In order to do this, we need to save and restore parser state easily. The
important pieces of state are:
* lexer position;
* lexical scope stack.
Lexer position can be saved/restored easily. We don't need to store the tokens
for the function body because swift does not have a preprocessor and we can
easily re-lex everything we need. We just store the lexer state for the
beginning and the end of the body.
To save the lexical scope stack, we had to change the underlying data
structure. Originally, the parser used the ScopedHashTable, which supports
only a stack of scopes. But we need a *tree* of scopes. I implemented
TreeScopedHashTable based on ScopedHashTable. It has an optimization for
pushing/popping scopes in a stack fashion -- these scopes will not be allocated
on the heap. While ‘detached’ scopes that we want to re-enter later, and all
their parent scopes, are moved to the heap.
In parseIntoTranslationUnit() we do a second pass over the 'structural AST'
that does not contain function bodies to actually parse them from saved token
ranges.
Swift SVN r5886
Because of '~=' lookahead and precedence parsing, we need to be able to parse pattern productions in expression position and validate them after name binding. Add an unresolved Expr node that can hold a subpattern for this purpose.
Swift SVN r5825
Introduce Pattern subclasses for the 'is T', 'T(<pattern>)', and '<expr>' pattern syntaxes we'll be introducing for pattern-matching "switch" statements. Also add an 'UnresolvedCalLPattern' to act as an intermediate for name lookup to resolve to a nominal type, oneof element, or function call expression pattern. Since we'll need to be able to rewrite patterns like we do expressions, add setters to AST nodes that contain references to subpatterns. Implement some basic walking logic in places we search patterns for var decls, but punt on any more complex type-checking or SILGen derived from these nodes until we actually use them.
Swift SVN r5780
Sub-patterns are now considered part of the enclosing pattern, so if the
parent pattern pointer is const, the child pointer will be too.
I changed the minimal number of files to make this work, but future code
should use "const Pattern *" when intended, and "Pattern *" only if they
intend to modify the pattern.
Swift SVN r5743
Improve our representations of casts in the AST and SIL so that 'as!' and 'is' (and eventually 'as?') can share almost all of the same type-checking, SILGen, and IRGen code.
In the AST, we now represent 'as!' and 'is' as UnconditionalCheckedCastExpr and IsaExpr, respectively, with the semantic variations of cast (downcast, super-to-archetype, archetype-to-concrete, etc.) discriminated by an enum field. This keeps the user-visible syntactic and type behavior differences of the two forms cleanly separated for AST consumers.
At the SIL level, we transpose the representation so that the different cast semantics get their own instructions and the conditional/unconditional cast behavior is indicated by an enum, making it easy for IRGen to discriminate the different code paths for the different semantics. We also add an 'IsNonnull' instruction to cover the conditional-cast-result-to-boolean conversion common to all the forms of 'is'.
The upshot of all this is that 'x is T' now works for all the new archetype and existential cast forms supported by 'as!'.
Swift SVN r5737
Open us 'a as! T' to allow dynamic casts from archetypes to archetypes, archetypes to concrete types, existentials to archetypes, and existentials to concrete types. When the type-checker finds these cases, generate new Unchecked*To*Expr node types for each case.
We don't yet check whether the target type actually makes sense with the constraints of the archetype or existential, nor do we implement the SILGen/IRGen backends for these operations. We also don't extend 'x is T' to query the new operation kinds. There's a better factoring that would allow 'as!' and 'is' to share more code. For now, I want to make sure 'x as! T' continues to work for ObjC APIs when we flip the switch to import protocol types.
Swift SVN r5611
Treat 'as' and 'is' as fixed-precedence binary operators, like we now do '=' and '? ... :'. However, since 'as' and 'is' max-munch-parse a type name on their RHS, we only parse them at the tail end of a SequenceExpr. This not only makes 'a = b as T' work as expected again, but also makes 'a += b as T' work, fixing <rdar://problem/13772819>.
Swift SVN r5514
We can save some source code noise and ASTContext allocation traffic by representing unsequenced assignments and ternaries using AssignExpr/IfExpr with the left and right subnodes nulled out, filling them in during sequence folding.
Swift SVN r5509
Parse '=' as a binary operator with fixed precedence, parsing it into a temporary UnsequencedAssignExpr that gets matched to operands and turned into an AssignExpr during sequence expr folding. This makes '=' behave like library-defined assignment-like binary operators.
This temporarily puts '=' at the wrong precedence relative to 'as' and 'is', until 'as' and 'is' can be integrated into sequence parsing as well.
