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
{{ message }}
Repository navigation
build-check: add a targets input for manual runs - #40
PRs and the weekly schedule do not change. They build each reference on its default target.
Manual runs (workflow_dispatch) get a new targets input:
empty: default target (same as before)
raspberrypi5 imx8mp-evk: every selected reference on each listed target
all: each reference on every entry in its own supported_targets
If a reference lists specific supported_targets and does not include the requested target, the pair is skipped and the detect job shows a notice. With all, a reference with supported_targets: "*" (or no list) uses its default target.
Target resolution moves from the build job into detect. The matrix is now {ref, target} pairs. Result rows and artifact names include the target, so one reference on several targets does not collide.
Ran the new detect logic locally against every reference. With no inputs, it gives the same targets as the old resolver. With a target list, it skips unsupported pairs. With all, it expands each reference's supported_targets.
This PR changes only the workflow, so the PR check itself should report no references to build. A manual run from this branch is the real test.
PRs and the weekly schedule still build each reference on its default
target. A manual run can now pass `targets` as a space-separated list,
or `all` to build each reference on every entry in its own
supported_targets. Pairs a reference does not support are skipped with
a notice.
Target resolution moves into the detect job, so the build matrix is a
list of {ref, target} pairs. Result rows and artifacts are keyed by
both, so one reference on several targets does not collide.
Repeated refs or targets made duplicate matrix jobs that collided on
artifact names. An unreadable avocado.yaml, or a ref that does not
exist, stopped detect so no ref built. Such a ref now stays in the
matrix and fails in its own build job.
The reason will be displayed to describe this comment to others. Learn more.
🔵 Needs a closer look
A non-mapping YAML configuration can abort detection and prevent healthy references from building.
No open issues 🎉
1 previously missed issue
In code that hasn't changed since last review
Validate YAML root type inside try to prevent global build failure
.github/workflows/build-check.yml:105
A valid YAML document with a non-mapping root, such as [qemuarm64], passes safe_load but raises AttributeError at y.get(...) outside the handler. Because resolution now runs in detect, this prevents every selected reference from building, not just the malformed reference. Validate the root type inside the try so the existing per-reference fallback handles this configuration error too.
The reason will be displayed to describe this comment to others. Learn more.
Checked the new targets input end to end: it reaches the python resolver through env and a quoted heredoc, and matrix.target only reaches run: blocks through env. I ran the resolver against all 35 references at this head. With no targets it picks the same target the old Resolve target step did for every reference, and targets=all expands to 72 jobs. Row files and artifact names now carry the target, so one reference on several targets no longer collides.
Nothing blocking. One thing worth considering as a follow-up: a mistyped target against references with an explicit supported_targets list skips every pair, so pairs ends up empty and the run finishes green with nothing built. Failing detect when a non-empty targets input yields zero pairs would keep red as the signal.
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
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.
Summary
workflow_dispatch) get a newtargetsinput:raspberrypi5 imx8mp-evk: every selected reference on each listed targetall: each reference on every entry in its ownsupported_targetssupported_targetsand does not include the requested target, the pair is skipped and thedetectjob shows a notice. Withall, a reference withsupported_targets: "*"(or no list) uses its default target.detect. The matrix is now{ref, target}pairs. Result rows and artifact names include the target, so one reference on several targets does not collide.Usage
gh workflow run build-check.yml -R avocado-linux/references \ -f refs="imx8mp-npu-pose" -f targets=allTesting
detectlogic locally against every reference. With no inputs, it gives the same targets as the old resolver. With a target list, it skips unsupported pairs. Withall, it expands each reference'ssupported_targets.