Celebi / Releases

CelebiChrono v1.0.0b2

Workflow & shell

  • Project-root-relative paths: paths starting with @/ (or bare @) in add-input, add-algorithm, and add-parameter-subtask resolve relative to the current project root. Tab completion understands @/ prefixes.
  • Selective impress/submit by object names: impress a b operates only on the named sub-objects; the no-argument form still processes all sub-objects collectively. submit accepts --runner <name> and positional object names, e.g. submit --runner cern a b.

Workaround improvements

  • --reference <algorithm> copies the referenced algorithm into a reference/ subdirectory of the workaround space for side-by-side code inspection.
  • --skip-input <task> skips mounting the impression of a specified input task.
  • Filelist manifest: workaround_preshell records the initial algorithm files to filelist.yaml; workaround_postshell copies listed files back and deletes files not in the manifest.

LHCb support

  • New LHCb Analysis Productions (AP) data-list task type (lhcb_ap_datalist): create-lhcb-ap-list creates the task with example AP configuration; add-apd-token stores the APD token in .celebi/config.local.json.
  • Deposit uploads .celebi/config.local.json alongside the impression so Yuki can generate lbapi_key.json server-side.
  • datalist and lhcb_ap_datalist tasks validate without requiring an algorithm.

Runner management

  • update-runner <name> modifies runner settings (URL, token, backend type, Kerberos, EOS mount point).
  • runners now displays the full runner configuration.

Fixes & quality

  • collect: start message, HTTP/connection errors surfaced, warning when nothing is collected, file counts and per-file skip/fail reasons reported.
  • Fix arc cache invalidation so add-algorithm/remove-input changes are visible immediately in ls.
  • Fix book-reana mid-upload connection-loss error message.
  • Fix status display ([coda][finished] for completed workflows).
  • Full-project pylint score brought to 10.00/10.

Yuki (DITE server)

  • Structured collect reports from all workflow backends ({collected, skipped, failed} with per-file reasons); /collect* endpoints return JSON.
  • LHCb AP support: lbapi_key.json generation, apd-login, EOS path handling; Kerberos auto-enabled for tasks with an AP predecessor.
  • CVMFS repository support in Snakefile rule resources and reana.yaml.
  • Runner management endpoints: PATCH /update-runner, GET /runners-config.
  • SSH runner backend: upload inputs and rawdata over SFTP, run Snakemake remotely over SSH, poll and download outputs back to Yuki Storage.
  • Fix input/rawdata staging into imp<short>/stageout/ and backend-type persistence across reloads.
  • Completed jobs stored with legacy status finished ([coda][finished] in the client).