To test seaborn, run `make test` in the root directory of the source distribution. This runs the unit test suite (using `pytest`, but many older tests use `nose` asserts). It also runs the example code in function docstrings to smoke-test a broader and more realistic range of example usage.
The full set of tests requires an internet connection to download the example datasets (if they haven’t been previously cached), but the unit tests should be possible to run offline.
完整的测试集需要 Internet 连接才能下载示例数据集(如果以前没有缓存过),但单元测试应该可以离线运行。
## Bugs
## 错误
Please report any bugs you encounter through the github [issue tracker](https://github.com/mwaskom/seaborn/issues/new). It will be most helpful to include a reproducible example on one of the example datasets (accessed through [`load_dataset()`](generated/seaborn.load_dataset.html#seaborn.load_dataset"seaborn.load_dataset")). It is difficult debug any issues without knowing the versions of seaborn and matplotlib you are using, as well as what [matplotlib backend](https://matplotlib.org/faq/usage_faq.html#what-is-a-backend) you are using to draw the plots, so please include those in your bug report.