This routine provides an AST-level operation that takes a base type
and a member, then computes the type the member will have when
accessed through that base, substituting (e.g.) generic arguments for
the archetypes in the type of the member. It is a cleaned up, slightly
generalized, and slightly more correct formulation of
TypeChecker::substMemberTypeWithBase(), which is now just a shim that
invokes TypeBase::getTypeOfMember().
Addresses <rdar://problem/14778376>.
Swift SVN r8198
We often won't need the complete substitutions, so only compute them
when needed. This also means that we don't need to serialize them into
module files.
Swift SVN r8194
Teach a BoundGenericType to compute its own substitutions, which
allows AST clients to create new bound generic types without the aid
of the type checker.
This eliminates the TypeChecker::validateTypeSimple() abomination as
well as the need for the BoundGenericType AST validation step. There
is still more cleanup to do in this area.
Note that BoundGenericType::getSubstitutions() now accepts a module
parameter, which is the place from which we will look for
conformances. This is a baby step toward properly modeling the
conformances as part of the bound generic type, and is nowhere near
complete.
Swift SVN r8193
No testcase because this is not testable now -- this function is only used in
diagnostics of value redefinitions, which are broken now for redefinitions of
member functions.
Swift SVN r8187
and remove DeclContext base class from FuncDecl, ConstructorDecl and
DestructorDecl
This decreases the number of DeclContexts to 7 and allows us to apply
alignas(8) to DeclContext.
Swift SVN r8186
This eliminates the need for the hackish
LazyResolver::resolveUnvalidatedType() entry point, making type
substitution on ASTs far more useful. As part of this, teach it to
handle conformances for existentials and archetypes.
Swift SVN r8169
This allows AST clients to perform type substitutions without
requiring a type checker. Test this by performing substitutions within
the AST verifier.
Swift SVN r8166
Introduce an AST operation that, given a type and a protocol, determines
whether the type conforms to the protocol and produces the protocol
conformance structure. Previously, this operation was only available
on the type checker, requiring many callbacks from the AST to the type
checker during AST substitution operations (for example).
Now, we only call back into the type checker when we hit a case where
we see an explicit conformance in the AST, but the actual
ProtocolConformance object has not yet been built due to lazy type
checking.
Note that we still require a resolver (i.e., a TypeChecker) in a few
places, although we shouldn't need it outside of lazy type
checking. I'll loosen up the restrictions next.
There's a minor diagnostics regression here that will be cleaned up in
a future commit.
Swift SVN r8129
This routine performs the substitutions necessary to get the actual
superclass for (e.g.) any class or substitutable type, including bound
generic classes and inner classes thereof. This routine still requires
a lazy resolver (i.e. a type checker).
Swift SVN r8101
The type substitution still depends fairly heavily on the type
checker, which is accessed via the LazyResolver abstract
interface. That interface should get simpler, and become unnecessary
except to deal with lazy type checking, over time. At that point, it
will become optional, and Type::subst() will be usable without a type
checker around.
Swift SVN r8099
Fixes two bugs in Clang importer and deserialization code that were found by
the verifier:
(1) FuncExprs were created with a null FuncDecl
(2) BoundGenericType that was created by Clang importer for UnsafePtr<> and
other magic types did not have substitutions.
Swift SVN r8073
If an OpaqueValueExpr is only used in one place, mark it as such. SIL
generation will then elide the retain/release pair associated with
each reference to the opaque value, instead consuming the value at the
point of use.
Swift SVN r8072
These helper expressions will eventually be used by SILGen to help
package up the optional values. I expect that we'll eventually have
library builtins for this, so consider this a stop-gap until those
appear.
As part of this, make OpaqueValueExpr a bit more usable: it can now
persist in the AST as a placeholder, but its uses must be within AST
subtrees of some specific introduction point (similarly to how Clang's
OpaqueValueExpr works).
Swift SVN r8051
ConstructorDecl::getBody() and DestructorDecl::getBody() return 'BraceStmt *'.
After changing the AST representation for functions, FuncDecl::getBody() will
return 'BraceStmt *' and FuncDecl::getFuncExpr() will be gone.
Swift SVN r8050
global variables used by functions in the capture list as well.
SILGen and other things that don't care about these (i.e., all
current current clients) filter the list to get what they want.
This is needed for future definite init improvements, and unblocked
by Doug's patch in r8039 (thanks! :)
No functionality change.
Swift SVN r8045
Diagnostics that point to the first bad token are marked with this option in
Diagnostics.def. Parser::diagnose treats the source location of such
diagnostics in a special way: if source location points to a token at the
beginning of the line, then it moves the diagnostic to the end of the previous
token.
This behaviour improves experience for "expected token X" diagnostics.
Swift SVN r7965
We want to enforce the invariant that invalid ValueDecls have type ErrorType.
This patch adds that check, and fixes up the last few places which violated
it.
Last part of <rdar://problem/10157547>.
Swift SVN r7960
getDisplayDecls() was introduced for ":print_module" and works slightly differently, e.g.
it will return the decls from a shadowed clang module, since we want to display them.
Swift SVN r7909
Essentially, fall back to the list of protocols (and possibly the superclass)
if the "inherited types" array is empty.
<rdar://problem/14788337>
Swift SVN r7884
