Christoph Deppisch created CAMEL-18698:
------------------------------------------
Summary: Add support for multiple data types on components
Key: CAMEL-18698
URL: https://issues.apache.org/jira/browse/CAMEL-18698
Project: Camel
Issue Type: New Feature
Components: camel-core
Reporter: Christoph Deppisch
Each component consumes/produces input and output data when processing the
exchange as part of the route. Usually a component reads data from the exchange
message (body and header) as input and produces specific message content as an
output.
The input and output of a component can be of various nature (e.g. Java POJO
objects, Json objects, String, binary data, InputStream, etc.)
In order to simplify the interrelationship between components each component
should allow well defined data types as input and output and provide basic
transformation logic.
Having the possibility to choose, a user can specify the *format* option on an
endpoint URI in order to specify the requested data type. Each component
defines a set of supported data types and provides logic to convert to and from
this data type.
{{from("kafka:topic?format=avro").to("aws2-s3:bucketName?format=string")}}
The example above lets the user choose one of the supported input/output data
formats on the components kafka and aws2-s3. The components may offer specific
logic to transform from/to the defined data type. Also there will be generic
data type transformation logic that Camel already provides when converting
types via TypeConverter.
The idea is to provide multiple supported data types for components and let the
component automatically apply transformation logic to/from the data type that
the user has chosen.
On the one hand there is the transformation EIP and Camel provides powerful
route DSL to perform transformations from one data type to another. But in many
cases the component itself knows best how to transform from/to the domain model
data type.
This is especially useful when dealing with components that use Java POJO
objects as input/output. This way the user is able to deal with more common
data types (e.g. String, Json) instead of having to know the domain model of
the individual component.
In particular this would be beneficial for Camel K and Kamelets (also see the
discussions in https://github.com/apache/camel-k/issues/1980) where declarative
YAML DSL is used to specify the route logic. In these declarative use cases the
usage of custom Java processors to transform from one data type to another is
not easily possible. Instead the components itself should provide predefined
data type support and offer multiple types that the user may choose from.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
