Create output file when generating resources with CLI

[adamrtalbot]

It would be useful to store a log of the generated resources when creating a platform item with seqerakit, so that we can re-use it later.

For an example, here I am adding a new compute environment and pipeline:

launch:
  - name: 'hello-world'
    workspace: 'myorg/myworkspace'
    pipeline: 'hello'

How do I fetch the results from this launch? If we generate an output structure similar to the input structure, we could do tw runs view on the output object, but currently seqerakit uses a fire-and-forget model.

Let's say we called the above yaml and added an output arg, we could store the outputs for future use.

seqerakit --output output.yml hello.yml

This may help with #83 as well as we could generate resources at creation time.

[ejseqera]

@adamrtalbot Is the output intended to be YAML or do you mean generate a log file that captures the stdout? Could you share a desired example?

seqerakit returns the CLI command invoked under the hood through the definitions provided in the YAML file.

[adamrtalbot]

A YAML, so we could re-use it as an input if we wanted to rerun the previous command e.g.

seqerakit hello-world.yml > hello-world-2.yml
hellow-world.yml == hello-world-2.yml

[adamrtalbot]

Some of this is covered with: #146 and #173 and #172 for the following fixes 🤦.

We can possibly coerce this yaml to json.

[ejseqera]

My only concern would be that I am not sure if all resources will allow you to generate an output that you can reuse to generate resources again, if I understand the intention correctly. Do you wanna work on the POC?

[adamrtalbot]

After 6 months, I can't actually picture how it would work now. I guess something like this:

• Create YAML with creds, CE, pipeline
• Run seqerakit

Then the decision:

• Output YAML that is the resolved union of the above?
• Output YAML that is the generic version, with specific features removed like credentials and workspace name?

Low priority right now imho.