The proposed power stroke LPE is as follow:
This LPE allows a modulated stroke from manually entered values.
Further development could allow direct width reading from a pressure-enabled device.
The user simply creates the line, it might for example be based on an outline.
Then the user applies the LPE to his shape, it overrides the normal stroke rendering and allows custom line width at different nodes of the path.
Only the stroke-width property of the path is overridden, the rest of the stroke properties keep working as usual, when relevant.
Se image above. User selects a nodes or a set of nodes and affects values to those. When a node is selected, you can see what its value is by the change in values in the number entry or by the state of the push button.
Additionally, hovering over a node could show the value of a node. Graphical feedback could even possibly set different icons for nodes that are set and others that are interpolated.
Comments by JohanE
This is very well doable, but would require a new "Nodepath" implementation that enables writing special values (preferably a struct) per node. There is already something like this, the "nodetypes" attribute, but it is too restrictive. Expanding this would enable this effect, and opens up many other possibilities. Nice idea !