Crud api to store data in memory or database or call an api to store the data

[bhumirjhaveri]

What is the question:
I want to use serverless workflow which is a crud api which will either store data in memory or db or call an another api, how can I do that with sdk-java version 7.0.0 or serverless specs 1?

[bhumirjhaveri]

We need more examples with 7.0.0 as action, state, functionref etc have been taken of and how to execute workflow correctly?

[fjtirado]

I feel the issue you wrote is more related with the specification than with the java SDK. New DSL extension mechanism to invoke custom APIs is described here

Anyway, since you asked for more examples of how to do things with the new DSL , please take a look to these ones https://github.com/serverlessworkflow/sdk-java/tree/main/impl/core/src/test/resources and https://github.com/serverlessworkflow/sdk-java/tree/main/impl/http/src/test/resources

[bhumirjhaveri]

Thanks for responding, so idea here is to specify the whole workflow in the specs yaml and from java/language, just start the workflow?

[fjtirado]

There are different things:

• The Serverless Workflow 1.0 spec, that specifies how to write workflows in yaml or json.
• The Serverless Worfklow 1.0 Java SDK, that allows you to load an in memory representation of that yaml/json
• The Serverless Workflow 1.0 Java reference implementation, that use the Java SDK to execute sample workflows written in yaml/json as per specification.

Your original question pertains to the specification domain: "how do I write a workflow that call a CURD API?". Since this functionality is not natively supported by the spec, I referred you to the extension mechanism (which is currently not supported by the reference implementation)

[bhumirjhaveri]

Thanks!
I am looking for 2nd option to use java-sdk. Load the yaml, have CRUD api and based on api call, perform respective workflow from the yaml.

[bhumirjhaveri]

Here is what I am thinking

document:
dsl: '1.0.0'
namespace: test
name: switch-example
version: '0.1.0'
use:
functions:
saveToolToDatabase:
input:
schema:
document:
type: object
properties:
toolId:
type: string
toolName:
type: string
toolDescription:
type: string
required: [toolId, toolName]
call: internal
with:
function: storeToolData

do:

• handle-request:
  switch:
  - create-tool:
  when: .callType == "create-tool"
  then: saveTool
  - update-tool:
  when: .callType == "update-tool"
  then: saveTool
  - default:
  then: handleUnknownOrderType
• saveTool:
  call: saveToolToDatabase
  with:
  toolId: "tool-1"
  toolName: "sample tool"
  toolDescription: "sample desc"
  then: end
• end: # End of workflow

saveToolToDatabase should be some java function where I actually write it to underlying database.

Let me know if this is currently supported or not. If not then is there a way out to implement this??

[fjtirado]

@bhumirjhaveri Can you ask the question in the specification repo?

[fjtirado]

@ricardozanini More than not planeed, I would say that language specific extensions should be discussed on the specification forum, not in the sdk one. Currently, there is not place in the specification for a Java function.

[ricardozanini]

It's not planned to add support for it here directly, which is why I closed the issue. I agree that @bhumirjhaveri can bring it to the spec.