[R-package] move R source files into R-package, reduce duplication in build_r.R

[jameslamb]

In this PR, I propose moving the R-specific C++ source files into the R-package directory. For CRAN builds, it's necessary for those files to be at the root level of the R-package's src/ directory (not src/src/ like we currently use when we copy in files). R CMD BUILD automatically compiles any .c, .cc, or .cpp files it finds in src/.

I also think there's a nice organizational benefit to having as much of the R package's code inside R-package/ as possible.

This PR also proposes some simplifications to build_r.R. We were repeating the string literal "lightgbm_r" a lot, and I think it would be better to store it in one constant at the top of the script.

This is only moving files around and will not change anything for our users.

How this helps us get to CRAN

This is the next PR on the road to #629 . I'm working on #2960 in my fork. That PR includes build steps for the R package which only use the CRAN toolchain (e.g. no Visual Studio, no CMake).

That PR on my fork right now has a large diff that would be difficult to review, and it needs a bit more work. So I'd like to propose the first batch of changes here.

[StrikerRUS]

There are some other files that should be updated:
https://github.com/microsoft/LightGBM/blob/master/CMakeLists.txt
https://github.com/microsoft/LightGBM/blob/master/.github/CODEOWNERS

[jameslamb]

There are some other files that should be updated:
https://github.com/microsoft/LightGBM/blob/master/CMakeLists.txt
https://github.com/microsoft/LightGBM/blob/master/.github/CODEOWNERS

I don't think anything in CMakeLists.txt needs to be updated. The R-specific parts only get run once that file has been copied into an R package, for example lightgbm_r/ directory created by Rscript build_r.R. That is why all R builds are working right now on CI without touching CMakeLists.txt (except Mac on Travis, which is failing across all branches for unknown reasons). We already don't support building the library for R by running cmake from the repo root, so I don't think this breaks anything.

Good catch on CODEOWNERS! Didn't even think of that. Now we'll be able to eliminate these three lines in that file:

https://github.com/microsoft/LightGBM/blob/master/.github/CODEOWNERS#L18-L20

[jameslamb]

Currently looking into this cpplint warning.

R-package/src/lightgbm_R.cpp:8:  R-package/src/lightgbm_R.cpp should include its header file R-package/src/lightgbm_R.h  [build/include] [5]

Here's the relevant source code in cpplint: https://github.com/cpplint/cpplint/blob/6e3e178b031169be858ba6a3cf96d263a04c2b17/cpplint.py#L2213-L2237

[jameslamb]

Currently looking into this cpplint warning.

R-package/src/lightgbm_R.cpp:8:  R-package/src/lightgbm_R.cpp should include its header file R-package/src/lightgbm_R.h  [build/include] [5]

Here's the relevant source code in cpplint: https://github.com/cpplint/cpplint/blob/6e3e178b031169be858ba6a3cf96d263a04c2b17/cpplint.py#L2213-L2237

it was this: cpplint/cpplint#139

The fix is using "lightgbm_R.h" instead of "./lightgbm_R.h", which I did in 1e97aae

[jameslamb]

Ok I think this is ready! @Laurae2 or @guolinke I won't merge until one of you has a chance to review

[StrikerRUS]

LGTM!

Removing my "changes requested" review status.

Adrdessed

[github-actions]

This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https://github.com/microsoft/LightGBM/issues including a reference to this.