Skip to content
dev

Migration Guide ​

Breaking changes and migration steps between releases. Read the section for the version you are upgrading to before updating your scripts and existing jobs.

v0.9 → v0.10 ​

v0.10 has two breaking changes. Both require action; neither alters your FHIR data.

1. aether.yaml is now a required positional argument ​

The --config flag has been removed, and Aether no longer discovers a config file implicitly. v0.9 searched --config → ./aether.yaml → ~/.config/aether/aether.yaml. v0.10 requires you to pass the config path explicitly, as the first positional argument of every command that needs configuration.

Before (v0.9):

bash
# Relied on the --config flag or implicit ./aether.yaml discovery
aether pipeline start your-crtdl.json
aether pipeline status <job-id>
aether pipeline continue <job-id>
aether job list

# or with the explicit flag
aether --config /path/aether.yaml pipeline start your-crtdl.json

After (v0.10):

bash
aether pipeline start    aether.yaml your-crtdl.json
aether pipeline status   aether.yaml <job-id>
aether pipeline continue aether.yaml <job-id>
aether job list          aether.yaml

What to update:

  • Scripts, cron jobs, and CI: insert the config path as the first argument of each aether subcommand, and remove any --config flag — it no longer exists.
  • No implicit discovery: a bare ./aether.yaml or ~/.config/aether/aether.yaml is not picked up automatically. Point at the file explicitly.

2. DIMP output directory renamed pseudonymized/ → dimp/ ​

The DIMP step now writes to jobs/<job-id>/dimp/ instead of jobs/<job-id>/pseudonymized/, matching the step name and the rest of the per-step layout. The file contents are unchanged — only the directory name moved.

What to update:

  • Downstream tooling and scripts that read pseudonymized output by path must switch jobs/<job-id>/pseudonymized/ → jobs/<job-id>/dimp/.

  • In-flight jobs from v0.9: a job whose DIMP step already ran has its data under pseudonymized/, but v0.10 looks under dimp/. To resume such a job, rename the directory first:

    bash
    mv jobs/<job-id>/pseudonymized jobs/<job-id>/dimp

    Jobs started fresh on v0.10 need no action.

Worth knowing (non-breaking) ​

  • Per-job logs: each job now writes its own log file into its job directory, so logs live alongside that job's data.
  • TORCH URL handling: scheme-less, host-prefixed URLs returned by TORCH are handled correctly, fixing double-host URLs.
  • Extraction reporting: status now shows the extraction summary instead of a stale cohort-size diagnostic.

See the v0.10.0 release notes for the full changelog.

Healthcare data integration made simple