Previously, global assembly was parsed expecting it to have
the template syntax. However global assembly has no inputs,
outputs, or clobbers, and thus does not have template syntax.
This is now fixed.
This commit also adds a compile error for using volatile
on global assembly, since it is meaningless.
closes#1515
Before, allocator implementations had to provide `allocFn`,
`reallocFn`, and `freeFn`.
Now, they must provide only `reallocFn` and `shrinkFn`.
Reallocating from a zero length slice is allocation, and
shrinking to a zero length slice is freeing.
When the new memory size is less than or equal to the
previous allocation size, `reallocFn` now has the option
to return `error.OutOfMemory` to indicate that the allocator
would not be able to take advantage of the new size.
For more details see #1306. This commit closes#1306.
This commit paves the way to solving #2009.
This commit also introduces a memory leak to all coroutines.
There is an issue where a coroutine calls the function and it
frees its own stack frame, but then the return value of `shrinkFn`
is a slice, which is implemented as an sret struct. Writing to
the return pointer causes invalid memory write. We could work
around it by having a global helper function which has a void
return type and calling that instead. But instead this hack will
suffice until I rework coroutines to be non-allocating. Basically
coroutines are not supported right now until they are reworked as
in #1194.
`--static` is no longer an option. Instead, Zig makes things as static
as possible by default. `-dynamic` can be used to choose a dynamic
library rather than a static one.
`--enable-pic` is a new option. Usually it will be enabled
automatically, but in the case of build-exe with no dynamic libraries
on Linux or freestanding, Zig chooses off by default.
closes#1703closes#1828
* in Zig build scripts, getOutputPath() is no longer a valid function
to call, unless setOutputDir() was used, or within a custom make()
function. Instead there is more convenient API to use which takes
advantage of the caching system. Search this commit diff for
`exe.run()` for an example.
* Zig build by default enables caching. All build artifacts will go
into zig-cache. If you want to access build artifacts in a convenient
location, it is recommended to add an `install` step. Otherwise
you can use the `run()` API mentioned above to execute programs
directly from their location in the cache. Closes#330.
`addSystemCommand` is available for programs not built with Zig
build.
* Please note that Zig does no cache evicting yet. You may have to
manually delete zig-cache directories periodically to keep disk
usage down. It's planned for this to be a simple Least Recently
Used eviction system eventually.
* `--output`, `--output-lib`, and `--output-h` are removed. Instead,
use `--output-dir` which defaults to the current working directory.
Or take advantage of `--cache on`, which will print the main output
path to stdout, and the other artifacts will be in the same directory
with predictable file names. `--disable-gen-h` is available when
one wants to prevent .h file generation.
* `@cImport` is always independently cached now. Closes#2015.
It always writes the generated Zig code to disk which makes debug
info and compile errors better. No more "TODO: remember C source
location to display here"
* Fix .d file parsing. (Fixes the MacOS CI failure)
* Zig no longer creates "temporary files" other than inside a
zig-cache directory.
This breaks the CLI API that Godbolt uses. The suggested new invocation
can be found in this commit diff, in the changes to `test/cli.zig`.
Unlike the other glibc source code checked into the repo, `csu/init.c`
did not have a license clause that allowed linking without restrictions.
`_IO_stdin_used` is the only symbol in the file and appears to be a 20
year old compatibility shim for the glibc 2.0 ABI. Obsolete in 2.1.
closes#2024
there's a new cli option `--main-pkg-path` which you can use to choose
a different root package directory besides the one inferred from the
root source file
and a corresponding build.zig API:
foo.setMainPkgPath(path)
* CLI: `-target [name]` instead of `--target-*` args.
This matches clang's API.
* `builtin.Environ` renamed to `builtin.Abi`
- likewise `builtin.environ` renamed to `builtin.abi`
* stop hiding the concept of sub-arch. closes#1526
* `zig targets` only shows available targets. closes#438
* include all targets in readme, even those that don't
print with `zig targets` but note they are Tier 4
* refactor target.cpp and make the naming conventions
more consistent
* introduce the concept of a "default C ABI" for a given
OS/Arch combo. As a rule of thumb, if the system compiler
is clang or gcc then the default C ABI is the gnu ABI.
We pass -MD -MF args to clang when doing `@cImport`, which
gives us a complete list of files that the C code read from.
Then we add these to the cache. So even when using `@cImport`
Zig's caching system remains perfect. This is a proof of concept
for the mechanism that the self-hosted compiler will use to
watch and rebuild files.
New CLI parameter: --c-source [options] [file]
It even works with `--cache on` when there are transitive dependencies.
Instead of `builder.addCExecutable`, use `builder.addExecutable` and pass
`null` for the root source file. Then use `builder.addCSourceFile`,
which takes the path to the C code, and a list of C compiler args.
