remove IRState.falseBlock

[WalterBright]

Shouldn't need it anymore.

[dlang-bot]

Thanks for your pull request, @WalterBright!

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

Testing this PR locally

If you don't have a local development environment setup, you can use Digger to test this PR:

dub run digger -- build "master + dmd#14943"

[ibuclaw]

Isn't the inverse needed for the else block when if (!__ctfe) ?

[WalterBright]

Yes, if we really wanted to support that. I'm on the fence about it. There are other __ctfe constructs, too. I figure we make this solid first.

[ibuclaw]

This should target stable, as it's a continuation of the fix for regression 23710 in all but name.

[WalterBright]

macOS 13.x x64 (M1), DMD (latest) Failing after 22m

Looks like a problem with master, not this PR. @RazvanN7 ?

[ibuclaw]

macOS 13.x x64 (M1), DMD (latest) Failing after 22m

Looks like a problem with master, not this PR. @RazvanN7 ?

Seems like dmd's yl2xp1 intrinsic doesn't work on macOS 13.x (emulating x86 on M1) with small inputs.

log1p(0x1p-16382): expected: 0x1p-16382 | got: 0x0p+0
****** FAIL release64 std.math.exponential
core.exception.AssertError@std/math/exponential.d(3765): Assertion failure
----------------

fabs(0x1p-16382) = 3.3621e-4932
fabs(0x1p-16382) <= 0.25 = true
yl2xp1(0x1p-16382) = 0x0p+0

[WalterBright]

@ibuclaw I made this for stable #14946 because my dev work on master needs this one.

[ibuclaw]

Shouldn't be required to have both, just merge stable then create a PR to git merge stable --no-ff on master.

[WalterBright]

Every time I've tried something like that I get a PR with ten thousand lines of difference. Just merge both PRs and things will be fine!

[ibuclaw]

Automated merging of stable fixes into master is something that needs looking into as it's quite silly to have bug fixes diverged from mainline for any lengthy period of time.

[WalterBright]

If git had a git undo command I'd be a lot more adventurous with it. But as it stands, it's easier to just delete 16 lines into a new PR and submit it.

[ibuclaw]

Can't do anything about Apple having buggy emulators except turn off fyl2x instructions on macOS.

dlang/phobos#8705

[WalterBright]

Thanks for identifying the problem and supplying a workaround.

[ibuclaw]

Superseded by #14947.

[ibuclaw]

If git had a git undo command I'd be a lot more adventurous with it. But as it stands, it's easier to just delete 16 lines into a new PR and submit it.

Every change you make with git goes into the reflog, you don't lose it. https://git-scm.com/docs/git-reflog

git reflog

Which lists both commit hashes which can also be referenced with HEAD@{N}.

Undo then becomes:

git reset --hard the-reflog-hash

Or

git reset --hard HEAD@{1}

[ibuclaw]

Other PR merged.

[WalterBright]

Thanks for reflog tip