test: Define BOOST_TEST_DISABLE_ALT_STACK only where Boost.Test uses it #36600

Boost.Test consults the BOOST_TEST_DISABLE_ALT_STACK macro only in its sigaction-based signal handling, which it selects when BOOST_HAS_SIGACTION is defined. This causes a build failure when cross-compiling for Windows with Clang after bitcoin/bitcoin#36509:

/home/runner/work/bitcoin-core-nightly/bitcoin-core-nightly/src/test/main.cpp:9:9: error: macro is not used [-Werror,-Wunused-macros]
    9 | #define BOOST_TEST_DISABLE_ALT_STACK
      |         ^

(not sure why GCC fails to catch this case in the CI.)

This PR fixes the issue.

<!--e57a25ab6845829454e8d69fc972939a-->

The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

<!--006a51241073e994b41acfe9ec718e94-->

External sites

• Code coverage & benchmarks: https://corecheck.dev/bitcoin/bitcoin/pulls/36600
• Ralph review <img src=https://review.fish.foo/assets/bitcoin/bitcoin/36600.png alt="Ralph review status"> https://git.fish.foo/bitcoin/bitcoin/pulls/36600#author-ralph

<!--021abf342d371248e50ceaed478a90ca-->

Reviews

See the guideline and AI policy for information on the review process.

If your review is incorrectly listed, please copy-paste <code>&lt;!--meta-tag:bot-skip--&gt;</code> into the comment that the bot should ignore.

<!--5faf32d7da4f0f540f40219e4f7537a3-->

Also see https://github.com/hebasto/bitcoin-core-nightly/pull/414.

Friendly ping @l0rinc @maflcko

lgtm ACK 3d446f5738b0b64ccea13f8cf907333ad807a2ac

Though, boost is scheduled for removal, so alternatives could be:

• Set -Wno-error=unused-macros in the Windows cross-builds temporarily; or
• Temporarily drop the Wunused-macros from the cmakelist

Set -Wno-error=unused-macros in the Windows cross-builds temporarily;

This seems like the proper solution here. Not sure why we'd revert changes in this repo, to fix a failure in a nightly repo (where it's trivial to work around this), and it seems like the underlying issue isn't diagnosed ("(not sure why GCC fails to catch this case in the CI.)").

Not sure why we'd revert changes in this repo

To be clear. I think if something is important enough that we'd revert changes here to "fix" it, then it (the CI job) should be in this repo, so it doesn't get broken in the first place.

Yeah, this seems pretty irrelevant, so my preference would be to just -Wno-error=unused-macros in the Windows cross-builds temporarily. But anything is fine here, no strong opinion.

Closed in favour of https://github.com/hebasto/bitcoin-core-nightly/pull/415.

Not sure why we'd revert changes in this repo

To be clear. I think if something is important enough that we'd revert changes here to "fix" it, then it (the CI job) should be in this repo, so it doesn't get broken in the first place.

Not related to this particular issue, but we still have some kinds of tests, such as actually running tests on *BSD platforms, that are not part of this repo's CI. Still, issues with such tests have to be addressed in this repo.

Yes. It's the same issue as #33438.