我有一个github存储库,我在其中开发了electrical calculation software。
最近我完全从PyQt5迁移到PySide2。
今天,我添加了Travis-CI以进行持续集成,以此作为对Githb存储库的一个钩子。这意味着当我进行一些更改时,travis-CI会在其云中的独立计算机中启动我的存储库的构建。
Travis CI由于以下原因而失败:
ImportError while importing test module '/home/travis/build/SanPen/GridCal/src/tests/test_branch_tolerance.py'. Hint: make sure your test modules/packages have valid Python names. Traceback: src/tests/test_branch_tolerance.py:1: in <module>
from GridCal.Engine import * src/GridCal/Engine/__init__.py:17: in <module>
from GridCal.Engine.basic_structures import * src/GridCal/Engine/basic_structures.py:23: in <module>
from GridCal.Engine.plot_config import LINEWIDTH, plt src/GridCal/Engine/plot_config.py:18: in <module>
matplotlib.use('Qt5Agg') ../../../virtualenv/python3.6.7/lib/python3.6/site-packages/matplotlib/cbook/deprecation.py:307: in wrapper
return func(*args, **kwargs) ../../../virtualenv/python3.6.7/lib/python3.6/site-packages/matplotlib/__init__.py:1297: in use
switch_backend(name) ../../../virtualenv/python3.6.7/lib/python3.6/site-packages/matplotlib/pyplot.py:230: in switch_backend
newbackend, required_framework, current_framework))
E ImportError: Cannot load backend 'Qt5Agg' which requires the 'qt5' interactive framework, as 'headless' is currently running
失败的文件plot_config.py
开头具有以下matplotlib配置:
import PySide2 # this line is necessary so that Matplotlib recognises that PySide is the Qt Backend
import matplotlib
matplotlib.use('Qt5Agg')
# matplotlib.rcParams['backend.qt5'] = 'PySide2' # this is not supported anymore
from matplotlib import pyplot as plt
这段代码是我在多个线程上发现的正确做法。但是,由于它们,在独立计算机上进行的自动化测试失败。
那么,告诉matplotlib使用PySide2的最佳方式是什么?
所需的软件包为(requirements.txt):
PySide2>=5.11
numpy>=1.14.0
scipy>=1.0.0
networkx>=2.1
pandas>=0.22
xlwt>=1.3.0
xlrd>=1.1.0
matplotlib>=3.1.0
qtconsole>=4.3.1
pyDOE>=0.3.8
pySOT>=0.2.1
openpyxl>=2.4.9
pulp>=1.6.8
smopy>=0.0.6
chardet>=3.0.4
scikit-learn>=0.18
geopy>=1.16
pytest>=3.8
h5py>=2.9.0