30
8c72ca9 to
71678f1
Compare
jeremybeard
changed the title
jeremybeard
force-pushed
the
dags-only-fallbacks
branch
from
October 16, 2026 18:40
71678f1 to
1d3b092
Compare
jeremybeard
marked this pull request as ready for review
October 16, 2026 19:11
wendtek
reviewed
Oct 16, 2026
| # validate deploy-type-specific inputs | ||
| # a custom image is deployed as it was built, so there is nothing to test or build, and dags-only never | ||
| # deploys an image; the CLI rejects the equivalent flags alongside --image-name | ||
| if [[ "${{ inputs.image-name }}" != "no-custom-image" ]]; then |
There was a problem hiding this comment.
Do we need to also validate the action before going into this block? Like, if the action is going to be delete-deployment-preview, should it matter what the rest of the variables are? The below if [[ ${{ inputs.action }} == delete-deployment-preview ]]; then branch doesn't seem to care what type it is, but we won't make it there if we exit here.
wendtek
approved these changes
Oct 16, 2026
wendtek
left a comment
wendtek
left a comment
There was a problem hiding this comment.
One possible nit. I leave it up to your judgement on whether it's a valid scenario.
neel-astro
approved these changes
Oct 19, 2026
| # history is already present from the fetch-depth: 0 checkout, so no fetch is needed; | ||
| # a `git fetch` here would also fail on private repos due to persist-credentials: false | ||
| if ! git cat-file -e "$GITHUB_EVENT_BEFORE" 2>/dev/null; then | ||
| echo "Commit $GITHUB_EVENT_BEFORE does not exist, falling back to image deploy." |
Collaborator
There was a problem hiding this comment.
this log is not consistent with the behavior now, correct? Maybe it was misleading before this PR as well, if I understand the PR description correctly 😅
…re it can't apply Co-Authored-By: Claude <[email protected]>
jeremybeard
force-pushed
the
dags-only-fallbacks
branch
from
October 19, 2026 13:19
1d3b092 to
fb382d1
Compare
A delete deploys nothing, so inputs that only shape a deploy shouldn't stop it. A delete step sharing a deploy step's inputs would otherwise start failing after an upgrade that newly rejects one of them, leaving the preview running. Co-Authored-By: Claude <[email protected]>
Without a diff, the action deploys everything deploy-type covers, which for dags-only is a DAG deploy rather than an image deploy. Co-Authored-By: Claude <[email protected]>
This branch is waiting to be deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
dags-onlydeployed nothing when there was no diff to go on (workflow_dispatch, a push that creates a branch, a missing diff base), where the other deploy types deploy everything. It now deploys DAGs in that case, when DAG deploys are enabled. The DAG deploy e2e test covers a dispatch.image-nameis now rejected where it can't apply, as the Astro CLI rejects the equivalent flags alongside--image-name:dags-only, which never deploys an image; this used to deploy nothing at allparseorpytestoninferandimage-and-dags(image-onlydoesn't pass them on)build-secretsA custom image is deployed as it was built, so none of these combinations did anything useful. Input validation tests cover each.