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Information in this section is out of date!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 entry point of the Taverna Workflow Bundle.

Defines the workflows and profiles of a Taverna Workflow Bundle. The main workflow is also normally defined, which would be the top-level workflow to execute. The profiles defines how these workflows can be realised and executed on different environments, one of which can be suggested as the main profile.

Bundle path and root files

The workflow bundle document in RDF/XML format should be in in /workflowBundle.rdf within the bundle archive.

If the archive is a workflow bundle, i.e. /mimetype is application/vnd.taverna.scufl2.workflow-bundle, then the META-INF/container.xml can define root files at alternative paths and media types. This specification requires that one of those formats is application/rdf+xml according to this specification.

Example META-INF/container.xml:

This defines the RDF/XML root file to be /workflowBundle.rdf - with workflowBundle.ttl being an alternate representation the resource in Turtle format.

SCUFL2-compliant workflow bundle writers:

  • Must set the bundle mimetype to application/vnd.taverna.scufl2.workflow-bundle
  • Must add a workflow bundle document in application/rdf+xml format
  • Should store the workflow bundle document in /workflowBundle.rdf
  • Must not contain a resource /workflowBundle.rdf which is not the workflow bundle document
  • If the application/rdf+xml representation is not in /workflowBundle.rdf, the writer must include META-INF/container.xml with the required <rootfile> entries.
  • META-INF/container.xml, if present, must contain one and only one rootfile with the media-type application/rdf+xml. rootfiles of other media-types may be included, but their formats and restrictions are not defined by this specification.
  • May Add additional representation of the workflow bundle document (and other documents). Alternates of the workflow bundle document should be included in the META-INF/container.xml, but only if they can be considered to fully specify the workflow bundle as in the RDF/XML format. (So for instance a text/html or image/png representation would not normally be considered a rootfile if it does not include all the structural information from the RDF/XML representation as specified here)

It is possible to include a workflow bundle document within a different kind of archive or bundle, for instance in a data bundle. In this case the bundle is not considered an application/vnd.taverna.scufl2.workflow-bundle - but producers of such archives:

  • Should store the workflow bundle document in /workflowBundle.rdf, unless the workflow bundle is not to be considered to have a 'main' or 'prominent' role within the archive. (For instance if the archive is a collection of workflow bundles).
  • Should have a mimetype and META-INF/container.xml resource which declares the archive's main entry point, like the data bundle document. The mime type must not be application/vnd.taverna.scufl2.workflow-bundle and the root files should not be the workflow bundle document.
  • Should link to the workflow bundle document from a resource within the archive which (ultimately) is linked to from one of the rootfile documents. Such documents should be in RDF/XML format.
  • Should declare the media type of the RDF/XML workflow bundle document as application/rdf+xml in its META-INF/manifest.xml

SCUFL2 compliant workflow bundle readers:

  • Should assume that /workflowBundle.rdf - if present - is the root workflow bundle in the application/rdf+xml format specified here.
  • Should assume that if the archive's mimetype is application/vnd.taverna.scufl2.workflow-bundle, then the rootfile in META-INF/container.xml with the media type {{application/rdf+xml}) is the root workflow bundle document.
  • May assume that any alternate formats listed as a rootfile in a application/vnd.taverna.scufl2.workflow-bundle archive would fully cover the specification of the RDF/XML representation, and read such formats instead.
  • May assume that any application/rdf+xml document with a xsi:type="WorkflowBundleDocument" can be parsed according to the Scufl2 XML schema

Identifiers

Workflow bundles and their resources must be declared with relative identifiers within the archive. In a application/vnd.taverna.scufl2.workflow-bundle archive, the workflow bundle must be identified as the root of the archive. If the Workflow Bundle document is in workflowBundle.rdf within the archive, the workflow identifier is ./. This should be achieved by setting xml:base="./" and rdf:about="".

This means that one can mint a URI to refer to resources within the bundle archive, including the workflow bundle, workflows and representations. If
http://example.com/myWfBundle.scufl2 returns a Scufl2 workflow bundle archive of the content type application/vnd.taverna.scufl2.workflow-bundle, then (assuming default structure of the archive):

Embedded workflow bundles

If the archive is another type of bundle which includes the workflow bundle (but is not primarily playing the role as the format for this workflow bundle), the local workflow identifier should be unique within the archive. This is easiest achieved by using the same folder technique as for the workflow representations:

  • workflowBundle1.rdf can define workflowBundle1/
  • exampleWorkflowBundles/hello.rdf defines exampleWorkflowBundles/hello/

Such embedded workflow bundles should include their constituent representations (such as workflow/HelloWorld.rdf) within that folder, for instance exampleWorkflowBundles/hello/workflow/HelloWorld.rdf to define exampleWorkflowBundles/hello/workflow/HelloWorld/ - but could also be shared among bundles, for instance both workflowBundle1.rdf and workflowBundle2.rdf might refer to workflow/Shared.rdf.

Global workflow bundle identifiers

Workflow bundles should declare a sameBaseAs reference to a globally unique non-informational URI.

