Pythonplot extension by j-fu · Pull Request #101 · WIAS-PDELib/GridVisualize.jl

j-fu · 2025-11-18T08:12:22Z

Addresses #62

- Introduce new PythonPlotType - Let PythonPlotType and PyPlotType have PyCommonType as supertype - Move pyplot code to ext/pycommon.jl which is included by both PyPlot and PythonPlot extensions - Disable some of the pyplot code for PythonPlot if it crashes.

Copilot

Pull Request Overview

This pull request adds support for PythonPlot.jl as an alternative to PyPlot.jl by refactoring the common plotting code and introducing a new extension module. The main changes involve creating a shared codebase for both Python-based plotting backends while maintaining backward compatibility with PyPlot.

Introduces PythonPlotType and AbstractPythonPlotterType to the type hierarchy for better code organization
Refactors PyPlot implementation by extracting common code into pycommon.jl shared by both PyPlot and PythonPlot extensions
Updates heuristic detection functions to distinguish between PyPlot and PythonPlot packages

Reviewed Changes

Copilot reviewed 6 out of 7 changed files in this pull request and generated 8 comments.

Show a summary per file

File	Description
src/dispatch.jl	Adds PythonPlot type hierarchy, updates plotter detection logic, and exports new types
src/GridVisualize.jl	Exports PythonPlotType to make it available to users
ext/pycommon.jl	New shared implementation file for Python-based plotters containing all plotting logic
ext/GridVisualizePythonPlotExt.jl	New extension module for PythonPlot.jl support
ext/GridVisualizePyPlotExt.jl	Refactored to use shared pycommon.jl instead of duplicated code
README.md	Updates documentation to mention PythonPlot support alongside PyPlot
Project.toml	Adds PythonPlot as a weak dependency with version compatibility constraints

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2025-11-18T08:18:48Z

src/GridVisualize.jl

 export plottertype, available_kwargs
 export default_plotter!, default_plotter
-export PyPlotType, MakieType, PlotsType, VTKViewType, PlutoVistaType, MeshCatType
+export PyPlotType, PythonPlotType, MakieType, PlotsType, VTKViewType, PlutoVistaType, MeshCatType


The new function ispythonplot is defined and used internally but not exported from the module (line 37), unlike the similar functions ispyplot, ismakie, isplots, etc. that are exported (line 33 in GridVisualize.jl). For consistency with the existing API, consider adding ispythonplot to the export list.

Copilot · 2025-11-18T08:18:48Z

README.md



-### [PyPlot](https://github.com/JuliaPy/PyPlot.jl):
+### [PyPlot/PythonPlot](https://github.com/JuliaPy/PyPlot.jl):


The section header mentions both PyPlot and PythonPlot but the link only points to PyPlot.jl (https://github.com/JuliaPy/PyPlot.jl). Consider also adding a link to PythonPlot.jl (https://github.com/stevengj/PythonPlot.jl) for completeness, for example: ### [PyPlot](https://github.com/JuliaPy/PyPlot.jl) / [PythonPlot](https://github.com/stevengj/PythonPlot.jl):

Suggested change

### [PyPlot/PythonPlot](https://github.com/JuliaPy/PyPlot.jl):

### [PyPlot](https://github.com/JuliaPy/PyPlot.jl) / [PythonPlot](https://github.com/stevengj/PythonPlot.jl):

Copilot · 2025-11-18T08:18:49Z

src/dispatch.jl

 Heuristically check if Plotter is PyPlot
 """
-ispyplot(Plotter) = (typeof(Plotter) == Module) && isdefined(Plotter, :Gcf)
+ispyplot(Plotter) = (typeof(Plotter) == Module) && isdefined(Plotter, :Gcf)  && !isdefined(Plotter, :plotshow)


Double space detected between :Gcf) and &&. For consistency with the rest of the codebase, use single space: isdefined(Plotter, :Gcf) && !isdefined(Plotter, :plotshow)

Suggested change

ispyplot(Plotter) = (typeof(Plotter) == Module) && isdefined(Plotter, :Gcf) && !isdefined(Plotter, :plotshow)

ispyplot(Plotter) = (typeof(Plotter) == Module) && isdefined(Plotter, :Gcf) && !isdefined(Plotter, :plotshow)

Copilot · 2025-11-18T08:18:49Z

src/dispatch.jl

 Create a  grid visualizer

-Plotter: defaults to `default_plotter()` and can be `PyPlot`, `Plots`, `VTKView`, `Makie` or `PlutoVista´.
+Plotter: defaults to `default_plotter()` and can be `PyPlot`, `PythonPlot`, `Plots`, `VTKView`, `Makie` or `PlutoVista´.


