Updated the runtest script to build libtangobos-partial.a if it hasn't already been built.
Added in signbit() and va_arg!()() to libtangobos-partial.a so more of the phobos dependent DStress tests pass.
_aaEq was added to runtime/internal/aaA.d which forwards to
TypeInfo_AssociativeArray.equals in genobj.d. On the codegen side, DtoAAEquals
was added to gen/aa.cpp and is called from EqualExp::toElem in gen/toir.cpp.
I assume that the frontend will produce an error if == is used on associative
arrays of different type.
This fixes DMD bug 1429.
The idea is to separate the notion of const from 'this variable can always be
replaced with its initializer' in the frontend. To do that, I introduced
Declaration::isSameAsInitializer, which is overridden in VarDeclaration to
return false for constants that have a struct literal initializer.
So
{{{
const S s = S(5);
void foo() { auto ps = &s; }
// is no longer replaced by
void foo() { auto ps = &(S(5)); }
}}}
To make taking the address of a struct constant with a struct-initializer
outside of function scope possible, I made sure that AddrExp::optimize doesn't
try to run the argument's optimization with WANTinterpret - that'd again
replace the constant with a struct literal temporary.
called on an `AliasDeclaration`; just printing the name will do. This fixes
#305, which otherwise tries to generate
{{{
class E {
void A() {
alias /* recurse into E->toCBuffer() */ m;
}
}
}}}
by way of an infinite recursion (causing a segfault when the stack runs out).
Recent optimization improvements made LLVM realize the store-to-null was
unavoidable, so it deleted all of main() and replaced it with 'unreachable'.
Because the body of main() no longer even contained a return instruction,
calling it caused random code to be ran instead. This happened to be the code
that links in the ModuleInfo on my machine, which then returned "successfully".
they were longs.
This was causing asm1_1 to fail when compiled with -O3 because it was
overwriting the spilled value of callee-saved register %rbx, which the runtime
was using as a pointer value at the time.
functions.
There's no need to waste cycles with extern(D), which we get to define
ourselves. Fixes tests/mini/asm8.d. (Since the asm abiret code already assumed
{xmm0, xmm1} returns)
The assembler was miscompiling "add" (specifically, the "add reg/mem, imm"
variations).
The change that caused this seems to have been made because without it, some
"add"s didn't compile at all.
This patch reverts the previous change, and makes sure assembler operands are
remapped correctly even though the input operands auto-generated due to
updating operations aren't explicitly used.
