Bug summary

I ran into this bizarre bug, I am not sure if it's a skill issue on my end, but I am trying to embed a plot in my qt6 application. Allowing the user to plot and replot for some reason keeps increasing the sizes of my chart elements, but only when I have display scaling on (e.g. 125%, 150% etc). To illustrate:

See below for the code, basically I am creating a new plot every time there is a mouse click on that row. Embedding print(self.chart.figure.dpi) shows that my plot's dpi keeps doubling every other time I trigger a click. Console output:

100.0
100.0
200.0
200.0
400.0
400.0

I can workaround this by explicitly setting the figure DPI, e.g. figure.dpi = 100 when creating it. But I assume that means I'm giving up display scaling

Code for reproduction

import matplotlib

matplotlib.use("Qt5Agg")
from matplotlib.backends.backend_qt5agg import FigureCanvasQTAgg
from matplotlib.figure import Figure
from PyQt6.QtWidgets import (
    QWidget,
    QVBoxLayout, QFrame
)

class GraphView(QWidget):
    def __init__(self):
        super().__init__()
        self.layout = QVBoxLayout()
        self.separator = None
        self.chart = MplCanvas()

        self.figures = None

        self.init_widget()

    def init_widget(self):
        self.init_separator()
        self.layout.addWidget(self.separator)
        self.layout.addWidget(self.chart)
        self.setLayout(self.layout)
    ...
    # Note this is the code that triggers when I click a row.
    def update_graph(self, figure):

        self.chart.deleteLater()
        self.chart = MplCanvas(fig=figure)
    
        self.layout.addWidget(self.chart)
        self.layout.update()

class MplCanvas(FigureCanvasQTAgg):
    def __init__(self, fig=Figure()):
        super(MplCanvas, self).__init__(fig)

Actual outcome

Figure DPI keeps dynamically doubling

Expected outcome

DPI shouldn't double like that, it's madness ☹️

Additional information

No response

Operating system

Windows, Linux

Matplotlib Version

3.6.3

Matplotlib Backend

QtAgg

Python version

3.11.4

Jupyter version

No response

Installation

pip