Feature Request: Further customise of our .lintr file

[manciniedoardo]

Feature Idea

What if we leveraged the functionality in {lintr} to make custom linters?

One example I can think of is related to our recent error messaging update. Now that we use {cli}, do we ever need to use stop() or warning()? If not, we could make a custom linter for them to avoid ever introducing them in our codebase.

@pharmaverse/admiral any thoughts?

Some thoughts:

• This is of course relevant for {admiraldev} as well.
• If we go ahead with this, it would be good to include some documentation about this. And also to include it maybe in the ext packages to try and enforce a certain way of doing things?

[bms63]

I'm curious how we can make this work for us especially since we are reaching more of a steady state. This way some of our programming conventions can be automatically checked and enforced.

.... but I feel like it should be a discussion rather than an issue.

[bundfussr]

This could be particularly useful for extension packages.

• We would need to find a way to centralize the customizations. At the moment we have a .lintr file in each package. If we copy the customizations to each package, it will be hard to maintain.
• If one of the custom linters fails, it should be clear from the message that it is an admiral specific one. Otherwise it could be confusing if using warning() is marked as an issue.

[manciniedoardo]

I guess we could control it through the .lintr file in the {admiraltemplate} repo?

[bms63]

We could have a central .lintr file in admiraldev that the workflow grabs when doing the check.

[bundfussr]

We could have a central .lintr file in admiraldev that the workflow grabs when doing the check.

If this is possible, I would prefer it. It seems better maintainable to me than a .lintr file in {admiraltemplate}.

[bms63]

@cicdguy possible to maintain a central .lintr file in admiraldev or in admiralci that all admiral repos use?