[ https://issues.apache.org/jira/browse/CAMEL-13026?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16726832#comment-16726832 ]
Christian Ribeaud commented on CAMEL-13026: ------------------------------------------- What you are describing is a workaround. I should basically have two {{CSVFormats}}, one for _empty_ file and one for _non-empty_ one. When writing out a _huge_ *CSV* file, you do not want to keep the content in the memory. Instead, you prefer to append to an existing file. And because it is possible to specify a header in {{CSVFormat}}, I would expect {{CsvMarshaller}} being smart enough to be able to handle such case. But, I see your point, maybe one DOES want to write out a new file for each message. And they all should contain the header. However, instead of making my routing more complicate, how about being able to: # Extend {{CsvMarshaller}} to customize it to my needs? (currently not possible) # Specify a message property to tackle my use case? > Header written out on each message when marshalling > --------------------------------------------------- > > Key: CAMEL-13026 > URL: https://issues.apache.org/jira/browse/CAMEL-13026 > Project: Camel > Issue Type: Improvement > Components: camel-csv > Reporter: Christian Ribeaud > Priority: Major > > When *CSV* marshalling, the header is written out on each message. > Additionally, {{CsvMarshaller}} is so opaque, making it difficult for > customization. -- This message was sent by Atlassian JIRA (v7.6.3#76005)