The discussion of workflow provenance exports came up on the CWL gitter channel and I received this advice from @stain:
"I would be biased, but I would of course recommend you use wfprov and put it in say JSON-LD format together with a wfdesc description of the (possibly implied) workflow and the values as individual files, then package it all as a ResearchObject bundle ZIP file - that way you can have multiple options for how to format things, e.g. you can include also a Galaxy-native representations and have a semi-structured file layout that users might understand (e.g. reflecting their workspace)
what is fun with Galaxy is that users build workflows "by accident" - so you can choose to either just model that as individual Process invocations without any overall workflow, or to imply the workflow from the values.
but the second might mean you have to keep historical values that since disappeared from the workspace
(or at least its provenance)
https://github.com/common-workflow-language/common-workflow-language/issues/84#issuecomment-195718238 shows a good representation approach by @ntijanic - however for interoperability I would do it in pure JSON-LD rather than in Schema Salad."
The discussion of workflow provenance exports came up on the CWL gitter channel and I received this advice from @stain:
"I would be biased, but I would of course recommend you use wfprov and put it in say JSON-LD format together with a wfdesc description of the (possibly implied) workflow and the values as individual files, then package it all as a ResearchObject bundle ZIP file - that way you can have multiple options for how to format things, e.g. you can include also a Galaxy-native representations and have a semi-structured file layout that users might understand (e.g. reflecting their workspace) what is fun with Galaxy is that users build workflows "by accident" - so you can choose to either just model that as individual Process invocations without any overall workflow, or to imply the workflow from the values. but the second might mean you have to keep historical values that since disappeared from the workspace (or at least its provenance) https://github.com/common-workflow-language/common-workflow-language/issues/84#issuecomment-195718238 shows a good representation approach by @ntijanic - however for interoperability I would do it in pure JSON-LD rather than in Schema Salad."