Swift SVN r5508
Change AssignStmt into AssignExpr; this will make assignment behave more consistently with assignment-like operators, and is a first step toward integrating '=' parsing with SequenceExpr resolution so that '=' can obey precedence rules. This also nicely simplifies the AST representation of c-style ForStmts; the initializer and increment need only be Expr* instead of awkward Expr*/AssignStmt* unions.
This doesn't actually change any user-visible behavior yet; AssignExpr is still only parsed at statement scope, and typeCheckAssignment is still segregrated from the constraint checker at large. (In particular, a PipeClosureExpr containing a single assign expr in its body still doesn't use the assign expr to resolve its own type.) The parsing issue will be addressed by handling '=' during SequenceExpr resolution. typeCheckAssignment can hopefully be reworked to work within the constraint checker too.
Swift SVN r5500
Instead of trying to parse '?' and ':' as separate placeholder exprs and matching them up during binary expr resolution, it's a bit cleaner to parse the entire '? ... :' middle expr of the ternary into a single placeholder node at parse time. Then binary expr resolution only ever has to consider a single sequence element.
Swift SVN r5499
Trailing closure syntax allows one to write a closure following any
other postfix expression, which passes the closure to that postfix
expression as an arguments. For example:
sort(fruits) { |lhs, rhs|
print("Comparing \(lhs) to \(rhs)\n")
return lhs > rhs
}
As a temporary limitation to work around the ambiguity with
if foo { ... } { ... }
we require trailing closures to have an explicit parameter list, e.g.,
if foo { || ... } { ... }
Swift SVN r5210
Because we synthesize AST nodes fairly often, and those synthesized
AST nodes rarely have useful source-location information, we shouldn't
be using the validity of source locations to describe the AST. In the
case of closures, use a bit instead. No functionality change.
Swift SVN r5205
This commit implements closure syntax that places the (optional)
parameter list in pipes within the curly braces of a closure. This
syntax "slides" well from very simple closures with anonymous
arguments, e.g.,
sort(array, {$1 > $0})
to naming the arguments
sort(array, {|x, y| x > y})
to adding a return type and/or parameter types
sort(array, {|x : String, y : String| -> Bool x > y})
and with multiple statements in the body:
sort(array, {|x, y|
print("Comparing \(x) and \(y)\n")
return x > y
})
When the body contains only a single expression, that expression
participates in type inference with its enclosing expression, which
allows one to type-check, e.g.,
map(strings, {|x| x.toUpper()})
without context. If one has multiple statements, however, one will
need to provide additional type information either with context
strings = map(strings, {
return $0.toUpper()
})
or via annotations
map(strings, {|x| -> String
return x.toUpper()
}
because we don't perform inter-statement type inference.
The new closure expressions are only available with the new type
checker, where they completely displace the existing { $0 + $1 }
anonymous closures. 'func' expressions remain unchanged.
The tiny test changes (in SIL output and the constraint-checker test)
are due to the PipeClosureExpr AST storing anonymous closure arguments
($0, $1, etc.) within a pattern in the AST. It's far cleaner to
implement this way.
The testing here is still fairly light. In particular, we need better
testing of parser recovery, name lookup for closures with local types,
more deduction scenarios, and multi-statement closures (which don't
get exercised beyond the unit tests).
Swift SVN r5169
We will handle Swift-function-to-ObjC-block bridging in SILGen as part of general Cocoa-to-Swift type bridging. Temporarily disable building swiftAppKit and tests that exercise block bridging until the new implementation lands.
Swift SVN r5090
There is a special-case hack to allow capture of 'this', which is
implicitly [byref] for structs. At the moment, most of the cases where
this hack is necessary are [auto_closure] parameters (for assertions
and &&/||).
Swift SVN r5047
Make IntegerLiteral, FloatLiteral, and StringLiteral own their own copies of their values so they don't depend on the AST. Remove the now-redundant IntegerValueInst, which only existed to be a non-AST-dependent variant of IntegerLiteral.
Swift SVN r5045
Fixes <rdar://problem/13723781>.
T(x) still has some lingering conversion behavior, so there's a type-checking ambiguity in classes that are constructible from super- or subclasses, like stdlib's File is from VFSObject. I cheesed around this for now by using keywords in the constructor forms that have ambiguities. This issue should go away when we finish making T(x) mean only construction.
Swift SVN r5002
One can now attach an initializer to a member variable. That value
will used as the default initialization for the member variable(s) it
initializes. Fixes <rdar://problem/12597404>.
Swift SVN r4989
Give the ternary a fixed precedence, parse '?' and ':' into SequenceExprs, and fold them into IfExprs as part of sequence folding. This allows assignment operators like '+=' to have precedence below the ternary as in C. Fixes <rdar://problem/13756211>.
Swift SVN r4983
Add assignment statements into the implicitly-defined default
constructor body to initialize all of the members appropriately, e.g.,
by calling the default constructor. For builtin types and class types,
introduce ZeroValueInitExpr to produce a "zero" value.
