Prior to this, the swift build didn’t understand what i686 is, or what to do about building it. This unblocks building, but will still run into build breaks.
This should resolve SR-8826! Ideally, we would be using CMake to tie together
all the dependent packages and get proper dependency tracking. Currently,
change to the swift runtime do not get tracked properly, and can result in
undefined references to symbols.
Update the build-script-impl to only build the armv7 variant by default. This
should prevent the community buildbots from failing which may not have the arm64
ICU available.
The path to the SDK can be computed from the NDK. Do so to reduce the
parameters needed to pass to CMake. More importantly, this will allow building
with multiple architectures simultaneously.
I changed all of the places that used end_borrow_argument to use end_borrow.
NOTE: I discovered in the process of this patch that we are not verifying
guaranteed block arguments completely. I disabled the tests here that show this
bad behavior and am going to re-enable them with more tests in a separate PR.
This has not been a problem since SILGen does not emit any such arguments as
guaranteed today. But once I do the SILGenPattern work this will change.
rdar://33440767
Really we ought to clear out "%t*", but running wildcards with 'rm'
scares me a little too much for that. This still fixes a handful of
tests that were relying on %t being an empty directory.
Note that this doesn't interfere with generated files persisting
between %target-run invocations; they'll be downloaded and then
re-uploaded as long as they're mentioned in the invocation.
The swift-5.0-branch of Clang/LLVM is closest to LLVM 7.0 (and we are
investigating merging up to the real 7.0 branch). Set the default Clang
version to match.
rdar://problem/43163910
This does not eliminate the entrypoints on SILBuilder yet. I want to do this in
two parts so that it is functionally easier to disentangle changing the APIs
above SILBuilder and changing the underlying instruction itself.
rdar://33440767
* Use 'parseAbstractFunctionBody()' for accessors as well. This
simplifies the implementation, and makes 'parseAbstractFunctionBody()'
the single point of parsing body of every 'AbstructFunctionDecl' types.
Most of this is just "remember to specify the inputs and outputs on
the command line, so remote-run can see them". A bit is "prefix
environment variables with '%env-'". And the last few are "yeah,
this was never going to work in a remote environment".
In the few cases where I couldn't think of anything reasonable, I just
marked the test as "UNSUPPORTED: remote_run", a new "feature".
The methods were never executed because DEBUG was never defined in
normal builds and the only way to create nodes is through generated
factory methods which provide the same safety `validate` was supposed to
ensure at the interface level.
