CRAN check warning about `_exit`, `abort` and `exit`

[yutannihilation]

https://cran.r-project.org/web/checks/check_results_string2path.html

  File ‘string2path/libs/string2path.so’:
    Found ‘_exit’, possibly from ‘_exit’ (C)
      Object: ‘rust/target/release/libstring2path.a’
    Found ‘abort’, possibly from ‘abort’ (C)
      Object: ‘rust/target/release/libstring2path.a’
    Found ‘exit’, possibly from ‘exit’ (C)
      Object: ‘rust/target/release/libstring2path.a’
  
  Compiled code should not call entry points which might terminate R nor
  write to stdout/stderr instead of to the console, nor use Fortran I/O
  nor system RNGs nor [v]sprintf.
  
  See ‘Writing portable packages’ in the ‘Writing R Extensions’ manual

[yutannihilation]

https://github.com/r-devel/r-svn/blob/b74106735de14725c7e5ebe41fce29da086a1781/src/library/tools/R/check.R#L4030-L4034

[yutannihilation]

• Add _R_CHECK_COMPILED_CODE_USE_OBJECTS_SYMBOL_TABLES_ env var to allow experimenting with PR#18789.
• Include symbols from static libraries used in PKG_LIBS in symbols.rds (PR#18789).

[yutannihilation]

https://bugs.r-project.org/show_bug.cgi?id=18789#c2

[yutannihilation]

After some thinking, I'm in the mood this cannot be fixed. Since Rust is not C, it cannot unwind to longjmp. So, the only possible strategy is just to abort and crash the R session. Are there any way to crash R gracefully...?

[yutannihilation]

Maybe I'm just confused. Let's rethink after enough sleep.

[yutannihilation]

hellorust fixed this so quickly. It seems this was just that the intermediate static lib was not cleaned up.

r-rust/hellorust@9a3ab01

[yutannihilation]

It seems this didn't work on the CRAN machine. But why...?

https://www.r-project.org/nosvn/R.check/r-devel-linux-x86_64-fedora-clang/string2path-00check.html

rm -Rf /data/gannet/ripley/R/packages/tests-clang/string2path/src/.cargo ./rust/target//release/build