Be sure to linkSystemLibrary("c") if you want libc headers to be
available.
Merge TestStep into LibExeObjStep. That was long overdue.
* better libc detection
This introduces a new command `zig libc` which prints
the various paths of libc files. It outputs them to stdout
in a simple text file format that it is capable of parsing.
You can use `zig libc libc.txt` to validate a file.
These arguments are gone:
--libc-lib-dir [path] directory where libc crt1.o resides
--libc-static-lib-dir [path] directory where libc crtbegin.o resides
--msvc-lib-dir [path] (windows) directory where vcruntime.lib resides
--kernel32-lib-dir [path] (windows) directory where kernel32.lib resides
Instead we have this argument:
--libc [file] Provide a file which specifies libc paths
This is used to pass a libc text file (which can be generated with
`zig libc`). So it is easier to manage multiple cross compilation
environments.
`--cache on` now works when linking against libc.
`ZigTarget` now has a bool field `is_native`
Better error messaging when you try to link against libc or use
`@cImport` but the various paths cannot be found. It should also be
faster.
* save native_libc.txt in zig-cache
This avoids having to detect libc at runtime on every invocation.
* When you do field access of a type which only has one possible
value, the result is comptime-known.
* StorePtr instructions which operate on pointers to types which
only have one possible value, the result is a comptime no-op.
closes#1554
* Separate LoadPtr IR instructions into pass1 and pass2 variants.
* Define `type_size_bits` for extern structs to be the same as
their `@sizeOf(T) * 8` and allow them in packed structs.
* More helpful error messages when trying to use types in
packed structs that are not allowed.
* Support arrays in packed structs even when they are not
byte-aligned.
* Add compile error for using arrays in packed structs when the
padding bits would be problematic. This is necessary since
we do not have packed arrays.
closes#677
* `type_size_store` is no longer a thing. loading and storing a pointer
to a value may dereference up to `@sizeOf(T)` bytes, even for
integers such as `u24`.
* fix `types_have_same_zig_comptime_repr` to not think that the
same `ZigTypeId` means the `ConstExprValue` neccesarily has the
same representation.
* implement `buf_write_value_bytes` and `buf_read_value_bytes` for
`ContainerLayoutPacked`
closes#1120
with this change, when you assign undefined, zig emits a few
assembly instructions to tell valgrind that the memory is undefined
it's on by default for debug builds, and disabled otherwise. only
support for linux, darwin, solaris, mingw on x86_64 is currently
implemented.
--disable-valgrind turns it off even in debug mode.
--enable-valgrind turns it on even in release modes.
It's always disabled for compiler_rt.a and builtin.a.
Adds `@import("builtin").valgrind_support` which lets code know
at comptime whether valgrind client requests are enabled.
See #1989
* also fix extern variables with initialiaztion values to generate runtime code
* remove the workaround in example/shared_library/mathtest.zig
* introduce the ability for global variables to have Weak and LinkOnce
linkage
* fix `@export` to work for non-functions. this code needs to be
audited though.
* fix comptime ptrcast not keeping bigger alignment
* fix linker warnings when targeting darwin
closes#1903
Mostly picking the same paths as FreeBSD.
We need a little special handling for crt files, as netbsd uses its
own (and not GCC's) for those, with slightly different names.
Previously, if a dereference instruction was an lvalue, it would fail to
typecheck that the value being dereferenced was indeed a pointer.
Although a little clunky, this change obviates the need for redundant
type checks scattered about the analysis.
this adds the prototype of panic to @import("builtin")
and then uses it to do an implicit cast of the panic
function to this prototype, rather than redoing all the
implicit cast logic.
closes#1894closes#1895
* Fixes breaches of the guarantee that `@sizeOf(T) >= @alignOf(T)`
* Fixes std.mem.secureZero for integers where this guarantee previously
was breached
* Fixes std.mem.Allocator for integers where this guarantee previously
was breached
Closes#1851Closes#1864
It's still best practice to put `@setEvalBranchQuota` at the top of
the comptime stack, but as Jimmi notes in #1949, when a function
can be called at comptime and also can be the top of the comptime stack,
this compile error is fundamentally unsound.
So now it's gone.
closes#1949
This deletes some legacy cruft, and produces leaner object files.
Example:
```
var x: i32 = 1234;
export fn entry() i32 {
return x;
}
```
This produces:
```
@x = internal unnamed_addr global i32 1234, align 4
@0 = internal unnamed_addr constant i32* @x, align 8
```
and @0 is never even used. After this commit, @0 is not produced.
