Iain Buclaw

@RazvanN7 going to rebase and finish this?

D will soon be a fixed language as far as this is concerned. https://github.com/dlang/dmd/pull/14538

> I'm dead in the water on this. I cannot repro any failures on my Ubuntu machine. None of the test machines build the compiler with -debug, so the crash...

@WalterBright this seems to be the reduction that confirms my observation. ``` #define pr16199_trigger(cond,func,args) _Generic (cond, default: func args) #define pr16199_skipped1(a) (1) /* This is incorrectly parsed by cparseExpression */...

> @tim-dlang you were right about the location of the segfault. It was the result of memory corruption. @ibuclaw your diagnosis was also correct. I was able to fix it....

> > Freebsd can't do any link/run tests - looks like importC is miscompiling zlib > > I seem to recall this was a pre-existing problem with FreeBSD. Something to...

> I'm mystified by this, as this PR is only about D code using #defines, it doesn't have anything to do with how the C preprocessor works. Maybe the FreeBSD...

This is the reproducer: ```c #define M16199C(X,S,M) ({ int __x; }) int pr16199c() { return 0; } ``` All functions after the macro definition get erroneously swallowed up, so the...

> damn, you're good, @ibuclaw ! This is the behaviour seen on FreeBSD that causes _nothing_ to be emitted. ```c #define M16199Da(TYPE,VAR) ((TYPE)(VAR)) #define M16199D(X,S,M) ({ int *__x = (X);...

``` ld: load command #3 extends beyond the end of the load commands file '../generated/osx/release/64/lexer.o' for architecture x86_64 ```