Don't declare some special params as DI references, but as DI locals
instead.
Also enable some Win32 checks in the args_cdb test, as they are working
now with LLVM 7.
And make sure the template instance names include the template
parameters, a regression after #2826.
Also prevent `TemplatedSymbol(...).TemplatedSymbol` names, just use
`TemplatedSymbol!(...)` instead.
Once again closer to GDC and Clang's output, and attaching static variables to the module scope prevents an assert from triggering in IR/DIBuilder.cpp:
checkGlobalVariableScope(): Assertion `CT->getIdentifier().empty() && "Context of a global variable should not be a type with identifier"'
Previously entries were made children of the compile unit, and both global and imported symbols were only accessible by using their mangled name, or since the latest commit by their qualified name.
This brings the DWARF output closer to GDC's.
By discarding (and warning about) all but the first definition.
compilable/test17352.d tests that the compiler accepts 2 distinct but
equivalent templated function instantiations (with an alias to a
different local of the same type and name) ending up with a common
mangled name.
See https://issues.dlang.org/show_bug.cgi?id=17352.
* Disable debug info for now Until I get it working.
* Don't recurse into uninstansiated templates
* Remove device only modules from objfiles.
* Remove typeid generation for dcompute.
* Dont generate typeinfo member functions
* Whitelist `__dcompute_reflect`
Static arrays of classes, pointers to class (and perhaps global class variables) ended up with a wrong DIType because of wrong CreateTypeDescription() calls for class types.
Ex.: Array!FuncDeclaration caused an infinite recursion crash in GDB, because for its « FuncDeclaration[SMALLARRAYCAP] smallarray » field DIBuilder::CreateSArrayType() had DIBuilder return a value type instead of a reference type for FuncDeclaration.
The part needing most attention was ddmd.root.ctfloat, ddmd.target (incl.
gen/target.cpp) and ddmd.builtin. The front-end is now prepared for
elaborate compile-time floating-point types to allow for proper cross-
compilation.
This version still uses the host's `real` type for compile-time reals,
except for MSVC hosts, which still use 64-bit doubles (when compiled with
DMD host compiler too).
Some other changes:
* semantic*() of Statements extracted from statement.d to statementsem.d
* mangle() -> mangleToBuffer()
* Identifier::string -> toChars()
* Token::float80value => floatvalue
* Dsymbol::isAggregateMember() => isMember()
* BoolExp is no more
* ddmd.root.ctfloat: LDC-specific CTFE builtins
For nested variables. Resolves issue #1933 at the expense of a debuginfo
regression wrt. captured this pointer, see PR #1963.
Also add DWARF deref for non-ref/out variables captured by reference.
