Add support for root detection via built-in project.el

[wedens]

Works with emacs >= 26

[coveralls]

Coverage increased (+0.1%) to 82.766% when pulling 7bc483b on wedens:patch-1 into 2e4aae5 on dajva:master.

[dajva]

Thanks a lot. Didn't know this existed.

Could you also add a test for it in rg-integration-test/project-root between the ffip and vc-backend test?
I think this should do it:

;; project.el
(rg-check-git-project-root)
(eval-after-load 'project
  (fmakunbound 'project-current))

[wedens]

@dajva I noticed commented test for projectile with a comment:

  ;; projectile require emacs 25.1 so can't test with cask since we
  ;; support emacs 24.4.

and package.el requires emacs >= 26. So, I wasn't sure I can test it.

I've tried to add the test, but CI is still failing and I'm not sure why

[dajva]

It should be fine testing this in versions, not supporting project.el. The test structure is a bit peculiar for this test but what will happen in that case is the vc-backend version will be invoked twice which is ok.

I think the problem actually lies in your patch. I looked at the project.el code and seems it doesn't return a string as all the other backends, but a cons cell of (symbol . dir-as-string). I think (you'll have to test this) that you need to use (cdr (project-current)) in you original patch.

[dajva]

(The problems with projectile lies entirely within cask)

[wedens]

Yeah, I forgot to add cdr 🤦‍♂️

CI is still failing for whatever reason. I don't have Cask, etc installed locally, but if I just call rg-project-root it seems to work as expected.

[dajva]

Ok, you can leave the tests. I'll have a look at this and then merge it.

[dajva]

Ok, this was entirely my fault. I used eval-after-load in the wrong way which accidentally worked for the other packages but not for project.el for some reason. I have merged this pr and pushed a fix for the eval-after-load problems.

Thanks