HMPlot

Offsets and dragging

Rows dragged with drag= move by offsets that views read through moved() and offset().

hm.Points(x=hm.moved("x"), y=hm.moved("y"), drag=True)

Functions

hm.moved(column)expression
A numeric table column plus each row's offset: what a view reads to follow drags, as in x={moved("weight")}. Offsets are in the column's units, so log(moved("weight")) moves through a log.
hm.offset(column)expression
Each row's offset in the column alone, 0 where it has none: set by dragging or from code.

Usage

drag on a point mark makes its rows draggable. true drags along every numeric axis of the view; "x", "y", "xy" or "xyz" name the axes, and a categorical axis is never dragged, so a strip plot drags along its numeric axis. A press on a draggable row starts a drag, anywhere else the plot pans, Shift-drag still brushes, and Escape cancels.

A drag never changes the table. It writes offsets, per-row deltas in each column’s own units, on the table’s rows, and every view of the table shares them. A view follows only where it reads them: moved("weight") is the column plus its offsets and offset("weight") the offsets alone, so x={moved("weight")} follows a drag and x="weight" does not, the dragged mark included. That is what lets one drag move a row in every cell of a scatterplot matrix: each cell’s x and y stand for its two fields, and moved("x") reads the offsets of the field the cell shows.

The drag goes back through the channel’s expression and the transforms above the mark, so the row lands under the pointer through log(moved("w")), a polar or map frame, or a Calculate column. A Filter that reads moved() hides and shows the row while it is dragged. A row of a Group drags its members: a mean or median moves them all by the change, a sum splits it in proportion to each member. Dragging a selected row moves the selection. Rows being dragged skip their transitions and stay under the pointer; Escape and resetOffsets() animate.

getOffsets(), resetOffsets() and applyOffsets() on a Data, Provider or Plot read, clear and apply the offsets; ondrag and ondragend report the dragged rows’ keys and each column’s change. In Python, table.offsets is a DataFrame of key, column and delta, table.apply_offsets() a new Table with the moved values, and table.set_offsets() and table.clear_offsets() change them. See docs/DATA.md, “Dragging”.

Related examples

Dragging points

Dragging cities on a map