[
https://issues.apache.org/jira/browse/BEAM-8280?focusedWorklogId=411410&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-411410
]
ASF GitHub Bot logged work on BEAM-8280:
----------------------------------------
Author: ASF GitHub Bot
Created on: 27/Mar/20 21:51
Start Date: 27/Mar/20 21:51
Worklog Time Spent: 10m
Work Description: robertwb commented on pull request #11232: [BEAM-8280]
Document Python 3 annotations support
URL: https://github.com/apache/beam/pull/11232#discussion_r399553756
##########
File path: website/src/documentation/sdks/python-type-safety.md
##########
@@ -23,38 +23,58 @@ Python is a dynamically-typed language with no static type
checking. Because of
The Apache Beam SDK for Python uses **type hints** during pipeline
construction and runtime to try to emulate the correctness guarantees achieved
by true static typing. Additionally, using type hints lays some groundwork that
allows the backend service to perform efficient type deduction and registration
of `Coder` objects.
-Python version 3.5 introduces a module called **typing** to provide hints for
type validators in the language. The Beam SDK for Python, based on Python
version 2.7, implements a subset of [PEP
484](https://www.python.org/dev/peps/pep-0484/) and aims to follow it as
closely as possible in its own typehints module.
+Python version 3.5 introduces a module called **typing** to provide hints for
type validators in the language.
+The Beam SDK for Python implements a subset of [PEP
484](https://www.python.org/dev/peps/pep-0484/) and aims to follow it as
closely as possible in its own typehints module.
+
+These flags control Beam type safety:
+- `--no_pipeline_type_check`
+
+ Disables type checking during pipeline construction.
+ Default is to perform these checks.
+- `--runtime_type_check`
+
+ Enables runtime type checking of every element.
+ This may affect pipeline performance, so the default is to skip these checks.
## Benefits of Type Hints
-The Beam SDK for Python includes some automatic type checking: for example,
some `PTransform`s, such as `Create` and simple `ParDo` transforms, attempt to
deduce their output type given their input. However, the Beam cannot infer
types in all cases. Therefore, the recommendation is that you declare type
hints to aid you in performing your own type checks if necessary.
+The Beam SDK for Python includes some automatic type checking: for example,
some `PTransforms`, such as `Create` and simple `ParDo` transforms, attempt to
deduce their output type given their input. However, the Beam cannot infer
types in all cases. Therefore, the recommendation is that you declare type
hints to aid you in performing your own type checks if necessary.
-When you use type hints, the runner raises exceptions during pipeline
construction time, rather than runtime. For example, the runner generates an
exception if it detects that your pipeline applies mismatched `PTransforms`
(where the expected outputs of one transform do not match the expected inputs
of the following transform). These exceptions are raised at pipeline
construction time, regardless of where your pipeline will execute. Introducing
type hints for the `PTransform`s you define allows you to catch potential bugs
up front in the local runner, rather than after minutes of execution into a
deep, complex pipeline.
+When you use type hints, the runner raises exceptions during pipeline
construction time, rather than runtime. For example, the runner generates an
exception if it detects that your pipeline applies mismatched `PTransforms`
(where the expected outputs of one transform do not match the expected inputs
of the following transform). These exceptions are raised at pipeline
construction time, regardless of where your pipeline will execute. Introducing
type hints for the `PTransforms` you define allows you to catch potential bugs
up front in the local runner, rather than after minutes of execution into a
deep, complex pipeline.
Review comment:
I might lead with this paragraph, maybe even giving a very concrete example
("for example, giving an error when trying to apply a PTransform that expects a
PCollection of strings to a PCollection of ints"). Then put the other paragraph
about Beam not always being able to infer types next.
----------------------------------------------------------------
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]
Issue Time Tracking
-------------------
Worklog Id: (was: 411410)
Time Spent: 11h 40m (was: 11.5h)
> re-enable IOTypeHints.from_callable
> -----------------------------------
>
> Key: BEAM-8280
> URL: https://issues.apache.org/jira/browse/BEAM-8280
> Project: Beam
> Issue Type: Bug
> Components: sdk-py-core
> Reporter: Udi Meiri
> Assignee: Udi Meiri
> Priority: Major
> Time Spent: 11h 40m
> Remaining Estimate: 0h
>
> See https://issues.apache.org/jira/browse/BEAM-8279
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
