Skip to end of metadata
Go to start of metadata

Scufl2 has moved to Apache (incubator) 

Information in this section is out of date!

 

The definition of a data link.

  • Bundle path: /workflow/\$workflowName.n3

Datalink should be described in the same file as their containing workflow.

Each datalink is uniquely identified by their source and destination ports, in addition to the optional merge position.

Datalinks are globally identified as <http://ns.taverna.org.uk/2010/researchObject/$researchObjectUUID/workflow/$workflow/datalink?from=$fromPort&to=$toPort&mergePosition=$mergePosition> - for instance <http://ns.taverna.org.uk/2010/researchObject/28f7c554-4f35-401f-b34b-516e9a0ef731/workflow/HelloWorld/datalink?from=processor/A/out/result&to=processor/B/in/db> defines the link in workflow "HelloWorld" from the output port "result" in the processor "A" going to the input port "db" for the processor "B". As there is no mergePosition there can't be any other links going to the "db" port.

As these URIs can be tricky to construct or maintain, feel free to use anonymous nodes, or construct alterative URIs as <datalink/$number>, the number here would not have any semantic meaning except it must be unique per datalink in that workflow.

workflow/Helloworld.n3:

This example defines three datalinks.

The first link _:datalink1 is just an anonymous node without an identifier. It defines a data link from the input port "yourName" to the processor input port "name". This link could also have been written embedded with the workflow:

The second datalink <datalink/5>> defines a link directly from the workflow input port "yourName" to the output port "results". Links must go from a scufl2:SendingPort sending to a scufl2:ReceivingPort, meaning from a workflow input port or processor output port, going to either a workflow output port or processor input port. Several links can receive from the same sending port.

Merges

Merges is a way Taverna allows you to connect several links to the same scufl2:ReceivingPort, that is to a processor input port or workflow output port. When executing, values from each link will be inserted into the specified scufl2:mergePosition in a new list.

You only need to specify scufl2:mergePosition if more than one link is connected to the same processor input port or workflow output port. If there is more than one link connected to the same receiving port, all of them need to specify a unique mergePosition, starting from 0, with no gaps. If you specify a single link to the port with a mergePosition of 0, that port input would still be wrapped in a list.

The third datalink, specified using the 'full' URI <datalink?from=processor/Hello/out/greeting&to=out/results&mergePosition=0>, defines the link from the output port "out" in processor "Hello", linking to the workflow output port "results". As now two links go to that receiving port, they both need to specify a unique mergePosition. The second link specifies position 1, and the third position 0. That means that the output port will receive a list Scufl2-DataLink. The second element 'yourName' would arrive first (as it is sent before "Hello" produces "greeting"), but it would be arriving in position 1 rather than 0.

URI templates not enough

The full URIs such as <http://ns.taverna.org.uk/2010/researchObject/28f7c554-4f35-401f-b34b-516e9a0ef731/workflow/HelloWorld/datalink?from=processor/A/out/result&to=processor/B/in/db> are meant to be helpful, not defining. The workflow definitions should be complete without having to be parse these URIs. Such URIs are however useful to be able to annotate or talk about workflow elements outside of the workflow definition.

Although a resource is specified using a full URI which uniquely identifies it, such as in the datalink above, the resource must still be defined, such as the datalink must define the properties scufl2:receivesFrom, scufl2:sendsTo (and optionally) scufl2:mergePosition. Similarly the input port <in/yourName> must still be defined with a scufl2:name "yourName".

The nested resources for this workflow, such as InputProcessorPort , OutputProcessorPort, DispatchStack, IterationStrategyStack and their children should be described in the same file as owning workflow itself. Additional metadata should be added to an /annotations/ file.

  • scufl2:receivesFrom (required) The scufl2:SendingPort this link is receiving data from. The port must be in the same workflow as the link.
  • scufl2:sendsTo (required) The scufl2:ReceivingPort this link is sending data to. The port must be in the same workflow as the link.
  • scufl2:mergePosition (optional) An integer, starting from 0. Must be set where more than one datalinks sendsTo the same ReceivingPort. The positions for a port must be sequentially assigned from 0 without gaps.
Labels
  • None