Anyone can generate such a URI using the form http://ns.taverna.org.uk/2010/workflowBundle/UUID/ - for instance http://ns.taverna.org.uk/2010/workflowBundle/28f7c554-4f35-401f-b34b-516e9a0ef731/ - including the trailing slash.

The semantics of sameBaseAs is a kind of recursive version of owl:sameAs - so also resources which URI start with the same will be included.

So if:

<./> = <http://ns.taverna.org.uk/2010/workflowBundle/28f7c554-4f35-401f-b34b-516e9a0ef731/>
<workflow/HelloWorld/ = <http://ns.taverna.org.uk/2010/workflowBundle/28f7c554-4f35-401f-b34b-516e9a0ef731/workflow/HelloWorld/>
<workflow/HelloWorld.rdf> = <http://ns.taverna.org.uk/2010/workflowBundle/28f7c554-4f35-401f-b34b-516e9a0ef731/workflow/HelloWorld.rdf>
{code]

This allows anyone to make a statement about any resource within the workflow bundle, even if the URL of the workflow bundle representation itself might change, be it on a local USB disk, DropBox folder, myExperiment, etc.

Updating the UUID

It is up to the software editing or creating the workflow to assign a new UUID as soon as any change has been done to any workflow, profile or workflow bundle, as this is the globally unique identifier for this workflow archive, and also the base URI for all the other resources in the archive.

Not implemented by API

Scufl2 API do not yet automatically update the workflow bundle identifier.SCUFL2-41

To update the URI, use workflowBundle.setSameBaseAs(WorkflowBundle.generateIdentifier()).

Example representation in RDF/XML

This example defines the workflow bundle "HelloWorld". It contains one workflow workflow/HelloWorld, which is also the main workflow. Any additional workflows are typically nested (and nested-nested, etc) workflows bound as activities in processors). Two execution profiles are provided, and profile/tavernaWorkbench is dedicated as the main profile.

Properties

  • name (required) gives the human readable title for this workflow archive. This is a subproperty of dc:title.
  • sameBaseAs (optional) gives a unique URI which is owl:sameAs with this workflow bundle and its children.
  • workflow (required) All workflows included in this bundle. Each workflow must have an rdfs:seeAlso link to the bundle resource that defines the workflow, typically workflow/workflowName.rdf corresponding to the non-information resource workflow/workflowName/.
  • mainWorkflow (optional) The reference to the top-level workflow of this bundle. It is valid to have a workflow bundle without a main workflow, for instance if the bundled workflows are unconnected "workflow fragments". If there is a mainProfile the workflow bundle must also have a mainWorkflow. The main workflow must always be listed under workflow.
  • profile (optional) profiles specifying how to execute the bundled workflows. In particular the profile provides a set of configured activities bound to the processors for a particular run environment. If no profiles are specified this is an abstract workflow bundle.
  • mainProfile (optional) the suggested main profile. Execution platforms unable to choose between the provided profiles can select this profile as a default. It is valid to have a workflow bundle without a main profile (even if it has other profiles), but any main profile must be listed under profile.
  • rdfs:seeAlso (optional) link to annotations about the workflow bundle and its content. Traditionally found in annotation/workflowBundle.rdf, which should contain further links to annotations from different sources, for instance annotation/myExperiment.rdf for annotations included from myExperiment.

The workflow bundle document is the starting point for finding all workflow bundle resources within the archive. Each of the workflows and profiles must therefore have a rdfs:seeAlso link to the bundle resource that defines it.

If alternate formats other than the required RDF/XML format is included in the bundle, these formats can therefore link to resources in other formats, for instance in an additional workflowBundle.ttl (Turtle format):

Parsing/writing

SCUFL2 compliant writers, when producing the workflow bundle document:

  • Should write the workflow bundle RDF/XML document according to the [SCUFL2 XML schema], use the default namespace xmlns="http://ns.taverna.org.uk/2010/scufl2#" and declare the xsi:type="WorkflowBundleDocument"
  • Must ensure the workflow bundle RDF/XML document is valid RDF/XML and includes the properties deemed required by this specification. Conforming to the XML schema should ensure this.
  • Should set the xml:base property to "./
  • Should set rdf:about to "" (or "./" if xml:base is not set))
  • Should declare a mainWorkflow and mainProfile
  • Must ensure that any workflow/profile has a relative rdfs:seeAlso link to a bundle resource in application/rdf+xml which defines
    that workflow/profile.

SCUFL2 compliant readers, when parsing a workflow bundle document:

  • Mayassume that a declared workflow/profile is defined in the referenced representation. For instance, if:

    then workflow/SomeNestedWorkflow.rdf must contain a workflow definition for workflow/SomeNestedWorkflow/.

  • May parse the /workflowBundle.rdf as RDF/XML
  • May parse the {/workflowBundle.rdf}} according to the XML schema if the xsi:type="WorkflowBundleDocument" is set on the rdf:RDF element.
Labels
  • None