ArchetypeMethodInst, like ProtocolMethodInst, actually needs to be modeled as returning a thick function type, in order to represent the metadata parameter to the witness table function. With that change to SILGen, implement ArchetypeMethodInst IRGen.
Swift SVN r4688
Doug found that the verifier would admit a DowncastInst that took an lvalue as the input, which is bogus. Refine the verifier rules for DowncastInst so it only admits class rvalues and destination types.
Swift SVN r4685
During type checking, fake up a "forwarding" substitution set for constructor decls that maps archetypes to themselves. This is goofy, but type-checking is the place that knows how to make ProtocolConformances, and I don't want to reinvent that. Use these forwarding substitutions in SILGen when emitting a generic allocating constructor so we can properly emit a call to the corresponding initializing constructor.
Swift SVN r4684
On second thought, having SILConstant be able to point to a TLCD is going
against the goal of making SILConstant be a "SILGen thing". I'll find another
approach.
Swift SVN r4680
Factor out the logic for properly storing a loadable or address-only rvalue into ManagedValue::forwardInto and ManagedValue::assignInto, and use those methods everywhere instead of trying to cut corners and directly emit store instructions.
Swift SVN r4670
Introduce an 'RValue' object in SILGen that behaves like a simplified version of IRGen's 'Explosion' object, holding a destructured tuple in a naturally destructured state. This cleans up a bunch of ad-hoc scattered code that destructured tuples in various places. Update SILGenFunction so that the expression visitors all pass around RValues, and update CallEmission to handle argument clauses as RValues instead of a pre-destructured string of arguments. Implement "emit into" optimization of TupleExpr and ScalarToTupleExpr, now that that's easy to do.
Swift SVN r4667
Create a new FallthroughStmt, which transfers control from a 'case' or 'default' block to the next 'case' or 'default' block within a switch. Implement parsing and sema for FallthroughStmt, which syntactically consists of a single 'fallthrough' keyword. Sema verifies that 'fallthrough' actually appears inside a switch statement and that there is a following case or default block to pass control to.
SILGen/IRGen support forthcoming.
Swift SVN r4653
Unfortunately, this regresses the repl when expressions like (1,2) are entered. This is because the repl is violating some invariants (forming dags out of ASTs, making ASDAG's which upset the type checker). I'm going to fix this next, but can't bring myself to do it in the same commit.
Swift SVN r4617
The Verifier wasn't actually verifying function bodies, because I neglected to visit the actual basic blocks after checking the entry point arguments in verifySILFunction. This revealed a SILType identity issue where TypeConverter::getLoweredType and SILType::getEmptyTupleType returned non-identical SILTypes for the empty tuple type; fix that by removing SILType::getEmptyTupleType, moving TypeConverter into SILModule, and forcing all SILType creation through TypeConverter.
Swift SVN r4616
At the top level, if 'operator' is followed by 'infix', 'prefix', or 'postfix', consider it a contextual keyword, and parse an operator decl following it that looks like:
operator {infix|postfix|prefix} <+> {
attributes…
}
Prefix and postfix operator decls currently admit no attributes. Infix operators have 'associativity {left|right|none}' and 'precedence <int>' attributes.
This patch implements parsing for operator declarations but does not yet attach the declared attributes to func decls for the operators.
Swift SVN r4596
Function and compound types have a bunch of extra calling convention and uncurrying info stuffed into a "SILTypeInfo" object that until now had to be fetched through the SILModule. Change the representation of SILType to be a PointerUnion of CanType and SILTypeInfo*, and move the uncurry level onto the SILTypeInfo for functions, so that SILTypeInfo is available directly through the SILType and SILType can go back to being pointer-sized.
Swift SVN r4582
Other SIL producers (like the parser) will probably need to be able to lower Swift Types to SILTypes, so make TypeLowering.h a public SIL header instead of an implementation detail of SILGen.
Swift SVN r4577
Represent class_method and super_method SIL values using an ObjCMethod variant of LoweredValue that holds the necessary information to emit an objc_msgSend* call when the value is apply-ed. You can't really use ObjC objects from SIL IRGen yet because SILGen doesn't yet visit external definitions from the Clang importer.
Swift SVN r4550
Archetype and protocol 'x.metatype' expressions eventually need to do a dynamic lookup, like 'x.metatype' for classes. These instructions represent that lookup.
Swift SVN r4534
Provide distinct syntax 'a as T' for coercions and 'a as! T' for unchecked downcasts, and add type-checker logic specialized to coercions and downcasts for these expressions. Change the AST representation of ExplicitCastExpr to keep the destination type as a TypeLoc rather than a subexpression, and change the names of the nodes to UncheckedDowncast and UncheckedSuperToArchetype to make their unchecked-ness explicit and disambiguate them from future checked casts.
In order to keep the changes staged, this doesn't yet affect the T(x) constructor syntax, which will for the time being still perform any construction, coercion, or cast.
Swift SVN r4498
If the result of an expression is address-only, we can store the result in-place if we have an uninitialized destination buffer for it, such as in a variable initialization or return. Modify SILGenFunction::visitExpr to write to the buffer of an Initialization if one is available instead of allocating a temporary.
A couple cases that still need to be implemented are destructuring tuple initializations (e.g. 'var (a, b) = (foo(), bar())') and apply expression results.
Swift SVN r4492
Use unique_ptr to manage the ownership of sub-initializations, add the ability for contiguous tuple initializations to break up into non-contiguous tuple initializations so that tuple expressions can be uniformly "emit-into"-ed, and formalize the different broad categories of initialization kind.
Swift SVN r4491
Instead of trying to be cute and lazily copy address-only values, always pass around ManagedValues at +1 (unless they represent lvalues). This cleans up some special-case clutter for address-only values.
Swift SVN r4488
Add code to the verifier to verify that the entry point BB for a function has the right argument types for its function type, return instructions match the function type, and branch instruction arguments match the branched-to BB.
Swift SVN r4487
All the other implicit conversions have been given their own instructions, so specialize this ill-specified instruction to its one remaining use, for converting compatible function types.
Swift SVN r4482
Set up IRGen to emit SIL code that uses top-level-code global variables. Add -sil-i to a bunch of Interpreter tests that use global variables.
Swift SVN r4480
In top-level code, global variables are notionally local variables of the "main" function, but we give them global storage as an implementation detail. Add the ability to represent physical global variables to SILConstant, and emit top-level-code global variable initializers and references in terms of the physical address.
Swift SVN r4479
FunctionConversionExpr was converting the type of SIL values to thick function types without emitting SIL to create the dummy context, which was confusing IRGen.
Swift SVN r4476
Now that we don't allow static methods to be invoked from instances we no longer need an AST node to represent an implicit instance-to-metatype conversion. MetatypeExpr encodes the explicit '.metatype' operation.
Swift SVN r4472
Swift doesn't yet type-check function thinness, so variables, arguments, and return values are always thick. When we store, pass as argument, or return a function value that SIL tracks as thin, emit a new 'thin_to_thick_function' instruction to represent converting the type to the thick type and adding a null context pointer.
Swift SVN r4470
