.Rprofile in admiral gets changed by styler

[esimms999-gsk]

In admiral@devel, the .Rprofile file in admiral gets changed by styler::style_pkg() and then shows up in the git pane. This is due to some blanks. I think that everything in admiral@devel should go through styler without changes; all pass except .Rprofile.

I think that admiral's .Rprofile comes from admiralci, but I have not figured out how. I expect a fix would need to be done somewhere in that process.

[bms63]

Can we just style it and commit it??

[esimms999-gsk]

@bms63 I don't think it is that straight forward. If I understand correctly, the .Rprofile file in admiral comes from admiralci ... and it is not a direct copy. If I "fix" .Rprofile in admiral, it will soon be over-written by some process in admiralci.

[manciniedoardo]

@esimms999-gsk is this still a problem?

[esimms999-gsk]

@manciniedoardo I just created a project with admiral, switched to the devel branch and ran styler::style_pkg(). Only .Rprofile gets changed. The lines changed by styler are a bit different now:

Interesting that it wants to comment out that source() statement; anyone know what is up with that? I know that source() is frowned upon in packages, but would not expect styler to make such a bold move.

This is not a huge problem ... but it is a bit messy in that styler being run on our production package should be clean, i.e. no changes. I think of it in the same way as if we ran our 697 tests and one of them failed and we said "yeah, we know about that, it isn't important".

[bms63]

My understanding is that the .Rprofile doesn't ship with the package, just like renv and .git stuff doesn't ship with package either. It is purely for us as developers to configure and track our environments. I believe the impact on users is zero.

If we want to track and attempt to fix, then I suggest we make an issue in admiralci and we can attempt to address it.

[galachad]

@esimms999-gsk should be fixed in last propagation.

[bms63]

Closing as fixed in latest propagation