The closing backtick uses an incorrect character (´ - acute accent) instead of a standard backtick (). Change PlutoVista´toPlutoVista`

Suggested change

Plotter: defaults to `default_plotter()` and can be `PyPlot`, `PythonPlot`, `Plots`, `VTKView`, `Makie` or `PlutoVista´.

Plotter: defaults to `default_plotter()` and can be `PyPlot`, `PythonPlot`, `Plots`, `VTKView`, `Makie` or `PlutoVista`.

Copilot · 2025-11-18T08:18:49Z

src/dispatch.jl

+    elseif ispyplot(Plotter)
+        return PyPlotType


Duplicate conditional check detected. Lines 144-145 are identical and both check ispyplot(Plotter) returning PyPlotType. The second check (lines 144-145) should be removed as the intended logic for ispythonplot is already handled correctly on lines 146-147.

Suggested change

elseif ispyplot(Plotter)

return PyPlotType

Copilot · 2025-11-18T08:18:49Z

src/pycommon.jl

+
+
+function initialize!(p, ::Type{T}) where {T <: AbstractPythonPlotterType}
+    PyPlotter = p.context[:Plotter]


The PyPlot-specific 3D plotting initialization PyPlot.PyObject(PyPlot.axes3D) was removed from the old PyPlotExt but not added to the common pycommon.jl. This line is required for 3D plotting support in PyPlot (see JuliaPy/PyPlot.jl#351). Since pycommon.jl is now shared by both PyPlot and PythonPlot, this initialization should be conditionally added only for PyPlot, possibly by checking the plotter type or having separate initialization methods.

Suggested change

PyPlotter = p.context[:Plotter]

PyPlotter = p.context[:Plotter]

# PyPlot-specific 3D plotting initialization (see JuliaPy/PyPlot.jl#351)

if ispyplot(PyPlotter)

# Import PyPlot only if needed

import PyPlot

PyPlot.PyObject(PyPlot.axes3D)

end

Copilot · 2025-11-18T08:18:50Z

ext/pycommon.jl

+        if size(triangulateio.segmentlist, 2) > 0
+            lines = []
+            rgb = []
+            markermax = maximum(triangulateio.segmentmarkerlist)
+            # see https://gist.github.com/gizmaa/7214002
+            for i in 1:size(triangulateio.segmentlist, 2)
+                x1 = triangulateio.pointlist[1, triangulateio.segmentlist[1, i]]
+                y1 = triangulateio.pointlist[2, triangulateio.segmentlist[1, i]]
+                x2 = triangulateio.pointlist[1, triangulateio.segmentlist[2, i]]
+                y2 = triangulateio.pointlist[2, triangulateio.segmentlist[2, i]]
+                push!(lines, [[x1, y1], [x2, y2]])
+                push!(rgb, frgb(PyPlotter, triangulateio.segmentmarkerlist[i], markermax))
+                ax.plot([x1, x2], [y1, y2], color = frgb(PyPlotter, triangulateio.segmentmarkerlist[i], markermax))
+            end


The lines variable is created and populated but never used. In the old PyPlot implementation, this was used with PyPlot.matplotlib.collections.LineCollection, but now individual ax.plot calls are made on line 916. Consider either removing the unused lines and rgb variables (lines 905-906, 915) or restoring the LineCollection approach for better performance when drawing many segments.

Copilot · 2025-11-18T08:18:50Z

src/pycommon.jl

+        for i in 1:length(rgb)
+            ax.plot([c1[i][1], c2[i][1]], [c1[i][2], c2[i][2]], color = rgb[i], linewidth = 3)
+        end
+        for i in 1:nbfaceregions
+            ax.plot(coord[1, 1:1] * gridscale, coord[2, 1:1] * gridscale; label = "$(i)", color = rgbtuple(cmap[i]))
+        end
+    end


In the refactored code, 2D gridplot boundary face region rendering was changed from using PyPlot.matplotlib.collections.LineCollection (more efficient for rendering many lines) to individual ax.plot calls in a loop (lines 295-296). This change may negatively impact performance when rendering grids with many boundary face regions. Consider restoring the LineCollection approach for better performance.

