This release expands the file formats Flowfile can read and write, makes the whole node palette searchable by tags, and brings a big wave of improvements to the in-browser editor (flowfile-editor) — alongside fixes to output paths and Save-As.
New file formats — Arrow IPC, NDJSON & Avro (#536)
Flowfile can now read and write three more formats, in both the visual editor and the Python API:
- Arrow IPC / Feather (
.arrow,.ipc,.feather), NDJSON / newline-delimited JSON (.ndjson,.jsonl), and Avro (.avro) - The Read Data and Write Data nodes detect and configure these alongside the existing CSV, Excel, and Parquet support
- Python API gains
read_ipc/read_ndjson/read_avro(plus lazyscan_ipc/scan_ndjson) andwrite_ipc/write_ndjson/write_avro - Pick a compression codec per format when writing — Parquet (zstd/snappy/gzip/lz4/brotli), IPC/Arrow (lz4/zstd), NDJSON (gzip/zstd), Avro (snappy/deflate)
Searchable node tags (#525)
- Every node now carries searchable tags, so typing concepts like
csv,vlookup,dedupe,window,s3, orkafkasurfaces the right node even when the term isn't in its name - Palette search now matches both node names and tags across the desktop and web designer
- In the browser editor, full-app-only nodes (Database, Cloud, REST API, Kafka, Google Analytics, SQL Query, Python Script, Window Functions) are now discoverable — hidden by default but revealed via a new "Show full-app nodes" toggle or by search, shown locked with a "Learn more" docs link, plus a node info popup
Browser editor (flowfile-editor) improvements
The in-browser WASM editor got a lot broader this release:
- Five new nodes — Formula, Cross Join, Union, Record ID, and Rename — broadening what you can build with zero backend (#518)
- Dynamic Rename node to bulk-rename columns by prefix, suffix, formula, or first-row values; a richer Formula function browser with a category tree, hover docs, and type-aware autocomplete; and fixed preview rendering for date/time/decimal/binary columns (#519)
- Delete connections by hovering an edge and clicking the remove button (#523)
- Explicit-only execution — opening a node's Settings or Table panel no longer silently runs your pipeline; data is computed only when you run it (Ctrl+E / Run Now / Apply / Fetch data). Single-click opens Settings, double-click opens Settings + Table (#521)
Fixes & improvements
- Output paths (desktop): output nodes now default to a real, writable directory (your last write folder or home) instead of an ambiguous
., and existing nodes carrying.are converted to an absolute path on open. Write history is tracked separately from the file browser (#529) - Save-As: saving a flow to the catalog under a new name now creates a copy instead of deleting the original, flow names can be free-form labels (auto-slugified on disk), and the catalog picker hides flows whose backing file was deleted (#522)
- Lighter installs: database writes no longer require pandas (SQLite uses
sqlite3, others use SQLAlchemy Core), and the Database Writer node now validates schema/table names inline (#528) - New documentation for the REST API Reader, Kafka Source, Google Analytics Reader, Window Functions, and SQL Query nodes (#535)
- The browser editor is now positioned as a standalone, embeddable npm package —
npm install flowfile-editor(#520) - Internal cleanup of the WASM editor — dead code and debug logging removed (#532)
- Dependency bumps: axios 1.15.2 → 1.16.0 and form-data 4.0.5 → 4.0.6 (frontend), vite / plugin-vue / vitest (wasm) (#511, #530, #531, #510)
Full changelog: v0.12.1...v0.12.2