HuangXingBo commented on a change in pull request #14741:
URL: https://github.com/apache/flink/pull/14741#discussion_r586340408
##########
File path: flink-python/setup.py
##########
@@ -321,12 +321,11 @@ def find_file_path(pattern):
license='https://www.apache.org/licenses/LICENSE-2.0',
author='Apache Software Foundation',
author_email='[email protected]',
- python_requires='>=3.5',
- install_requires=['py4j==0.10.8.1', 'python-dateutil==2.8.0',
'apache-beam==2.23.0',
+ python_requires='>=3.6',
+ install_requires=['py4j==0.10.8.1', 'python-dateutil==2.8.0',
'apache-beam==2.27.0',
'cloudpickle==1.2.2', 'avro-python3>=1.8.1,<=1.9.1',
'jsonpickle==1.2',
- 'pandas>=0.24.2,<1; python_full_version < "3.5.3"',
- 'pandas>=0.25.2,<1; python_full_version >= "3.5.3"',
- 'pyarrow>=0.15.1,<0.18.0', 'pytz>=2018.3',
'numpy>=1.14.3,<1.20'],
+ 'pandas>=0.25.2,<1', 'pyarrow>=0.15.1,<0.18.0',
'pytz>=2018.3',
+ 'numpy>=1.14.3,<1.20', 'fastavro>=0.21.4,<0.24'],
Review comment:
`fastavro` hasn't no longer provided manylinux2010.whl since `0.24.2`,
which makes some old version machines must have gcc if using newer version of
`fastavro`. In fact, the `fastavro` dependency comes from `apache-beam`, and it
is used in connectors. We can limit it to a relatively old version to solve the
problem of `gcc` absence, and it will not have any impact on Pyflink. What do
you think?
The 2.0.0 version of `pyarrow` is incompatible with the 1.20.x version of
`numpy`. This problem was resolved in the unreleased 2.0.1 version of
`pyarrow`. So here we limit the version of `numpy` to less than 1.20.
https://issues.apache.org/jira/browse/ARROW-10855?page=com.atlassian.jira.plugin.system.issuetabpanels%3Aall-tabpanel
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