ZeroValueInitExpr still needs a representation in SIL. Until then,
actual generation of this AST is suppressed.
Swift SVN r4895
Add an IsaInst to represent type tests, and implement SILGen for IsSubtypeExpr AST nodes. Get rid of SuperIsArchetypeExpr because it's not really necessary to have it different from IsaSubtype--the SIL and IR behavior is identical.
Swift SVN r4855
We use three tag bits on Expr*, Stmt*, Decl*, TypeBase* and SILTypeInfo*, and four on DeclContext*, so set the alignment of the pointed-to types formally with alignas(N) instead of relying on operator new passing down the right alignment to the allocator. Get rid of the informal T::Alignment members of these classes and pass alignof(T) to their allocators. Fix the 'operator new' of DeclContext subclasses so that we can actually use the four tag bits PointerLikeTypeTraits<DeclContext*> claims are available.
Swift SVN r4587
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
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
Implement the syntax 'if x then y else z', which evaluates to 'y' if 'x' is true or 'z' if 'x' is false. 'x' must be a valid logic value, and 'y' and 'z' must be implicitly convertible to a common type.
Swift SVN r4407
Pack the uncurry level onto SILConstant, and modify SILConstant constructors to determine the natural uncurry level of their referenced entity by default. While we're here, improve how SILConstant represents different kinds of constants by using a real enum for the kind. Type closure entry points by giving them an extra curry level for their context. The implementation of ApplyInst and ClosureInst hasn't been updated yet, so tests that lower closures to SIL are temporarily broken.
Swift SVN r4354
We use uniqued "locators" to describe how a particular constraint was
derived, starting from the constraints generated on expressions (e.g.,
the argument to a function application must be convertible to the
input type of the function being applied) and narrowing down to
specific parts of the types (e.g., the result type of tuple element
2).
Note that locators are not present in all constraints yet, nor are
they actually used for anything other than slightly-more-interesting
debug dumps.
Swift SVN r4353
'super.constructor' shouldn't be referenceable without being called, and 'super.constructor(...)' shouldn't return a value. Require super.constructor expressions to be called at parse time, and wrap the call expression in a new RebindThisInConstructorExpr that represents consuming the delegated-to constructor by using it to reassign 'this'. This should theoretically allow super.constructor to an ill-behaved self-modifying ObjC class to work. It's also necessary to support delegating constructors of value types.
Swift SVN r4326
Replace the more specific Super*RefExpr nodes with a single SuperRefExpr that resolves members of 'this' relative to its superclass. Add an OtherConstructorDeclRefExpr for referring to a constructor as called from another constructor, and use it to represent resolved 'super.constructor' expressions. (It should also be able to represent delegating constructors for free, if we decide we want to add syntax for that.)
Swift SVN r4286
This node represents a type parameter list application in an unresolved expr context. The type checker will use these to explicitly bind type variables of generic types.
Swift SVN r4046
Add support to the constraint checker for typechecking UnresolvedSuperMemberRef expressions and constructing SuperMemberRef or SuperCall expressions as appropriate. We’ll also need a GenericSuperMemberRefExpr to refer to properties of generic supertypes, but in the interests of demo expedience I’m leaving that case partially-implemented for now.
Swift SVN r4020
Analyze an expression of the form [<tuple contents>] into a call to T.convertFromArrayLiteral(<tuple contents>) for some T conforming to an ArrayLiteralConvertible protocol. Because of some limitations in the constraint checker and protocol conformance checkers, it currently does an ad-hoc conformance check using member constraints. It also currently fails to typecheck for generic container types, and does not provide a default fallback to 'T[]' if unable to deduce a better type from context.
Swift SVN r3953
Provide a BridgeToBlockExpr AST node as a temporary representation of func-to-block conversions. In Sema, when we see an [objc_block] type, insert a BridgeToBlock node and coerce the subexpression to the non-block func type.
Swift SVN r3897
Recognize super.constructor calls in IRGen and generate them as calls to the superclass initializing constructor. Note that super.constructor code still won't execute just yet because classic IRGen doesn't generate initializing constructors.
Swift SVN r3859
Set up AST nodes for 'super.<identifier>', 'super.constructor', and 'super[<expr>]' expressions, and implement parsing for them without any sema or backend support.
Swift SVN r3847
We have no intention of ever supporting actual semicolon statements
(separators, statements no), nor do we ever want to because that would
mean the behavior of the program would potentially change if semicolons
were naively removed.
This patch tracks the trailing semicolon now in the decl/expr/stmt,
which will enable someone to write a good "swift indent" tool in the
future.
Swift SVN r3824
