[R-package] Avoid bashisms (non-POSIX code) in R-package/configure

[smoser]

On systems that did not have bash as /bin/sh this would end up causing strange compilation errors.

Fixes #6743.

[jameslamb]

Thanks for the fix!

I was under the impression that R CMD check would run checkbashisms and that that would catch stuff like this (#6743 (comment)).

But it seems not. I just ran it (on my mac, with Homebrew's checkbashisms v2.23.7),

checkbashisms ./R-package/configure

It caught these += issues and some others. Do any of these look relevant?

possible bashism in ./R-package/configure line 344 ([^] should be [!]):
    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{
possible bashism in ./R-package/configure line 348 ([^] should be [!]):
          /^X\(\/\/\)[^/].*/{
possible bashism in ./R-package/configure line 462 ([^] should be [!]):
    sed '/^.*\/\([^/][^/]*\)\/*$/{
possible bashism in ./R-package/configure line 526 (unsafe echo with backslash):
  case `echo 'xy\c'` in
possible bashism in ./R-package/configure line 1792 (should be VAR="${VAR}foo"):
    LGB_CPPFLAGS+=" -DMM_PREFETCH=1"
possible bashism in ./R-package/configure line 1827 (should be VAR="${VAR}foo"):
    LGB_CPPFLAGS+=" -DMM_MALLOC=1"
possible bashism in ./R-package/configure line 1853 (should be >word 2>&1):
    if command -v brew &> /dev/null; then
possible bashism in ./R-package/configure line 1857 (should be >word 2>&1):
        brew --prefix libomp &>/dev/null && ac_brew_openmp=yes

The sed ones are code-generated by autoconf and out of our control. But those two output-redirection ones about brew are in our control.

Also... can you please try testing your branch on whatever system you're on that led to the errors reported in #6743?

git clone --recursive [email protected]:microsoft/LightGBM.git
cd ./LightGBM
sh build-cran-package.sh --no-build-vignettes
R CMD INSTALL ./lightgbm_*.tar.gz

[smoser]

@smoser please read the following Contributor License Agreement(CLA). If you agree with the CLA, please reply with the following information.
...

@microsoft-github-policy-service agree company="Chainguard"

[smoser]

The sed ones are code-generated by autoconf and out of our control. But those two output-redirection ones about brew are in our control.

Fixed the output-redirection (those are definitely not posix). I'll test later, but the changes here are absolutely improvement on their own.

[smoser]

Also... can you please try testing your branch on whatever system you're on that led to the errors reported in #6743?

$ git clone --recursive https://github.com/microsoft/LightGBM.git
$ cd LightGBM
$ git remote add smoser https://github.com/smoser/LightGBM.git
$ git checkout smoser/fix-plus-equal-not-posix
$ git log HEAD^.. --oneline
6f3b008a Remove bashisms in R-package/configure
$ sh build-cran-package.sh --no-build-vignettes
Building lightgbm with R executable: R
Removing files not needed for CRAN
Removing unknown pragmas in headers
Cleaning sed backup files
* checking for file ‘lightgbm_r/DESCRIPTION’ ... OK
* preparing ‘lightgbm’:
* checking DESCRIPTION meta-information ... OK
* cleaning src
* running ‘cleanup’
* checking for LF line-endings in source and make files and shell scripts
* checking for empty or unneeded directories
* looking to see if a ‘data/datalist’ file should be added
* building ‘lightgbm_4.5.0.99.tar.gz’

Done building R-package

Then copied the lightgbm_4.5.0.99.tar.gz into a container where I previously had the problem.

$ R CMD INSTALL ./lightgbm_4.5.0.99.tar.gz   
* installing to library '/usr/lib/R/library'
ERROR: dependencies 'R6', 'data.table', 'jsonlite' are not available for package 'lightgbm'

$ R -e "install.packages('R6', repos = 'https://cran.rstudio.com/')"
$ R -e "install.packages('data.table', repos = 'https://cran.rstudio.com/')"
$ R -e "install.packages('jsonlite', repos = 'https://cran.rstudio.com/')"

$ R CMD INSTALL ./lightgbm_4.5.0.99.tar.gz
...
* installing to library '/usr/lib/R/library'
* installing *source* package 'lightgbm' ...
** using staged installation
checking location of R... /usr/lib/R
checking whether MM_PREFETCH works... yes
checking whether MM_MALLOC works... yes
configure: creating ./config.status
config.status: creating src/Makevars
** libs
using C++ compiler: 'g++ (Wolfi 14.2.0-r5) 14.2.0'
using C++17
...

Note, I clearly don't know what I'm doing with R dependencies.. I'm sure there is a better way to install the necessary deps, but i muddled my way through there.

The output of configure looks sane and its building now, so this LGTM.

[jameslamb]

Nah that's great, thank you very much! Just one more thing... it looks like you didn't push the changes you made related to output-redirection (I still just see += changes). Could you push those?

[jameslamb]

Thanks very much for the report and fix!