llvm::Optional lives in "llvm/ADT/Optional.h". Like Clang, we can get
Optional in the 'swift' namespace by including "swift/Basic/LLVM.h".
We're now fully switched over to llvm::Optional!
Swift SVN r22477
A shortcut to skip validation if the type is set does not work since the
underlying type might not be set.
No tests since I can not reproduce outside of Xcode.
rdar://18502983
Swift SVN r22428
When trying to implement deduplication of results, found and fixed an issue
with loose checks for generic overload checking.
rdar://17995317
Swift SVN r21276
declarations exhaustively, including protocol conformances.
Refactor and simplify copyAssociatedUSRs() to copy each USR one by one.
Manually copy the string array into the sink allocator, to work around
a clang-600 merge failure.
Swift SVN r20910
If the completion result is a declaration, it contains the USR of the specified declaration and all overridden declarations.
Additional tests on SourceKit side.
This addresses <rdar://problem/17600891>.
Swift SVN r20877
This includes proper printing support as well as proper platform checking
when seeing if a decl is unavailable. A few other places in the code will
now use AvailabilityAttr::isUnavailable instead of just checking the
is-unavailable-always flag (and not always checking the platform).
No new tests yet because this doesn't include /parsing/ the other fields
of AvailabilityAttr. That will come next, at which point we'll test each
of the cases that has been switched over to use
AvailabilityAttr::isUnavailable.
Part of <rdar://problem/17024498>
Swift SVN r20844
The former is for debugging, the latter is for detailed presentation to users.
swift -print-ast will continue using printEverything, as will swift-ide-test,
but all other features should use printVerbose.
Swift SVN r20432
Expose Substitution's archetype, replacement, and conformances only through getters so we can actually assert invariants about them. To start, require replacement types to be materializable in order to catch cases where the type-checker tries to bind type variables to lvalue or inout types, and require the conformance array to match the number of protocol conformances required by the archetype. This exposes some latent bugs in the test suite I've marked as failures for now:
- test/Constraints/overload.swift was quietly suffering from <rdar://problem/17507421>, but we didn't notice because we never tried to codegen it.
- test/SIL/Parser/array_roundtrip.swift doesn't correctly roundtrip substitutions, which I filed as <rdar://problem/17781140>.
Swift SVN r20418
optional & alias types for closure type parameters.
This allows code completion placeholder expansion to properly expand
closure parameters utilizing a typealias, e.g. dispatch_block_t.
Update and add test for the above.
Work for <rdar://problem/15860693>.
Swift SVN r20206
Completely disable the AST transformation for single-expression closures. When
this transformation picks up an incomplete expression, the resulting AST is
almost guaranteed to have type mismatches, and the type checker just marks
everything with error types.
rdar://17193319 rdar://17086137
Swift SVN r20153
attribute is a "modifier" of a decl, not an "attribute" and thus shouldn't
be spelt with an @ sign. Teach the parser to parse "@foo" but reject it with
a nice diagnostic and a fixit if "foo" is a decl modifier.
Move 'dynamic' over to this (since it simplifies some code), and switch the
@optional and @required attributes to be declmodifiers (eliminating their @'s).
Swift SVN r19787
Introduce the new BooleanLiteralConvertible protocol for Boolean
literals. Take "true" and "false" as real keywords (which is most of the
reason for the testsuite churn). Make Bool BooleanLiteralConvertible
and the default Boolean literal type, and ObjCBool
BooleanLiteralConvertible. Fixes <rdar://problem/17405310> and the
recent regression that made ObjCBool not work with true/false.
Swift SVN r19728
We want to complete foo(p: (Int, Int)) as foo({#(p): (Int, Int)#}), not
as foo({#Int#}, {#Int#}). Also, make initializers go through the
function parameter code path rather than the enum one, since we don't
want to recursively find parameters in tuples.
<rdar://problem/17461862>
Swift SVN r19514
Leading paren in constructor patterns should be included in the description
text, but not in the name.
This addresses the first part of <rdar://problem/17004638>.
Swift SVN r19114
LLVM's system_error.h has been changed to forward to the standard
version of the same. Update usage for the minor API changes that this
entails.
Based in part on a patch by Justin Bogner.
Swift SVN r18832
-Hide vars that have a private type.
-Hide functions that have a parameter with private type or a parameter name with leading underscore.
-Minor change in StringUTF16.swift to avoid printing "func generate() -> IndexingGenerator<_StringCore>".
rdar://17027294
Swift SVN r18623
instead, so it picks up substituted types. BodyParamPatterns is now only
used for the local non-api parameter name.
Consolidate the two variations of addFunctionCallPattern() into one function.
Update the tests accordingly.
This addresses <rdar://problem/16956363>.
Swift SVN r18481
This reverts commit r18426.
Per feedback from Jordan this does the wrong thing because it aborts the entire visitation, not just the current visitation path.
Swift SVN r18433
if there's no parameter API name. This is for display purposes only.
Update all relevant tests accordingly.
This addresses <rdar://problem/16768768>.
For example:
class X {
func f(a: Int, b: Int) { }
}
Would previously display like this in code completion in Xcode:
f(<#Int#>, b: <#Int#>)
The local parameter name, while not API, often still conveys meaning
to the user. So it's now included like this:
f(<#a: Int#>, b: <#Int#>)
Swift SVN r18403
We already don't try to detect and filter out shadowed declarations due to performance reasons, so cache the visible declarations contained
in a specific ImportedModule and collect all the results by recursing via calling Module::forAllVisibleModules().
This is more efficient because we avoid doing the work to collect all the global results for both Cocoa and AppKit, and we save on memory usage for the cache.
This also fixes a correctness issue where if you imported a module that was transitively imported by a previous import (e.g. Cocoa and AppKit) you would get duplicate results.
For this test case:
----------
import Cocoa
import AppKit
import Foundation
func foo() {
<ESC>
}
----------
We go from
- 7.05 secs to 1.43 secs to collect the results
- 9.0 secs to 2.4 to ultimately show them in Xcode for first-time invocation.
Swift SVN r18344
A use-after-free could happen in the following scenario:
- code completion caches results from module A;
- code completion returns cached results -- result set 1;
- module is rebuilt;
- another code completion request arrives, cache is invalidated and re-filled.
Old results are freed, even though someone could be still using the result
set.
rdar://16953614
Swift SVN r18329
The leading paren is included for display purposes only, not inserted
in the code if already present. It makes the displayed text in the
code completion list symmetrical with respect to open/close parens.
Add markups when printing annotation chunks so it becomes testable.
Update tests accordingly, and include tests for leading parens.
This addresses <rdar://problem/16918310>.
Swift SVN r18126
