[
https://issues.apache.org/jira/browse/BEAM-7198?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Valentyn Tymofieiev updated BEAM-7198:
--------------------------------------
Description:
The name of ToStringCoder class [1] is confusing, since the output of encode()
on Python3 will be bytes. On Python 2 the output is also bytes, since bytes and
string are synonyms on Py2.
ToBytesCoder would be a better name for this class.
Note that this class is not listed in coders that constitute Public APIs [2],
so we can treat this as internal change. As a courtesy to users who happened
to reference a non-public coder in their pipelines we can keep the old class
name as an alias, e.g. ToStringCoder = ToBytesCoder to avoid friction, but
clean up Beam codeabase to use the new name.
[1]
https://github.com/apache/beam/blob/ef4b2ef7e5fa2fb87e1491df82d2797947f51be9/sdks/python/apache_beam/coders/coders.py#L344
[2]
https://github.com/apache/beam/blob/ef4b2ef7e5fa2fb87e1491df82d2797947f51be9/sdks/python/apache_beam/coders/coders.py#L20
cc: [~yoshiki.obata] [~chamikara]
was:
The name of ToStringCoder class [1] is confusing, since the output of encode()
on Python3 will be bytes. On Python 2 the output is also bytes, since bytes and
string a synonyms on Py2.
ToBytesCoder would be a better name for this class.
Note that this class is not listed in coders that constitute Public APIs [2],
so we can treat this as internal change. As a courtesy to users who happened
to reference a non-public coder in their pipelines we can keep the old class
name as an alias, e.g. ToStringCoder = ToBytesCoder to avoid friction, but
clean up Beam codeabase to use the new name.
[1]
https://github.com/apache/beam/blob/ef4b2ef7e5fa2fb87e1491df82d2797947f51be9/sdks/python/apache_beam/coders/coders.py#L344
[2]
https://github.com/apache/beam/blob/ef4b2ef7e5fa2fb87e1491df82d2797947f51be9/sdks/python/apache_beam/coders/coders.py#L20
cc: [~yoshiki.obata] [~chamikara]
> Rename ToStringCoder into ToBytesCoder
> --------------------------------------
>
> Key: BEAM-7198
> URL: https://issues.apache.org/jira/browse/BEAM-7198
> Project: Beam
> Issue Type: Sub-task
> Components: sdk-py-core
> Reporter: Valentyn Tymofieiev
> Priority: Minor
>
> The name of ToStringCoder class [1] is confusing, since the output of
> encode() on Python3 will be bytes. On Python 2 the output is also bytes,
> since bytes and string are synonyms on Py2.
> ToBytesCoder would be a better name for this class.
> Note that this class is not listed in coders that constitute Public APIs [2],
> so we can treat this as internal change. As a courtesy to users who happened
> to reference a non-public coder in their pipelines we can keep the old class
> name as an alias, e.g. ToStringCoder = ToBytesCoder to avoid friction, but
> clean up Beam codeabase to use the new name.
> [1]
> https://github.com/apache/beam/blob/ef4b2ef7e5fa2fb87e1491df82d2797947f51be9/sdks/python/apache_beam/coders/coders.py#L344
> [2]
> https://github.com/apache/beam/blob/ef4b2ef7e5fa2fb87e1491df82d2797947f51be9/sdks/python/apache_beam/coders/coders.py#L20
> cc: [~yoshiki.obata] [~chamikara]
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
