GH-46791: [C++] Add Status::OrElse, IntoStatus<T> and ToStatus

[pitrou]

Rationale for this change

In #46711 (comment) it was mentioned that the macro RETURN_NOT_OK_ELSE is confusing and can easily be misunderstood. We would like a better way to conditionally chain error-handling code if a Status does not indicate success.

What changes are included in this PR?

1. Add a type trait IntoStatus<T> that can be implemented to provide conversions from other error-like types
2. Add a global ToStatus function that calls the aforementioned type trait
3. Add a Status::OrElse method that calls a functor on error
4. Remove the RETURN_NOT_OK_ELSE macro

Are these changes tested?

Yes.

Are there any user-facing changes?

No, the RETURN_NOT_OK_ELSE was not supposed to be called by third-party code as it's not prefixed with ARROW_.

• GitHub Issue: [C++] Add Status::OrElse #46791

[github-actions]

⚠️ GitHub issue #46791has been automatically assigned in GitHub to PR creator.

[pitrou]

@zanmato1984@bkietz@paleolimbot What do you think?

[pitrou]

Ah, clang is being picky.

/Users/runner/work/arrow/arrow/cpp/src/arrow/status.h:363:16: error: cannot pass object of non-trivial type 'arrow::Status' through variadic method; call will abort at runtime [-Wnon-pod-varargs]
on_error(*this);
^
/Users/runner/work/arrow/arrow/cpp/src/parquet/arrow/writer.cc:392:31: note: in instantiation of function template specialization 'arrow::Status::OrElse<(lambda at /Users/runner/work/arrow/arrow/cpp/src/parquet/arrow/writer.cc:391:7)>' requested here
WriteRowGroup(0, 0).OrElse([&](...) { PARQUET_IGNORE_NOT_OK(Close()); }));
^

[paleolimbot]

These changes seems reasonable to me! The behaviour seems to line up exactly with Rust's .or_else() and I agree the macro was difficult to understand. The general maintenance around status handling is also great!

[pitrou]

The behaviour seems to line up exactly with Rust's .or_else()

You can guess it's intended :)

[zanmato1984]

This looks nice! A few nits.

[bkietz]

LGTM, just some nits and an idea for an alternate approach

[pitrou]

@github-actions crossbow submit -g cpp

[pitrou]

@github-actions crossbow submit -g cpp

[github-actions]

Revision: 65723c8

Submitted crossbow builds: ursacomputing/crossbow @ actions-dcfad22df7

Task	Status
example-cpp-minimal-build-static
example-cpp-minimal-build-static-system-dependency
example-cpp-tutorial
test-build-cpp-fuzz
test-conda-cpp
test-conda-cpp-valgrind
test-cuda-cpp-ubuntu-22.04-cuda-11.7.1
test-debian-12-cpp-amd64
test-debian-12-cpp-i386
test-fedora-39-cpp
test-ubuntu-22.04-cpp
test-ubuntu-22.04-cpp-20
test-ubuntu-22.04-cpp-bundled
test-ubuntu-22.04-cpp-emscripten
test-ubuntu-22.04-cpp-no-threading
test-ubuntu-24.04-cpp
test-ubuntu-24.04-cpp-bundled-offline
test-ubuntu-24.04-cpp-gcc-13-bundled
test-ubuntu-24.04-cpp-gcc-14
test-ubuntu-24.04-cpp-minimal-with-formats
test-ubuntu-24.04-cpp-thread-sanitizer

[pitrou]

@zanmato1984@paleolimbot@bkietz I've tried to address all your comments and suggestions, can you take a look again?

[paleolimbot]

This is a nice improvement on the ergonomics of working with the Status! I took a look through for anything out of place although some of the C++ metaprogramming is over my head here 🙂

[zanmato1984]

LGTM

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 3 benchmarking runs that have been run so far on merge-commit 0140089.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details.