This takes a bit longer since the interpreted part has to do more work
but it saves a round trip through the compiler by allowing `zig2 build`
to be the final step. 1-2-3, done.
For me this is currently failing due to compilation errors generated by
GCC when compiling zig2.c but in theory if those are fixed, it should
work!
* synchronize zig1.c from zig-wasi external project
* change the way argv works to avoid absolute paths
* autodetect isatty
* compiler_rt: disable some functions when object format is C
* add missing flag from config.zig.in
The next problem is that compiling compiler_rt.c with gcc gives
"conflicting types" errors for `__eqhf2` and friends.
This ensures that the Zig version will be re-computed when jumping
through the source tree, which is especially important if bisecting
across AstGen- or other changes that must not use the old cache.
* Old cmake option: `-DZIG_SKIP_INSTALL_LIB_FILES=ON`
* New cmake option: `-DZIG_NO_LIB=ON`
* Old build.zig option: `-Dskip-install-lib-files`
* New build.zig option: `-Dno-lib`
Motivation is making build commands easier to type.
CMake recognizes the CMAKE_PREFIX_PATH environment variable for some
things, and also the CMAKE_PREFIX_PATH cache variable for other things.
However, it does not relate these two things, i.e. if the environment
variable is set, CMake does not populate the cache variable in a
corresponding manner. Some package systems, such as Homebrew, set the
environment variable but not the cache variable. Furthermore, the
environment variable follows the system path separator, such as ':' on
POSIX and ';' on Windows, but the cache variable follows CMake's array
behavior, i.e. always ';' for a separator.
Closes#13242
build.zig: add a 'compile' step to compile the self-hosted compiler
without installing it.
Compilation: set cache mode to whole when using the LLVM backend and
--enable-cache is passed.
This makes `zig build` act the same as it does with stage1. Upside is
that a second invocation of `zig build` on an unmodified source tree
will avoid redoing the compilation again. Downside is that it will
proliferate more garbage in the project-local cache (same as stage1).
This can eventually be fixed when Zig's incremental compilation is more
robust; we can go back to having LLVM use CacheMode.incremental and rely
on it detecting no changes and avoiding doing the flush() step.
* CMakeLists: pass `-Dstrip` for release zig builds
* pass -target and -mcpu to zig1. works around llvm on freebsd
incorrectly detecting "freestanding" instead of "freebsd" for the
native OS.
* ci.ziglang.org is now responsible for creating aarch64-macos tarballs
rather than Azure.
This is a simplification of the cmake build script which introduces a
new "stage3" target that is built by default, which builds and installs
a stage3 zig.
It greatly simplifies the build instructions for Zig, making it conform
to the regular cmake routine, while still producing a stage3 artifact.
Now instead of zig.h being baked into the compiler binary, it is a
header file distributed along with all the other header files
distributed with Zig.
Closes#11643
This commit reworks the LLVM/Clang/LLD discovery process for CMake. The
biggest changes are that:
1. We search for LLVM from most preferred directory to least, skipping
any `llvm-config` that is the wrong version, or that doesn't
support the requested link mode ("static" or "shared").
2. `ZIG_PREFER_CLANG_CPP_DYLIB` has been renamed to `ZIG_SHARED_LLVM`,
to better align with `ZIG_STATIC_LLVM`.
3. We only search for LLVM in the same directory alongside LLVM.
4. LLVM's link mode is forwarded to Clang, so that we can look for the
appropriate shared/static libraries.
5. We use `--link-static` when querying `--system-libs` from llvm-config,
so that this will include libz and other dependencies for
statically linking LLD
CMake has a surprising default behavior where looking up a library by
multiple names gives the name order higher priority than the directory
search order.
For example, if your system provides "llvm-config-14" and
CMAKE_PREFIX_PATH includes "llvm-config", CMake will always end up
choosing the system-provided llvm-config-14.
This change add NAMES_PER_DIR to request the more sensible behavior:
directory search order has higher priority than name order, so
CMAKE_PREFIX_PATH always wins over system-provided tools/libraries.
LLVM 14 deprecated a bunch of C API functions in preparation for opaque
pointer changes. However, they did not actually implement opaque pointer
semantics, so the deprecations are largely masturbatory. I have nothing
against masturbation, I am just busy trying to get the self hosted
compiler done for 0.10.0, so we will come back to this later.
Rather than lowering float negation as `0.0 - x`.
* Add AIR instruction for float negation.
* Add compiler-rt functions for f128, f80 negation
closes#11853
After doing performance testing, it seems that multi-compilation-unit
compiler-rt did not bring the performance improvements that we expected
it to. The idea is that it makes linking faster, however, it incurred a
cost in the frontend that was not offset by any gains in linking.
Furthermore, the single-object compiler-rt (with -ffunction-sections and
--gc-sections) ends up being fewer bytes on disk and so it's actually
the same or faster linking speed than the multi-compilation-unit
version.
So we are planning to keep using single-compilation-unit compiler-rt for
the foreseeable future, but may experiment with this again in the
future, in which case this commit can be reverted.
