[
https://issues.apache.org/jira/browse/FLINK-20340?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jingsong Lee updated FLINK-20340:
---------------------------------
Description:
We've thought about using StreamFormat, but we don't have splittable
StreamFormat implementation at present. In order to avoid functional
regression, still write DelimitedInputFormat.
> From FLINK-20295: We can also use Java's
> {{BufferedReader(InputStreamReader())}} to parse the lines. That is a bit
> less performant than out own fast parsing DelimitedInputFormat, but it
> supports different charset encodings properly. Currently, the
> DelimitedInputFormat fails on UTF-16 and some other charsets.
was:We've thought about using StreamFormat, but we don't have splittable
StreamFormat implementation at present. In order to avoid functional
regression, still write DelimitedInputFormat.
> Use StreamFormat instead of DelimitedInputFormat in
> DeserializationSchemaAdapter
> --------------------------------------------------------------------------------
>
> Key: FLINK-20340
> URL: https://issues.apache.org/jira/browse/FLINK-20340
> Project: Flink
> Issue Type: Improvement
> Components: Connectors / FileSystem
> Affects Versions: 1.12.0
> Reporter: Jingsong Lee
> Priority: Major
>
> We've thought about using StreamFormat, but we don't have splittable
> StreamFormat implementation at present. In order to avoid functional
> regression, still write DelimitedInputFormat.
> > From FLINK-20295: We can also use Java's
> > {{BufferedReader(InputStreamReader())}} to parse the lines. That is a bit
> > less performant than out own fast parsing DelimitedInputFormat, but it
> > supports different charset encodings properly. Currently, the
> > DelimitedInputFormat fails on UTF-16 and some other charsets.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
