…k to thin and thick to thick representations on subtype context

…rtain representation convertions do not crash

Co-authored-by: Xiaodi Wu <13952+xwu@users.noreply.github.com>

…on map when computing necessary bindings of abstract cnditional requirements

SR-12853

This changes the hash from the Python `hash` function to SHA1.  Python 3
adopted SIP as the hashing algorithm which changed the hash results.  We
could no longer use a different python interpreter across the builds.  A
stable hashing algorithm allows easier migration.

…the requirement

…of a global variable

It is too slow, rdar://64215520

…n type has error type

…-build is enabled.

Explicitly use the path to pyhon2.7 (although the tool can support python 3.x, the path is much much unwieldy as `"C:\Program Files (x86)\Microsoft Visual Studio\Shared\Python37_64\python.exe"`

…S stdlib.

rdar://problem/64301166

Private and internal classes shouldn't have ABI constraints on their concrete vtable layout, so if methods
don't have overrides in practice, we can elide their vtable entries.

…ctionType

…#32332)

Adds support for three more basic infix operators: `%`, `&`, and `|`.

Co-authored-by: Michael Forster <forster@google.com>

…lude-os-stdlib

…ype-diag

[SR-12955] [Sema] Don't diagnose invalid conversion if fnType has error

Update WindowsBuild.md

build: remove unnecessary variable (NFC)

…table-entries

IRGen: Don't reify internal vtable entries that are marked non-overridden.

gyb: use a stable hash for Syntax

When running build-script with --swift-inspect, the Package.resolved
file is created in tools/swift-inspect.  Ignore it.

…276374405140f3

[build-script] llvm-dsymutil's component is dsymutil.

…5528e7ec1c27c2

[build-script] Print out inferred products/build order when --verbose-build is enabled.

to match the one specified in LLVM's Mach-O parser.
Otherwise LLDB could not possibly find it!

This name is used by the swift -modulewrap subcommand, which is currently unused
on Darwin, and primarily intended for use under Linux. However, it may be useful
to better support static archives (.a) files with Swift debug info in the
future. To fully support this, dsymutil and LLDB need to know to look for Swift
AST sections in Mach-O objects other than .dSYM bundled.

Implementation note: It would be nice to get the section name out of libObject,
but with the current architecture this needs a major refactoring that didn't
seem justified, given that there is an end-to-end test to prevent such a mishap
in the future.

<rdar://problem/63991514>

Swift calls the architecture x86_64, OpenBSD calls it amd64. If we use
run_cpu in lit.cfg as-is, then we may need to duplicate lines in each
test for 'x86_64' and 'amd64', which puts a maintenance burden on unit
test developers to ensure they are duplicating changes to each line.

Instead, alias 'amd64' to 'x86_64' for `run_cpu`, but keep the platform
module path referring to 'amd64', in order to distinguish the target
architecture name and the Swift architecture name. This is particularly
relevant for the %target-.*-name pseudovariables used, which should
reference the Swift architecture names.

However, some unit tests are directly referencing %target-cpu directly,
which would break the aliasing. This is done only for swiftinterface
files, so a new substitution is defined in lit.cfg for these variables,
and the affected unit test cases are migrated.

…e the index expression isn't a tuple or paren expression (#32356)

[test/Syntax/round_trip_stdlib.swift] Exclude the test for an ASAN build

vim: improve syntax highlighting rules

but Solution node->type map is all or nothing, so can be cheaper.

During witness table instantiation, the witness table is constructed
several sources: the pattern, the resilient witnesses, the private
data, and default implementations. The private data area is the only
one that was being zeroed out; the rest we rely on always filling in
the data from the conformance descriptor and provided info.

However, witness table instantiation uses the presence of a NULL
pointer for a particular witness in the resulting table to indicate
that no witness fulfilled that requirement, so that it can fill in the
default witnesss. Except that, without zeroing that part of the table
beforehand, we aren't guaranteed to have a NULL pointer for witness
entries that the client (protocol conformance) did not know about at
the time it was compiled.

Zero out the entire witness table so default implementations can be
filled in appropriately.

Fixes rdar://problem/64295849.

The CI is switching to ICU 67 since there are fixes in the latter version of ICU that are needed.  This should be further improved in the future to possibly use the the unicode.org releases.

This reverts commit 499ed05.

This changes the num-threads passed to 1 to avoid a multithreaded output
issue on windows

Allow user control over libxml2 usage via `LLVM_ENABLE_LIBXML2` rather
than assuming that if it is found, it should be enabled.

…-default-zeroing

[Runtime] Zero out the entire witness table during instantiation.

[Sema] Solution nodeTypes from MapVector to DenseMap.

…epresentation conversions.

Update WindowsBuild.md

build: do not assume that XML2 should be enabled

…npect-products

[Gardening] Ignored Package.resolved for swift-inspect.

Ignore onFastPath builtin while looking for valid static initializer of a global variable

…ditionalRequirements

IRGen: Lookup the conformance of an archetype in the right substitution map when computing necessary bindings of abstract cnditional requirements

…-for-all-modules

reland: Emit coverage mappings for all modules

[test] Handle architecture aliasing for OpenBSD.

Call prepareExtensions to make sure we load in any
new extensions brought in by modules loaded
after-the-fact. This isn't an issue for normal
compilations as we load all the modules up-front
in import resolution, but clients such as the
LLDB REPL may load in modules later.

Resolves rdar://64040436.

Fix the hardcoded Swift AST section / segment name for Mach-O

[SourceKit][SR-12837] Add effective access level of references

…se (#32364)

[SR-12723][Sema] Validate function type param representations thick-to-thin conversions

Disable tests that occasionally flake on Linux.

SR-13021 tracks re-enabling these tests.

Now that we use the LLVM mono-repo, we don't need to worry about clang's
version number. Also, git has the ability to estimate the safe number of
digits a hash can be truncated to and now git estimates that large
projects like LLVM and Linux "require" 12 digits for safe commit hash
abbreviation. Let's stay a little ahead of the curve and statically
truncate to 15.

The rule-of-zero is your friend. Also, switch to `class` and make the
`value` private while we're touching these types.

[Basic] Revise version printing