[
https://issues.apache.org/jira/browse/SYSTEMML-1658?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Mike Dusenberry updated SYSTEMML-1658:
--------------------------------------
Description:
Currently, a DML print statement results in a System.out.println call in the
JVM. For Python, this output will not be routed back as Python standard out,
and thus it will not be visible in some environments, such as a Jupyter
notebook. In a Python repl, the output will be visible, but only because the
JVM will be printing to the same terminal as the Python process is running
from, and thus the user will see the output. Since a large amount of ML work
is performed within a notebook using Python, we should figure out how to route
this output back to the Python std out.
Relevant:
* https://www.py4j.org/py4j_java_gateway.html#py4j.java_gateway.launch_gateway
*
https://github.com/apache/spark/blob/a36a76ac43c36a3b897a748bd9f138b629dbc684/python/pyspark/java_gateway.py#L36
*
https://github.com/bartdag/py4j/blob/262a20477a8a302cae0ede597c93b953a3fc2702/py4j-python/src/py4j/java_gateway.py#L320
*
https://github.com/bartdag/py4j/blob/262a20477a8a302cae0ede597c93b953a3fc2702/py4j-python/src/py4j/java_gateway.py#L596
was:Currently, a DML print statement results in a System.out.println call in
the JVM. For Python, this output will not be routed back as Python standard
out, and thus it will not be visible in some environments, such as a Jupyter
notebook. In a Python repl, the output will be visible, but only because the
JVM will be printing to the same terminal as the Python process is running
from, and thus the user will see the output. Since a large amount of ML work
is performed within a notebook using Python, we should figure out how to route
this output back to the Python std out.
> Route DML print statements from JVM back to Python
> --------------------------------------------------
>
> Key: SYSTEMML-1658
> URL: https://issues.apache.org/jira/browse/SYSTEMML-1658
> Project: SystemML
> Issue Type: Improvement
> Components: APIs
> Reporter: Mike Dusenberry
>
> Currently, a DML print statement results in a System.out.println call in the
> JVM. For Python, this output will not be routed back as Python standard out,
> and thus it will not be visible in some environments, such as a Jupyter
> notebook. In a Python repl, the output will be visible, but only because the
> JVM will be printing to the same terminal as the Python process is running
> from, and thus the user will see the output. Since a large amount of ML work
> is performed within a notebook using Python, we should figure out how to
> route this output back to the Python std out.
> Relevant:
> * https://www.py4j.org/py4j_java_gateway.html#py4j.java_gateway.launch_gateway
> *
> https://github.com/apache/spark/blob/a36a76ac43c36a3b897a748bd9f138b629dbc684/python/pyspark/java_gateway.py#L36
> *
> https://github.com/bartdag/py4j/blob/262a20477a8a302cae0ede597c93b953a3fc2702/py4j-python/src/py4j/java_gateway.py#L320
> *
> https://github.com/bartdag/py4j/blob/262a20477a8a302cae0ede597c93b953a3fc2702/py4j-python/src/py4j/java_gateway.py#L596
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)