build: remove usage of BOOST_NO_CXX98_FUNCTION_BASE

[fanquake]

Supposedly this is no-longer needed (and hasn't been ported to CMake), so remove it's usage here. Oringinally used to suppress warnings about functionality deprecated/removed from the standard library, which was still supported by some compilers.

[DrahtBot]

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

Code Coverage

For detailed information about the code coverage, see the test coverage report.

Reviews

See the guideline for information on the review process.

Type	Reviewers
ACK	hebasto

If your review is incorrectly listed, please react with 👎 to this comment and the bot will ignore it on the next update.

[maflcko]

Did you test this with the minimum required boost version 1.71?

[fanquake]

Did you test this with the minimum required boost version 1.71?

Only with 1.81.0. (depends) which still has the problematic behaviour. Used a CMake CI (which has already dropped this) with Boost 1.73.0 as a proxy for anything lower (https://github.com/hebasto/bitcoin/actions/runs/9257726144/job/25466464660).

[maflcko]

The cmake run is using g++-11, but g++-12 is required according to https://github.com/boostorg/config/pull/430/files

[fanquake]

@hebasto how did you test this macro is no-longer required in the CMake build (given it wasn't ported)?

[hebasto]

@hebasto how did you test this macro is no-longer required in the CMake build (given it wasn't ported)?

https://github.com/hebasto/bitcoin/actions/runs/9283914497

[hebasto]

Did you test this with the minimum required boost version 1.71?

It is 1.73 since #29066.

[hebasto]

Concept ACK.

[maflcko]

Looks like 880d4aa affected depends, not sure if it even reproduces outside of that?

[maflcko]

I could only reproduce inside of depends, which has boost 1.81, which is fixed, so LGTM, I guess. 🤷

[hebasto]

ACK fc47935.

[maflcko]

I stopped seeing the warning after commit 3b2acfc

[maflcko]

It would be good to properly explain this, because this still happens, at least with at least

• --disable-suppress-external-warnings
• depends boost 1.73
• g++-12 and clang-18

[fanquake]

It would be good to properly explain this,

The explanation is that it still happens, it never should have been incorrectly dropped from the CMake build (where it can now be re-added), and this can be closed.

[maflcko]

Does --disable-suppress-external-warnings only work with depends? (I can't seem to get the warning on Jammy with clang-18 outside of depends)

Should --disable-suppress-external-warnings be enabled in some CI tasks?

[maflcko]

If it is clear, that this can only happen inside of depends, where it is fixed after 1.81, then I think it is fine to merge. However, I am trying to figure out if this can happen outside of depends.

[fanquake]

Does --disable-suppress-external-warnings only work with depends?

It should not be depends specific.

Should --disable-suppress-external-warnings be enabled in some CI tasks?

That depends on what we are trying to achieve, and if we are planning on patching external libraries / code and upstreaming relevant changes etc. It's also likely to lead to more CI "breakage" on distro version, or other changes, that may not be relevant for us.

[maflcko]

That depends on what we are trying to achieve

Wouldn't it be good to know about issues, such as the one here (BOOST_NO_CXX98_FUNCTION_BASE) as early as possible? Not sure how to achieve that, other than setting --disable-suppress-external-warnings in at least some CI tasks.

[fanquake]

Wouldn't it be good to know about issues, as early as possible?

I think if we are going to do that, then enabling it on a rolling / nightly distro type CI, i.e fedora rawhide, would be the most useful, otherwise, I think it's unlikely to actually turn much up, given packages are pinned in depends, and in the distro. Note that *--suppress-external-warnings was another feature not ported to CMake, as apparently it isn't needed; so someone would have to look at porting it again, so it's possible to achieve this same functionality, of exposing all warnings.

[maflcko]

I think it's unlikely to actually turn much up, given packages are pinned in depends

at least for the sanitizer builds, the compiler is bumped regularly, so it seems good to check those bumps when they happen.