I have observed the standard library tests overflowing the default WASI
stack as of the previous commit. As best as I can tell, this isn't
directly our fault: LLVM is just emitting less efficient code in debug
builds with the new codegen patterns.
After fixing some issues with inline assembly in the C backend, the std
cleanups have the side effect of making these functions compatible with
the backend, allowing it to be used on linux without linking libc.
* introduce LazyPath.cwd_relative variant and use it for --zig-lib-dir. closes#12685
* move overrideZigLibDir and setMainPkgPath to options fields set once
and then never mutated.
* avoid introducing Build/util.zig
* use doc comments for deprecation notices so that they show up in
generated documentation.
* introduce InstallArtifact.Options, accept it as a parameter to
addInstallArtifact, and move override_dest_dir into it. Instead of
configuring the installation via Compile step, configure the
installation via the InstallArtifact step. In retrospect this is
obvious.
* remove calls to pushInstalledFile in InstallArtifact. See #14943
* rewrite InstallArtifact to not incorrectly observe whether a Compile
step has any generated outputs. InstallArtifact is meant to trigger
output generation.
* fix child process evaluation code handling of `-fno-emit-bin`.
* don't store out_h_filename, out_ll_filename, etc., pointlessly. these
are all just simple extensions appended to the root name.
* make emit_directory optional. It's possible to have nothing outputted,
for example, if you're just type-checking.
* avoid passing -femit-foo/-fno-emit-foo when it is the default
* rename ConfigHeader.getTemplate to getOutput
* deprecate addOptionArtifact
* update the random number seed of Options step caching.
* avoid using `inline for` pointlessly
* avoid using `override_Dest_dir` pointlessly
* avoid emitting an executable pointlessly in test cases
Removes forceBuild and forceEmit. Let's consider these additions separately.
Nearly all of the usage sites were suspicious.
This reverts commit c75e11bf6aa67f2ca62b9b6677d134592777bfec.
Caused this problem on some machines:
"this step declares an upper bound of 9126805504 bytes of memory,
exceeding the available 7515721728 bytes of memory".
Instead the next commit will disable std lib tests with the C backend on
Windows.
I observed clang taking 8G to compile the output from the std lib tests
using the C backend. This commit should make the Windows CI stop failing
due to OOM.
The CI now runs C backend tests in addition to compiling them. It uses
-std=c99 -pedantic -Werror in order to catch non-conformant C code.
This necessitated disabling a test case that caused a C compile error,
in addition to disabling a handful of warnings that are already being
triggered by Zig's C backend output for the behavior tests.
The upshot is that I was able to, very cleanly, integrate the C backend
tests into the build system, so that it communicates via the test runner
protocol along with all the other behavior tests.
Make the test targets use options that match the actual options of
CompileStep. This makes the code more straightforward, and ends up
making fewer tests incorrectly skipped. For example, now the CI runner
on Windows will no longer skip self-hosted x86_64 backend tests.
* remove setName, setFilter, and setTestRunner. Please set these
options directly when creating the CompileStep.
* removed unused field
* remove computeOutFileNames and inline the logic, making clear the
goal of avoiding state mutations after the build step is created.
These functions are problematic in light of dependencies because they
run and install, respectively, for the *owner* package rather than for
the *user* package. By removing these functions, the build script is
forced to provide the *Build object to associate the new step with,
making everything less surprising.
Unfortunately, this is a widely breaking change.
see #15079
It is reasonable to pass -Dskip-non-native when unable to run foreign
binaries, however there is no option for being able to run foreign
static binaries but unable to run foreign dynamic binaries. This can
occur when qemu is installed but not cross glibc.
std.Build.addTest creates a CompileStep as before, however, this kind of
step no longer actually runs the unit tests. Instead it only compiles
it, and one must additionally create a RunStep from the CompileStep in
order to actually run the tests.
RunStep gains integration with the default test runner, which now
supports the standard --listen=- argument in order to communicate over
stdin and stdout. It also reports test statistics; how many passed,
failed, and leaked, as well as directly associating the relevant stderr
with the particular test name that failed.
This separation of CompileStep and RunStep means that
`CompileStep.Kind.test_exe` is no longer needed, and therefore has been
removed in this commit.
* build runner: show unit test statistics in build summary
* added Step.writeManifest since many steps want to treat it as a
warning and emit the same message if it fails.
* RunStep: fixed error message that prints the failed command printing
the original argv and not the adjusted argv in case an interpreter
was used.
* RunStep: fixed not passing the command line arguments to the
interpreter.
* move src/Server.zig to std.zig.Server so that the default test runner
can use it.
* the simpler test runner function which is used by work-in-progress
backends now no longer prints to stderr, which is necessary in order
for the build runner to not print the stderr as a warning message.
Instead of using `zig test` to build a special version of the compiler
that runs all the test-cases, the zig build system is now used as much
as possible - all with the basic steps found in the standard library.
For incremental compilation tests (the ones that look like foo.0.zig,
foo.1.zig, foo.2.zig, etc.), a special version of the compiler is
compiled into a utility executable called "check-case" which checks
exactly one sequence of incremental updates in an independent
subprocess. Previously, all incremental and non-incremental test cases
were done in the same test runner process.
The compile error checking code is now simpler, but also a bit
rudimentary, and so it additionally makes sure that the actual compile
errors do not include *extra* messages, and it makes sure that the
actual compile errors output in the same order as expected. It is also
based on the "ends-with" property of each line rather than the previous
logic, which frankly I didn't want to touch with a ten-meter pole. The
compile error test cases have been updated to pass in light of these
differences.
Previously, 'error' mode with 0 compile errors was used to shoehorn in a
different kind of test-case - one that only checks if a piece of code
compiles without errors. Now there is a 'compile' mode of test-cases,
and 'error' must be only used when there are greater than 0 errors.
link test cases are updated to omit the target object format argument
when calling checkObject since that is no longer needed.
The test/stage2 directory is removed; the 2 files within are moved to be
directly in the test/ directory.
