stricter test for has_description

[MichaelChirico]

Just got bit by find_package landing on a DESCRIPTION in a parent directory which is a folder.

Changed the has_description test to be a bit stricter.

[AshesITR]

Could you add a test, e.g. using a mini Project with file structure

DESCRIPTION # file
dir
  |- DESCRIPTION # directory
  |  \- dummy_file # to have git track the directory
  \- R
      \- some_file.R

And testing the output of find_package(path = "dir/R/some_file.R")?
The root DESCRIPTION could be made a proper DESCRPITION file in dcf should we decide to check even more thoroughly in the future.

[MichaelChirico]

Sorry, I'm not following what we're trying to test exactly. We run lint_package on dir and it finds a lint in some_file, is that right? Don't we need R to be a sibling of the "correct" DESCRIPTION then?

[AshesITR]

I would keep the test atomic and only test that find_package returns the correct (expected) package directory.
No lints expected for this test IINM.

[MichaelChirico]

I was thinking to run lint_package to get at the behaviour through the "public" (exported) API, but of course we could directly test find_package as well. WDYT

[AshesITR]

Ah I see what you're getting at. That's a good point, would also be okay. In that case the DESCRIPTION directory should have a second-degree sibling directory named R which contains sources with lints.

[AshesITR]

@MichaelChirico Do you want to add a test for this so we can ship it with 3.0.0?
The change itself looks good to me.

[AshesITR]

LGTM