Rework frexp as per #1329

[ckormanyos]

A billion no-hang runs locally. Lets progress here.

See also #1329

[ckormanyos]

OK the first draft of a non-hanging, yet clunky, version is in and cycling in CI.

Next steps involve:

• Add dedicated numerical checks at 32, 64, 128 bit. Testing on frexp is a bit light anyway and we could use some more.
• Optimize it, as there is ample room for improvement.

Cc: @mborland

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.8%. Comparing base (e72fd34) to head (1cb398e).

Additional details and impacted files

@@            Coverage Diff            @@
##           develop   #1330     +/-   ##
=========================================
+ Coverage     98.8%   98.8%   +0.1%     
=========================================
  Files          278     279      +1     
  Lines        18034   18088     +54     
  Branches      1918    1926      +8     
=========================================
+ Hits         17812   17867     +55     
+ Misses         222     221      -1     

Files with missing lines	Coverage Δ
include/boost/decimal/detail/cmath/frexp.hpp	95.3% <100.0%> (+3.3%)	⬆️
test/github_issue_1329.cpp	100.0% <100.0%> (ø)

... and 3 files with indirect coverage changes

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e72fd34...1cb398e. Read the comment docs.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[ckormanyos]

Hi Matt (@mborland), there are a few things woth discussing that are appearing in CI/CD.

On old GCC7 running on Drone, we hit the constexpr iteration limit of something like 262,144 when compiling erfc. This after the few trivial lines of code in frexp (albeit it might have a few new loop iterations now). I don't think the new frexp has huge amounts of new iterations. But I'm not sure if we should dive deeper or just increase the compiler limits on that run.

There is also verbose text regarding a note about some alignment running wild all over the place in CI/CD. It is ugly. I also get it locally. So we should really clean this up

Other than that, just the usual headache of CI...

[mborland]

If we can crank up the depth on GCC-7 that's fine with me. I don't think there's anything we can do about the alignment notifications since they're not warnings, just really annoying notifications about a change from like GCC 4.something

[ckormanyos]

If we can crank up the depth on GCC-7 that's fine with me.

Fine. I don't actually know where the compiler switches for Drone on that runner can be found.

Do you have any tip for that?

[mborland]

If we can crank up the depth on GCC-7 that's fine with me.

Fine. I don't actually know where the compiler switches for Drone on that runner can be found.

Do you have any tip for that?

Actually, I think the best move is to dump GCC-7 going into our first boost release. We already don't support GCC7 32-bit because it can't hang.

[ckormanyos]

If we can crank up the depth on GCC-7 that's fine with me.

Fine. I don't actually know where the compiler switches for Drone on that runner can be found.

Do you have any tip for that?

Actually, I think the best move is to dump GCC-7 going into our first boost release.

Yes please. Preferably immediately

[mborland]

If we can crank up the depth on GCC-7 that's fine with me.

Fine. I don't actually know where the compiler switches for Drone on that runner can be found.

Do you have any tip for that?

Actually, I think the best move is to dump GCC-7 going into our first boost release.

Yes please. Preferably immediately

Done, and updated the docs accordingly on this branch.

[ckormanyos]

Done, and updated the docs accordingly on this branch.

Thanks Matt.

Now something seemingly unrelated, kind of weird, is happenning on some of the runner environments in GHA? Or is that in Boost's CMake env?

[cppalliance-bot]

An automated preview of the documentation is available at https://1330.decimal.prtest3.cppalliance.org/libs/decimal/doc/html/index.html

If more commits are pushed to the pull request, the docs will rebuild at the same URL.

2026-02-04 21:35:49 UTC

[mborland]

Done, and updated the docs accordingly on this branch.

Thanks Matt.

Now something seemingly unrelated, kind of weird, is happenning on some of the runner environments in GHA? Or is that in Boost's CMake env?

It's an ubuntu problem. Happens from time to time.

[ckormanyos]

It's an ubuntu problem. Happens from time to time.

Got it. Thanks