Release v0.22.3

One flow open in two places now stays in sync, the canvas notebook can move to its own window, and the Python editors show better signature help.

Two tabs on one flow stay in sync

Open the same flow in two browser tabs, or in the designer and a popped-out notebook, and each now follows the other's changes. Every flow carries a revision number that moves on each change that lands: an edit, an undo or redo, a run starting or ending, a save. Each window follows the flow's change feed and reloads the canvas when another window changes it, keeping its own viewport and selection. Its own changes are skipped. A flow closed elsewhere drops out of the tab bar, and a Save As in another window is followed to the flow's new id.

Two edits that used to overwrite each other silently are now refused:

• Settings saves. A settings drawer remembers the settings it opened with. If the node changed in another window in the meantime, saving asks: Keep mine overwrites, Discard mine closes the drawer so the next open shows the current settings, and closing the dialog keeps the drawer as it is.
• Node ids. Adding a node with an id already on the canvas used to replace that node; it is now refused. A notebook push also numbers new nodes above every id the canvas has held, including deleted nodes and undone placements, so a push can't reuse an id another window still refers to.

Notebook pop-out

Pop out in the Code panel moves the flow's notebook to its own window: a native window in the desktop app, a browser window on the web. The dock closes and shows Show window and Bring back until the window is closed; Return to designer in the window reopens the dock on that flow. One window holds a flow's notebook, and its kernel session, at a time. Closing the main desktop window closes its notebook windows too.

Signature help in the Python editors

Inside a call's parentheses in the Python Script and notebook editors, the popup shows the full docstring right after the (, then only the current parameter's description and the docstring's summary once an argument is typed. The highlighted parameter follows a name= or a partial name, not just the argument's position. Parameter descriptions are read from Google and numpydoc docstrings. The popup stays out of the way while you type a string or a member access like df., where column and member completions show instead.

Everything else

• The repo and the docs have a roadmap: ROADMAP.md, with the details on the docs site.
• The notebook docs page has more screenshots and a short round-trip video.

pip install flowfile, or the desktop app.