The line-ending is hard-coded into the TextOutputFormat. You will have to either extend this class and override #writeRecord(), or convert your POJOs to a Tuple and use the CsvOutputFormat, which supports setting the line delimiter (called recordDelimiter).

On 09/05/2019 08:32, Papadopoulos, Konstantinos wrote:

Kind reminder

*From:*Papadopoulos, Konstantinos <konstantinos.papadopou...@iriworldwide.com>
*Sent:* Monday, May 06, 2019 5:31 PM
*To:* user@flink.apache.org
*Subject:* writeAsFormattedText sets only Unix/Linux line endings

Hi all,

We are developing an application using Flink DataSet API focusing on generating a CSV file from a dataset of POJOs using writeAsFormattedText and a custom TextFormatter.

During the testing of our application, we observed that the files generated consist of Unix line endings (i.e., ā€˜\nā€™) even if the execution environment is Windows (i.e., expected line endings ā€˜\r\nā€™).

Since this is critical to our solution, is there any way to configure the line endings of the files generated?

Thanks in advance,

Konstantinos


Reply via email to