Suggested change

for i in 1:length(rgb)

ax.plot([c1[i][1], c2[i][1]], [c1[i][2], c2[i][2]], color = rgb[i], linewidth = 3)

end

for i in 1:nbfaceregions

ax.plot(coord[1, 1:1] * gridscale, coord[2, 1:1] * gridscale; label = "$(i)", color = rgbtuple(cmap[i]))

end

end

# Use LineCollection for efficient rendering of boundary face region lines

segments = [[[c1[i][1], c1[i][2]], [c2[i][1], c2[i][2]]] for i in 1:length(c1)]

lc = PyPlotter.matplotlib.collections.LineCollection(segments, colors = rgb, linewidths = 3)

ax.add_collection(lc)

for i in 1:nbfaceregions

ax.plot(coord[1, 1:1] * gridscale, coord[2, 1:1] * gridscale; label = "$(i)", color = rgbtuple(cmap[i]))

end

Move pycommon to src, make py extensions trivial This in principle makes it possible to use pyplot and pythonplot at the same time, though it doesn't seem to work for other reason

j-fu added 3 commits October 22, 2025 14:31

start pythonplot extension

44c35c2

- Introduce new PythonPlotType - Let PythonPlotType and PyPlotType have PyCommonType as supertype - Move pyplot code to ext/pycommon.jl which is included by both PyPlot and PythonPlot extensions - Disable some of the pyplot code for PythonPlot if it crashes.

Merge branch 'main' into jf/pythonplot

d993718

fix pythonplot issues, PyCommonType-> AbstractPythonPlotterType

fe4eb4e

j-fu requested a review from Copilot November 18, 2025 08:12

Copilot started reviewing on behalf of j-fu November 18, 2025 08:13 View session

Copilot finished reviewing on behalf of j-fu November 18, 2025 08:17

Copilot AI reviewed Nov 18, 2025

View reviewed changes

j-fu added 7 commits November 18, 2025 07:35

address copilot's suggestions

89f28ff

fix missing docstring mentions

de7213f

Modify architecture a bit

258ff31

Move pycommon to src, make py extensions trivial This in principle makes it possible to use pyplot and pythonplot at the same time, though it doesn't seem to work for other reason

bump version + changelog entry

57c9b48

runic

c9e786c

fix docs

2d85c4d

Fix pythonplot in notebooks: return gcf() from reveal

2e7fc7e

j-fu merged commit dbb8943 into main Nov 19, 2025
9 checks passed

j-fu deleted the jf/pythonplot branch November 19, 2025 04:38

j-fu mentioned this pull request Dec 13, 2025

Add PythonPlot backend #62

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Pythonplot extension#101

Pythonplot extension#101
j-fu merged 10 commits intomainfrom
jf/pythonplot

j-fu commented Nov 18, 2025

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Nov 18, 2025

Uh oh!

Copilot AI Nov 18, 2025

Uh oh!

Copilot AI Nov 18, 2025

Uh oh!

Copilot AI Nov 18, 2025

Uh oh!

Copilot AI Nov 18, 2025

Uh oh!

Copilot AI Nov 18, 2025

Uh oh!

Copilot AI Nov 18, 2025

Uh oh!

Copilot AI Nov 18, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants



		### [PyPlot](https://github.com/JuliaPy/PyPlot.jl):
		### [PyPlot/PythonPlot](https://github.com/JuliaPy/PyPlot.jl):

	ispyplot(Plotter) = (typeof(Plotter) == Module) && isdefined(Plotter, :Gcf) && !isdefined(Plotter, :plotshow)
	ispyplot(Plotter) = (typeof(Plotter) == Module) && isdefined(Plotter, :Gcf) && !isdefined(Plotter, :plotshow)

	Plotter: defaults to `default_plotter()` and can be `PyPlot`, `PythonPlot`, `Plots`, `VTKView`, `Makie` or `PlutoVista´.
	Plotter: defaults to `default_plotter()` and can be `PyPlot`, `PythonPlot`, `Plots`, `VTKView`, `Makie` or `PlutoVista`.



		function initialize!(p, ::Type{T}) where {T <: AbstractPythonPlotterType}
		PyPlotter = p.context[:Plotter]

-    PyPlotter = p.context[:Plotter]
+    PyPlotter = p.context[:Plotter]
+    # PyPlot-specific 3D plotting initialization (see JuliaPy/PyPlot.jl#351)
+    if ispyplot(PyPlotter)
+        # Import PyPlot only if needed
+        import PyPlot
+        PyPlot.PyObject(PyPlot.axes3D)
+    end

Conversation

j-fu commented Nov 18, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Copilot AI Nov 18, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 18, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 18, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 18, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 18, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 18, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 18, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 18, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants