[R-package] DESCRIPTION changes to address CRAN feedback

[jameslamb]

This addresses feedback from our first ever submission to CRAN!

#629 (comment)

• The note about "possible misspellings" can be fixed by putting LightGBM in quotes. I remember we had to do this when we first submitted uptasticsearch
• the error from "not installed on i386" is from setting Biarch: false in DESCRIPTION. I thought that flag could be a way to publish a package that doesn't support 32-bit, but I don't think that works. Let's remove that field and we'll get a better error on CRAN 32-bit checks.

This also proposes adding a cran-comments.md to track feedback from CRAN. Since the CRAN upload process involves back-and-forth with humans (not just automated checks), a lot of projects keep this type of file to hold a record of the things CRAN has asked them to do in the past. We do this in uptasticsearch.

[jameslamb]

@guolinke Could we submit to CRAN again? I want to be sure that this PR fixes the "misspelling" issue and gets us a different failure on Windows.

I just built a package that could be submitted, from this branch:

lightgbm_3.0.0-1.tar.gz

[jameslamb]

trying to fix this issue

* checking top-level files ... � NOTE
Non-standard file/directory found at top level:
  ‘cran-comments.md’

The regex for .Rbuildignore is really confusing.

[guolinke]

@jameslamb
package lightgbm_3.0.0-1.tar.gz does not pass the incoming checks automatically, please see the following pre-tests:
Windows: https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwin-builder.r-project.org%2Fincoming_pretest%2Flightgbm_3.0.0-1_20200811_054600%2FWindows%2F00check.log&data=02%7C01%7Cguolin.ke%40microsoft.com%7C9eb68df3881447f3508b08d83db1e59d%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637327181582245654&sdata=lT9Os6Tj1ZzwYp1qPLaBAv0mv20e4iw7kR4NC%2Bh5KLI%3D&reserved=0
Status: 1 ERROR, 2 NOTEs
Debian: https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwin-builder.r-project.org%2Fincoming_pretest%2Flightgbm_3.0.0-1_20200811_054600%2FDebian%2F00check.log&data=02%7C01%7Cguolin.ke%40microsoft.com%7C9eb68df3881447f3508b08d83db1e59d%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637327181582245654&sdata=sv0pGz%2FHlZ4PL3cG4FNMkverD1RS5ctco32NhFoeGc0%3D&reserved=0
Status: 2 NOTEs

[StrikerRUS]

Thanks for trying so hard!

LGTM, but please check my inline comment: #3298 (comment).

[jameslamb]

thanks @guolinke ! Ok I just added notes to cran-comments.md explaining my guesses for what went wrong. It seems quoting 'LightGBM' did eliminate the NOTE about misspellings, but the package is still not being installed on i386 architecture.

I just pushed another set of changes. At your earliest convenience, could you try another submission with this tarball?

lightgbm_3.0.0-1.tar.gz

[guolinke]

@jameslamb it takes a long time to get the result this time...
package lightgbm_3.0.0-1.tar.gz does not pass the incoming checks automatically, please see the following pre-tests:
Windows: https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwin-builder.r-project.org%2Fincoming_pretest%2Flightgbm_3.0.0-1_20200812_014949%2FWindows%2F00check.log&data=02%7C01%7Cguolin.ke%40microsoft.com%7C532ac2dabbe54eef150f08d83e9aec36%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637328182419417997&sdata=VNuGhBy55Yehcy%2FT8ludQKFY6bRKjgoS9tqvlPdOJpg%3D&reserved=0
Status: 1 ERROR, 1 NOTE
Debian: https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwin-builder.r-project.org%2Fincoming_pretest%2Flightgbm_3.0.0-1_20200812_014949%2FDebian%2F00check.log&data=02%7C01%7Cguolin.ke%40microsoft.com%7C532ac2dabbe54eef150f08d83e9aec36%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637328182419417997&sdata=aRLlC%2BVkwl7WvHiAt%2FGsbsSdGcaU9sqxKjbFieHY4e0%3D&reserved=0
Status: 1 NOTE

[jameslamb]

Thanks @guolinke . Ok this is good, I think this PR can be merged once CI finishes building. The 1 NOTE on both platforms can be safely ignored...it's just something that CRAN shows when you submit a package that isn't on CRAN yet.

The ERROR on Windows would be fixed by #3187 . Now we have absolute, definitive evidence that #3187 is required...before this, I was pretty sure but not certain.

Whenever you or @shiyu1994 pick that up, let me know if there's anything I can do to help. I can definitely create a CI job for 32-bit R based on #3187 (comment), and could help with any other setup.

[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.