Conversation
| - name: routput | ||
| path: /tmp/routput | ||
| s3: | ||
| key: /plot_gen.tgz |
There was a problem hiding this comment.
What would be a sensible way to generalise this kind of thing?
It would be good to not put the responsibility on the users to makes/find writable directories.
Could we provide templates that ensure the job runs in a tmp dir and has loops like
outputs:
artifacts:
{{% for output in outputs %}}
- name: {{ output.name }}
path: /tmp/{{job_id}}/{{ output.filename }}
s3:
key : {{ output.filename }}
{% endfor %}There was a problem hiding this comment.
Are you wanting these to also be user-facing examples? Is this thinking ahead to what we've discussed about potentially locking down templates? I didn't realise these were really intended to exemplify that.
This was only intended to exemplify getting an artifact in and putting one out, not generalise to many output artifacts. This script just tars an entire directory and puts it in a single file in the default S3 bucket, so there is only one output and it doesn't need a loop.
Closes #45