Warnings vs Errors

[willthames]

Warnings vs Errors

[willthames]

Yes, that's a standard feature.

From the README.md (it could probably be clearer):

To add standards that are advisory, don't set the version. These will cause a message to be displayed but won't constitute a failure.

Hopefully that helps, let me know if you need any more info

[ansiblejunky]

Yes, I think it definitely needs to be made more clear on the docs. tbh, I am still not fully understanding. So there are 2 versions - the standards version and the version that is inside each rule defined in the standards file? if the standards version is greater than the rule version then it becomes a warning? Perhaps you can give some better example and explanation?

[willthames]

I talk a bit about versioned standards at https://willthames.github.io/2016/06/28/announcing-ansible-review.html

This means that you can update your standards without roles developed against earlier standards failing.

[ansiblejunky]

Thanks @willthames, I think I understand now. We generally don't set the Standards in our playbooks because we always want to apply the latest version of the lint rules. So in our standards.py we set the latest version to 1.0 here:

# Update this every time standards version increase
latest_version = Standard(dict(
    check=noop,
    name="No-op check to ensure latest standards version is set",
    version="1.0",
    types=[]
))

And then, the rules we want to FAIL in our standards.py we make sure they have version set as we do here:

tasks_are_named = Standard(dict(
    name="Tasks and handlers must be named",
    check=lintcheck('ANSIBLE0011'),
    version="0.0",
    types=["playbook", "task", "handler"],
))

Then when the playbook or task file breaks that rule, a FAIL will occur for that specific rule. If we want the rule to only be "suggestive" and warn developers, we remove the version from the above rule.