This test uses %swiftc_driver_plain instead of %swiftc_driver or
%target-build-swift because it's testing what the driver does when
invoked with no -swift-version argument, but that means we have to
manually set up things like a module cache path. This will then fail
if it's the first Driver test run in a clean build directory (because
of where lit.py decides to put temporary files).
rdar://problem/38354843
Syntax nodes are designed as reusable blocks to construct Swift source
code. For this reason, we don't track absolute position in each node;
instead, the absolute position should be calculated on the fly when
needed. We recently found absolute positions are useful to bridge with
sourcekitd, which typically speaks in the language of line, column and
offset. Therefore, this patch tries to add a computed
property on SyntaxNode to get its absolute position.
To compute the absolute position of a SyntaxNode from scratch requires tree traversal.
However, getting the absolute position from these added APIs doesn't necessarily
mean we'll traverse since SyntaxData will actively cache the computed position.
Also, since we recursively compute the absolute position, all the position caches
for a SyntaxNode's previous siblings and ancestors will be populated as well during
the calculation of this SyntaxNode.
Assign the location of a switch statement's subject expression to all of
its case statements.
This improves the debugger's stepping behavior in switch statements.
Stepping into a switch now goes directly to the first matching case
(possibly one with a `where` clause that may or may not match). It's
still possible to set breakpoints within `where` clauses.
rdar://35628672
Port multiple test-cases from `test/Constraints/closures.swift`
to SILGen to verify that closures are inferred as having `Void`
result type, and there is no function conversion thunk emitted
to convert from `() -> T` to `() -> Void` for literal closures.
Without this fix, the type reconstruction logic chokes on
demangled subtree of the kind:
kind=DeclContext
kind=Function
kind=Module, text="swift_ide_test"
kind=PrivateDeclName
kind=Identifier, text="_0830B395847924C73A6666B87EFD2ADF"
kind=Identifier, text="patatino"
Fixes the compiler part of <rdar://problem/38248403>
Otherwise, the plus_zero_* tests will have plus_zero_* as a module name, causing
massive FileCheck problems.
The reason why I am doing it with the main tests is so that I can use it when
syncing branches/etc.
radar://34222540
Emit enum copy/destroy methods only when codegen demands them; they previously got emitted immediately when TypeInfo is instantiated, which led to many functions getting emitted that were never used. Also, make it so that the symbol name includes the full type of the enum instance the outlined functions operate on, so it's more obvious what they'e being used for and they can be ODRed across translation units.
If a type has the same layout as one of the basic integer types, or has a single refcounted pointer representation, we can use prefab value witness tables from the runtime instead of instantiating new ones. This saves quite a bit of code size, particularly in the Apple SDK overlays, where there are lots of swift_newtype wrappers and option set structs.
If we could find a binding for type variable representing closure result
let's also add implicit `Void` type as another pontential binding to attempt
because there is an implicit conversion from `() -> T` to `() -> Void`
so finding `Void` early is going to help avoid function conversions down
the line.
Resolves: rdar://problem/37790062
swift-ci
Michael Gottesman
Michael Ilseman
Huon Wilson
Karoy Lorentey
Pavel Yaskevich
Nathan Hawes
Jordan Rose
Graydon Hoare
Ben Cohen
Slava Pestov
Slava Pestov
Davide Italiano
Huon Wilson
Xi Ge
Vedant Kumar
Graydon Hoare
Harlan
Pavel Yaskevich
Davide Italiano
Adrian Prantl
Joe Groff
Michael Gottesman
Andrew Trick
Robert Widmann
Erik Eckstein
Joe Groff
Sho Ikeda
David Zarzycki