如何在pyqtgraph remoteview上绘制切片的numpy数据数组

Question

I am having trouble with the RemoteGraphicsView() functionallity in PyQTGraph. I have a numpy ndarray which I wish to plot on a RemoteGraphicsView (for speed as it runs in a seperate process). I want to plot a slice of the data but it fails with a TypeError

TypeError: must be string or single-segment read-only buffer, not numpy.ndarray

The following is a code snippet which demonstrates the problem I have in a larger program. As can be seen the data is a numpy.ndarray which is sliced.

import pyqtgraph as pg
import numpy as np
import pyqtgraph.widgets.RemoteGraphicsView

app = pg.mkQApp()

datview = pg.widgets.RemoteGraphicsView.RemoteGraphicsView(debug=False)
datview.pg.setConfigOptions(antialias=True) 
allplots = datview.pg.GraphicsLayout()
datview.setCentralItem(allplots)

w1 = allplots.addPlot(row=0,col=0)

layoutWG = pg.LayoutWidget()
layoutWG.addWidget(datview)
layoutWG.show()

data = np.random.randn(10000,100)
curve1 = w1.plot(pen='r')

now = pg.ptime.time()
for n in range(100):
    curve1.setData(data[:,n])
    app.processEvents()

app.exec_()

Any help would be much appreciated.

Answer 1

Looks like the interprocess communication system requires a contiguous array. This should work:

    curve1.setData(np.ascontiguousarray(data[:,n]))

Alternatively, you can define the data such that the slices you want are already contiguous:

data = np.random.randn(100,10000)
...
for n in range(100):
    curve1.setData(data[n])

I also recommend a couple of changes to speed things up:

# Prevent lookups to curve1.setData from immediately requesting and returning
# the method proxy every time it is called
curve1._setProxyOptions(deferGetattr=True)

for n in range(100):
    # Use _callSync='off' to prevent the main process waiting for a return 
    # value
    curve1.setData(np.ascontiguousarray(data[:,n]), _callSync='off')