inform conversion ranking about auto_closure, and rip out the
CR_AutoClosure implicit conversion. Our conversion ranking is
now completely trivial: we either have an identity conversion
or an invalid one - no crazy implicit conversions (still have
tuple shuffles etc though).
Swift SVN r1192
- introduce the concept of qualifiers on l-value types
- teach overload resolution and coercion how to drop explicitness
- require explicitness on normal [byref] arguments
- make 'this' [byref(implicit)]
- special-case '&' as a unary operator in the parser to make it
produce an expression which type-checks as turning implicit l-values
into explicit ones.
Obvious missing pieces:
- updating LangRef
- we should really complain about ever trying to rvalue-convert an
explicit l-value
- maybe qualification should play a part in overload resolution
- we should diagnose attempts to declare unary '&' operators
- there's a test case in expressions.swift which suggests my logic is
slightly off
But I am out of time, and these will have to wait.
Swift SVN r1119
instead .
1,2d
1i
Change the 'walk' APIs to use virtual dispatch instead of
blocks. This makes it much easier to thread information
through the API, as well to extend the API to handle
recursive visiting of other structures (e.g., Decls
and Patterns).
Swift SVN r1101
A function argument clause is now one or more patterns (which
must be parenthesized and explicitly type all positions) not
separated by arrows; the first arrow then separates off the
return type.
Revisions to language reference forthcoming.
Swift SVN r1099
in a FuncExpr. This coincidentally means that we can
create the FuncExpr before we create the ArgDecls and
thus avoid the need to reparent them all.
Swift SVN r1081
inconsistent (so it needs to be killed) but that we want to retain source
range information and not kill the entire subtree that contains it.
Swift SVN r1073
some redundancy in the code that always has to create an overload set
or a single declrefexpr depending on what various lookups produce.
Swift SVN r1044
a tuple with one element and no labels. This form is
treated specially in essentially every case, so it might
as well be its own expression kind.
Swift SVN r1021
anonymous member is actually one of these.
A func decl is curried over all of its parameter clauses,
as long as they're written without parentheses. So the body
of a func is the body of the "most curried" function:
func foo(x:int) -> (y:int) -> int {
// This is the body of the function that takes 'y'.
// It returns an int.
}
func bar(x:int) -> ((y:int) -> int) {
// This is the body of the function that takes 'x'.
// It returns a functon of type (y:int) -> int.
}
Swift SVN r993
function into all of the expression nodes. Re-implement
Expr::getStartLoc() in terms of this function, and add it's brother
Expr::getEndLoc(), removing the specialized implementations.
Clean up the source ranges of implicitly-created tuple expressions in
the process.
Swift SVN r855
forcing them to int64 all the time. Now the integer_literal_type in the standard library
is only used to resolve ambiguity in expressions. Among other amazing things, we now
correctly accept:
var x8 : int8 = 4
without a cast.
This still isn't doing any bounds checking and isn't checking the inferred type to tell if
it is compatible with integer literals (i.e. has a primitive conversion function from the
LLVM IR type). Enough for now though.
This is progress towards rdar://10357238.
Swift SVN r814
Also use the new getAdvancedLoc() method instead of hacking
on SMLoc directly.
Also fix the warning/note/error methods to forward through ASTContext
instead of being replicated everywhere.
Swift SVN r750
