Normally, run the tests via the Makefile in the parent directory. Go there, then make test, make test_opt or make miri. Setting RUST_BACKTRACE=1 should give some location information on the source of an error.

• div/ contains various hand-picked files with challenges

• chj-home/ contains the patches of a public repository, as created by git format-patch using git version 2.47.3

• errors/ contains patches that test the error response

• bugs/ contains patches that trigger unresolved bugs; they are not run as part of the normal test suite from the Makefile. But you can run them via:

    test/run-test-for-input-dir test/bugs/
  

  Or just run split-patch on a file directly.

  When resolving a bug, move the file that triggered it to errors/ or div/ (depending on whether split-patch now correctly reports an error or splits it correctly, respectively).

Add the file to the right directory. Then run

test/split-patches-in-dir /opt/chj/bin/split-patch test/"$input_dir" test/"$input_dir"-expected
git add test/"$input_dir" test/"$input_dir"-expected && git commit

Verify that the Rust version matches via

test/run-test-for-input-dir test/"$input_dir

then if it doesn't, decide how to resolve (e.g. if decided that it's actually producing the correct result, move the files from test/"$input_dir"-current/"$name_of_patch_file" to the corresponding ...-expected dir).