Users can execute a workflow with the command:
This will load the current project and work out where my-workflow is from there.
When we do this, by default the output path will be my-workflow/output.json . Which is odd because the workflow lives in workflows/my-workflow. It means a messy new diff has been created and its not a nice experience.
It would be nicer to default the output to tmp/output.json, relative to the workflow.yaml file.
What about input paths? If I do -s input.hson should that be relattve to workflow/tmp? cwd? Perhaps if you do ./state.json we'll do it relative to the workspace root, but if you just do state.json we load it from /tmp. Then again its not hard to do -s tmp/input/json.
So what if we say that all input and output paths for a workflow are relative to the workflow yaml? If no path is included we do tmp/state.
Users can execute a workflow with the command:
This will load the current project and work out where my-workflow is from there.
When we do this, by default the output path will be
my-workflow/output.json. Which is odd because the workflow lives inworkflows/my-workflow. It means a messy new diff has been created and its not a nice experience.It would be nicer to default the output to
tmp/output.json, relative to the workflow.yaml file.What about input paths? If I do
-s input.hsonshould that be relattve to workflow/tmp? cwd? Perhaps if you do./state.jsonwe'll do it relative to the workspace root, but if you just dostate.jsonwe load it from /tmp. Then again its not hard to do-s tmp/input/json.So what if we say that all input and output paths for a workflow are relative to the workflow yaml? If no path is included we do tmp/state.