This fixes a bug: Zig was creating invalid LLVM IR when one of these
globals that shouldn't exist takes the address of a thread local
variable. In LLVM 8.0.0rc2, it would produce a linker error. But
probably after my bug report is solved it will be caught by the IR
verifier.
https://bugs.llvm.org/show_bug.cgi?id=40652
This deletes some legacy cruft, and produces leaner object files.
Example:
```
var x: i32 = 1234;
export fn entry() i32 {
return x;
}
```
This produces:
```
@x = internal unnamed_addr global i32 1234, align 4
@0 = internal unnamed_addr constant i32* @x, align 8
```
and @0 is never even used. After this commit, @0 is not produced.
This fixes a bug: Zig was creating invalid LLVM IR when one of these
globals that shouldn't exist takes the address of a thread local
variable. In LLVM 8.0.0rc2, it would produce a linker error. But
probably after my bug report is solved it will be caught by the IR
verifier.
https://bugs.llvm.org/show_bug.cgi?id=40652
There's an unfortunate footgun in the current design of error sets.
The debug info type for every error set is the same as the debug info
type of the global error set, which is essentially an enum forward
declaration. The problem is that when we "replace" the forward
declaration with the final value, once we know all the possible errors,
we have to update the pointers of every error set.
So the footgun is that if you ever copy the debug info type of the
global error set, you have to add the address of the pointer to a list
of pointers that need to be updated once we "replace" the forward
declaration. I activated the footgun when I introduced the optimization
that `?anyerror` types are the same size as `anyerror` types (using 0 as
the null value), because I introduced a pointer copy of the global error
set debug info type, but forgot to add it to the list.
I'm sure that there is a better way to code this, which does not have
the footgun, but this commit contains only a fix, not a reworking of the
logic.
closes#1937
this would work if @llvm.sadd.with.overflow supported
vectors, which it does in trunk. but it does not support
them in llvm 7 or even in llvm 8 release branch.
so the next commit after this will have to do a different
strategy, but when llvm 9 comes out it may be worth coming
back to this one.
Previously, std.debug.assert would `@panic` in test builds,
if the assertion failed. Now, it's always `unreachable`.
This makes release mode test builds more accurately test
the actual code that will be run.
However this requires tests to call `std.testing.expect`
rather than `std.debug.assert` to make sure output is correct.
Here is the explanation of when to use either one, copied from
the assert doc comments:
Inside a test block, it is best to use the `std.testing` module
rather than assert, because assert may not detect a test failure
in ReleaseFast and ReleaseSafe mode. Outside of a test block, assert
is the correct function to use.
closes#1304
this should actually improve CI times a bit too
See the description at the top of std/os/startup.zig (deleted in this
commit) for a more detailed understanding of what this commit does.
closes#1764
This adds another boolean to the test matrix; hopefully it does not
inflate the time too much.
std.event.Loop does not work with this option yet. See #1908
Previously, it pointed to the start of the current token, but this made
it difficult to tell where the error occurred when it was, say, in the
middle of a string.
This commit contains everything from the copy-elision-2
branch that does not have to do with copy elision directly,
but is generally useful for master branch.
* All const values know their parents, when applicable, not
just structs and unions.
* Null pointers in const values are represented explicitly,
rather than as a HardCodedAddr value of 0.
* Rename "maybe" to "optional" in various code locations.
* Separate DeclVarSrc and DeclVarGen
* Separate PtrCastSrc and PtrCastGen
* Separate CmpxchgSrc and CmpxchgGen
* Represent optional error set as an integer, using the 0 value.
In a const value, it uses nullptr.
* Introduce type_has_one_possible_value and use it where applicable.
* Fix debug builds not setting memory to 0xaa when storing
undefined.
* Separate the type of a variable from the const value of a variable.
* Use copy_const_val where appropriate.
* Rearrange structs to pack data more efficiently.
* Move test/cases/* to test/behavior/*
* Use `std.debug.assertOrPanic` in behavior tests instead of
`std.debug.assert`.
* Fix outdated slice syntax in docs.
When doing multible codegen passes (such as building
compiler_rt and then something else) the TypeInfo cache code would point
to types from the prev code gen (such as the prev 'bool' type), giving
us errors like "expected type 'bool', but found type 'bool'"
This disabling of caching might have a performance hit, but correctness is better than
speed, so let's have this for now, until someone optimizes this
correctly (probably in stage2)
* bitreverse - give bswap behavior
* bitreverse, comptime_ints, negative values still not working?
* bitreverse working for negative comptime ints
* Finished bitreverse test cases
* Undo exporting a bigint function. @bitreverse test name includes ampersand
* added docs entry for @bitreverse
Prevents LLVM from generating debug info for
struct member functions with a pointer as the
first parameter as though the first parameter
were the implicit "this" pointer from C++.