Thank you all for the help and clarification! I'll go fix our code then.

(I know this is a very contrived example with void operator!=, but that is what CVise spat out, and the actual failures are related to comparison operators too)

I tried out D135772, and our build got significantly farther than before! I unfortunately discovered another piece of code that pre babdef27c503c0bbbcc017e9f88affddda90ea4e Clang and GCC accept in C++20 mode, but Clang trunk does not (https://godbolt.org/z/q1q4nfobK):

In D126907#3849583, @erichkeane wrote:

In D126907#3846591, @erichkeane wrote:

In D126907#3844788, @BertalanD wrote:

Hi @erichkeane,

This change broke compilation of this program (https://godbolt.org/z/KrWGvcf8h; reduced from https://github.com/SerenityOS/ladybird):

template<typename T, typename U>
constexpr bool IsSame = false;

template<typename T>
constexpr bool IsSame<T, T> = true;

template<typename T>
struct Foo {
    template<typename U>
    Foo(U&&) requires (!IsSame<U, Foo>);
};

template<>
struct Foo<void> : Foo<int> {
    using Foo<int>::Foo;
};

Foo<void> test() { return 0; }

<source>:18:27: error: invalid reference to function 'Foo': constraints not satisfied
Foo<void> test() { return 0; }
                          ^
<source>:10:24: note: because substituted constraint expression is ill-formed: value of type '<dependent type>' is not contextually convertible to 'bool'
    Foo(U&&) requires (!IsSame<U, Foo>);
                       ^

Thanks for the report! I'll look into it ASAP.

Quick note: I believe I understand the cause of this, which requires a bit more work than I otherwise would have expected. I have a candidate patch I'm running through my testing right now that should fix this, but it still needs cleaning up. Expect it in the next day or two if all goes well.

This change broke compilation of this program (https://godbolt.org/z/KrWGvcf8h; reduced from https://github.com/SerenityOS/ladybird):

I'll be mostly AFK for a couple of hours, so feel free to revert immediately if this commit starts breaking bots.

Added driver tests, tweaked help messages.

Cleaned up writeStub() implementations by taking the GOT/lazy pointer address as a parameter. Addressed review comments from Jez.

thank you!

rebased and started addressing review comments

Submitted in 3493f1a10791.

Thank you for the write-up, it was very interesting!

Sort DataInCodeSection's contents as suggested by thakis.

It kind of makes sense to require the data in code table in the output to be sorted too, so that tools can binary sort in _it_.

Changed the readValuelambda into a standalone function and removed a stray #undef.

Removed ObjFile::optimizationHints.

Addressed review comments, tweaked the comment on applyOptimizationHints and updated the description to include that the previous overhead was close to 100 milliseconds.

Rebased in order to re-run precommit checks before I push.

Thank you for the review!

Improved comments.

Rebase on top of D132865 and D132947.

Addressed review comments.

Fix CI by properly ignoring error during the first iteration in PrintDyldInfo().

pretty-print the table with proper column alignment

Attempt to fix the Windows build.

• Moved code for finding the next fixup into a helper function (findNextPageWithFixups).
• Folded printMachOChainedFixups into PrintDyldInfo.
• Added test to ensure that we can handle objects that don't use chained fixups.
• Minor comment tweaks.

Added diagnostics for too old deployment targets, rebase target address overflows, and tweaked comments.

I cleaned the patch up a bit and rebased it. D132560 depends on this for tests.

Updated StubsSection::addEntry() comment.

Call in.stubs->addEntry() directly, and use needsBinding() to check if a stub is needed.