Ronald L. Sprouse

Perhaps you are right that partial overlap is a better general solution for lines. For my use case (selecting time ranges of audio waveforms) the distinction is not very important...

The `_hit_rect_against_index` approach does *not* include partial overlaps. For example, this selection for ``` p = figure(tools=['box_select,reset'], plot_width=400, plot_height=400) p.line([0, 1, 2, 3], [0, 1, -1, 0]) ``` selects only...

For dense time series like yours or mine I agree that not selecting partial overlaps is at least as intuitive as including them in the selection, and it should generally...