v0.9.0

Minor Changes

• fa895fa: Every deploy now says how much CI time Scry took, warns when story execution is over budget, and records the time with the build (ISSUES.md #54).

  What went wrong before: a 461-story Storybook preview spent 19-21 minutes rendering stories one at a time (40 dialogs waited 15 s each), used 61% of an account's GitHub Actions minutes in a month, and nothing in the run said so. No line gave the execution time, no workflow set timeout-minutes, and the build recorded no time at all.

  What you see now:

  • A duration line after the capture: Story execution: 461 stories in 3.5 min (4 workers), budget 5.8 min.
  • Over the budget (120 s + 0.5 s × declared stories; change it with SCRY_EXECUTE_BUDGET_BASE_S / SCRY_EXECUTE_BUDGET_PER_STORY_S), a ::warning:: annotation on the run naming the time, the budget and where it went. A warning only: the exit code does not change.
  • The build records its CI timings (analyze, execute, archive, upload and deployer total; story counts; time lost per reason; sbcov and deployer versions; runner kind; Actions run id and attempt; budget). Anything that could not be measured is left out, never sent as 0. The first part rides on the existing build request; the final record goes to a new upload-service route. An upload service without that route is told apart from an error: the upload service does not record CI timings yet; not stored, once, and the deploy result is unchanged.
  • Whole-job time, when the job may read Actions (permissions: actions: read) and GITHUB_TOKEN is passed: CI time recorded: deployer 4.4 min, job 6.1 min so far (Actions API). Otherwise deployer time only (job start unknown: <reason>). This never fails a deploy.
  • SCRY_CONCURRENCY and SCRY_RENDER_TIMEOUT_MS (or concurrency / renderTimeoutMs in .storybook-deployer.json) are forwarded to scry-sbcov 0.7+ as --concurrency / --render-timeout; with an older scry-sbcov they are not sent and the log says so.
  • Generated workflows install @scrymore/scry-deployer@^0.9.0, set timeout-minutes: 20 on the Storybook job and grant actions: read. The push workflow had no permissions block before; the one it has now lists contents: read, actions: read and packages: read, and every other scope (for example write access your own added steps relied on) is dropped: add what your steps need.
  • timeout-minutes: 20 assumes scry-sbcov 0.7 (four stories at a time, 5 s render limit), which 0.9.0 installs. If you run an older scry-sbcov (SCRY_SBCOV_CMD, or your own pin) on a large Storybook, one run can take longer than 20 minutes and be stopped: raise the limit in your workflow.

  What you must do:

  1. Regenerate your workflows: npx -y @scrymore/scry-deployer@^0.9.0 update-workflows (the older ^0.7.0 and ^0.8.0 ranges do not pick up 0.9.0). If you keep a hand-written workflow, add timeout-minutes: 20 to the Storybook job, actions: read to its permissions (a permissions block drops every scope it does not list, so keep contents: read), and GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} to the deploy step's env.
  2. Nothing else. Without step 1 the deployer still records its own time and says the job start is unknown.