Mathias LANG

> C++ interop tests Heisenbug filed as https://issues.dlang.org/show_bug.cgi?id=21130 Please don't open bug reports that can't be fixed. In this case, absolutely nothing we can do. And re-triggering the CI will...

CC @ibuclaw @rainers @WalterBright : Do you have any other mangling changes coming up ? Asking because @ibuclaw has submitted patches to libiberty (used by GDB & other coreutils) and...

> It shouldn't be. Just test for the Y case first, if it isn't, then do what is done now. In any case, I said I'd implement it in Phobos....

> Also this works: https://issues.dlang.org/show_bug.cgi?id=20607 I wish we just went with `extern (C) shared static this` instead of `pragma(crt_constructor)`. But at the time it would have been a silent change,...

There seem to be at least two bugs in the current implementation: Something to do with package, and something with template. Looking into it.

Of course the issue is forward references. Currently an `auto` function `A` calling another function `B` will trigger `B`'s semantic3 which is needed to infer attributes. If `A` and `B`...

No I don't think that's wise. I'd rather delay this until we have time for a proper fix than put another half-baked feature in the compiler. Not to mention, a...

The commit title needs to be `Fix [Issue] XXX`, the PR title doesn't matter to the bot (because changelog is generated from git log), which is why you can see...

> Whoops! The default windows makefile targets only build the compiler. I thought so too, but actually invoking `druntime/win*.mak` led to infinite recursion. Will get to it soonish.

> I prefer a warning to the user that this may happen in the case of a parameter that is marked `in`, but not `return`, yet is returned. But why...