PR Summary

(replacement for PR #10153)

The embedded FigureCanvasWx currently can't be set to a fixed size.
When it is added to a sizer with proportion=0 and wx.EXPAND not set, then it will not expand over it's defined size, as expected. But when the visible area is reduced below the defined size, it will re-scale itself to fit the visible area because 'self.GetClientSize()' is used to read the size.
This could happen if the window size is reduced or a splitter is used.
When placed on a scrolled window, the problem does not occur, as then the VirtualSize will not change, even when the widget is not 100% visible.

This is not a big issue as the behaviour can easily be modified in a derived class. It would be nice though, to have the correct behaviour by default.

See attached file embedding_in_wx4_fixed.py for an example. With the modification from the pull request the size will stay fixed.
The only change is the sizer.Add call with proportion=0 and without EXPAND:

self.figure = Figure(figsize=(5, 4), dpi=100)
...
self.canvas = FigureCanvas(self, -1, self.figure)
self.sizer = wx.BoxSizer(wx.VERTICAL)
#self.sizer.Add(self.canvas, 1, wx.TOP | wx.LEFT | wx.EXPAND)
self.sizer.Add(self.canvas, 0, wx.TOP | wx.LEFT)

embedding_in_wx4_fixed.py.txt

I could commit the example file, but that's probably not worth it.

If the canvas is directly used as single child of a frame (i.e. instead of a panel), the behaviour is not modified by the pull request. This could be changed, but I don't see a use case for this.

Btw.: I just noticed that the embedding_in_wx4.py example includes an EVT_PAINT handler. I don't think that this is required, as the plot is static. Instead, it causes unnecessary renderings.

Also, the structure of fourier_demo_wx.py is a bit unusual. The usual hierarchy would be Frame->Panel->Sizer->Canvas. Actually it's now Frame->Sizer->Window->Canvas where the Window does not have any function and the sliders are drawn on the plain frame background.

I could submit a PR to update the demos, if desired.

PR Checklist

• Has Pytest style unit tests
• Code is PEP 8 compliant
• New features are documented, with examples if plot related
• Documentation is sphinx and numpydoc compliant
• Added an entry to doc/users/next_whats_new/ if major new feature (follow instructions in README.rst there)
• Documented in doc/api/api_changes.rst if API changed in a backward-incompatible way