1. Store the `PKG_CONFIG` variable in Makefile.sub (or try to get it from the ENV var PKG_CONFIG in mkmf.rb)
2. Try to use --msvc-syntax, with a fallback to replacing -Lxxx with -libpath:xxx. --msvc-syntax has been in pkgconf since 1.4.0 (released 7 years ago). pkg-config (freedesktop), does not support it, hence the fallback.
3. The `try_ldflags` passes these `ldflags` as the `opt` parameter to the `link_command`, not as `ldflags`. Unix systems are forgiving in that regard, MSVC is not: as a result as passing them as `opt`, they (specifically the `/libpath:xxx` ones) end up passed before the `-link` command to `cl.exe` and it throws because it ignores it and therefore can't find the lib.
```
cl : Command line warning D9002 : ignoring unknown option '-libpath:C:/Users/julien/.conan2/p/libff3726d89a6255c/p/lib'
```
Introduce a new mkmf option `--target-rbconfig` to specify the RbConfig
file for the deployment target platform. This option is useful for
cross-compiling Ruby extensions without faking the global top-level
`RbConfig` constant.
By replacing `ALLOBJS` suffix with intermediate file suffixes instead
of roughly removing by wildcards. Made `cleanlibs` append `.dSYM`
suffix for each word in `TARGET_SO`, not the end of the entire list.
The `POSTLINK` variable had been used in the `LINK_SO` variable, which
is used to link shared extension libraries. However, the `POSTLINK`
variable had not been defined in the generated Makefile, so extension
libraries were not properly post-processed. It was not a critical issue
for the existing `POSTLINK` usage for darwin platforms, but it would be
a problem for Wasm/WASI platform, which requires *mandatory*
post-processing for shared extension libraries.
"AR" was well propagated to the enc.mk and mkmf, but "LD" was not. This
caused the dynamic libraries to be linked with a linker found in the PATH,
which could be different from the one used in the Ruby build process.
This is especially important for cross-compilation, where the host
linker may not be compatible with the target system. (e.g. WebAssembly)
* We always need use_sigaltstack=0 because Ruby registers sigaltstack
handlers
* We also need to disable leak detection (unless RUBY_FREE_AT_EXIT is
set - I might experiment later with automatically enabling leak
detection if RUBY_FREE_AT_EXIT is set).
Burning it into the built ruby binary in this way avoids people needing
to remember to start their Ruby program with these flags all the time.
We also need a small fix in mkmf to make sure that test programs also
don't have leak detection enabled (this is never desirable)
[Bug #20256]
The extconf.rb in mysql2 gem repeats `dir_config('mysql')`, without
and with the default path. The third call returns the former results
even with the default path. Since it does not check the results of
the third call, that `nil` is passed to `find_library` as a path, and
fails with `NoMethodError`.
This commit is to add an extra option to enable verbose mode (V=1) in the
generated `Makefile` at runtime of the Ruby to print compiler command lines by
the commands below when building native extensions. It's possible to enable the
verbose mode by setting the environment variable `MAKEFLAGS="V=1"`[1]
implemented in GNU make. However, I wanted to make a consistent user-interface
not depending on the specific make's implementation.
```
$ ruby /path/to/extconf.rb -- --with-verbose
```
You can also add the extra option via rake-compiler gem.
```
$ rake compiler -- --with-verbose
```
If the extra option is not given, the value of the
`RbConfig::CONFIG["MKMF_VERBOSE"]` enabled by the configure option below is
used.
```
$ ./configure --enable-mkmf-verbose
```
For the unit tests, updated the following files.
* The `test/mkmf/test_configuration.rb` was created to test the cases with the
`configuration` method and this implementation.
* Updated the `TestMkmf#assert_separately` to set the extra
arguments in `test/mkmf/base.rb`. Updated tests using the `assert_separately`.
* Added tests for `MakeMakefile#with_config` in the `test/mkmf/test_config.rb`.
[1] https://www.gnu.org/software/make/manual/html_node/Variables_002fRecursion.html
Fixes [Bug #19695]
Co-authored-by: Nobuyoshi Nakada <nobu@ruby-lang.org>
The warning against `-undefined dynamic_lookup` is just a warning yet,
and many gems seem to pay no attention to warnings. Until it fails
actually, keep it as a migration path, except for standard extension
libraries and bundled extension gems.
For the macOS -bundle_loader linker option, we need a path to the
Ruby exectuable. $(RUBY) is not necessarily a path since it could
be a command line invocation. That happens during build with
runruby.rb and can happen post installation if the user passes
the --ruby option to a extconf.rb. Use $(bindir) to locate
the executable instead.
Before installation, $(bindir) doesn't exist, so we need to be
able to override $(BUILTRUBY) in such situations so test-spec
and bundled extensions could build. Use a new mkmf global,
$builtruby, to do this; set it in fake.rb and in extmk.rb.
Our build system is quite complex...
ld64 shipped with Xcode 14 emits a warning when using `-undefined
dynamic_lookup`.
```
ld: warning: -undefined dynamic_lookup may not work with chained fixups
```
Actually, `-undefined dynamic_lookup` doesn't work when:
1. Link a *shared library* with the option
2. Link it with a program that uses the chained-fixup introduced from
macOS 12 and iOS 15
because `-undefined dynamic_lookup` uses lazy-bindings and they won't be
bound while dyld fixes-up by traversing chained-fixup info.
However, we build exts as *bundles* and they are loaded only through
`dlopen`, so it's safe to use `-undefined dynamic_lookup` in theory.
So the warning produced by ld64 is false-positive, and it results
failure of option checking in configuration. Therefore, it would be an
option to ignore the warning during our configuration.
On the other hand, `-undefined dynamic_lookup` is already deprecated on
all darwin platforms except for macOS, so it's good time to get rid of
the option. ld64 also provides `-bundle_loader <executable>` option,
which allows to resolve symbols defined in the executable symtab while
linking. It behaves almost the same with `-undefined dynamic_lookup`,
but it makes the following changes:
1. Require that unresolved symbols among input objects must be defined
in the executable.
2. Lazy symbol binding will lookup only the symtab of the bundle loader
executable. (`-undefined dynamic_lookup` lookups all symtab as flat
namespace)
This patch adds `-bundle_loader $(RUBY)` when non-EXTSTATIC
configuration by assuming ruby executable can be linked before building
exts.
See "New Features" subsection under "Linking" section for chained fixup
https://developer.apple.com/documentation/xcode-release-notes/xcode-13-release-notes
Note this change is only for `configure.ac`, not for Windows using
`win32/configure.bat`.
```
$ ./configure --help | grep mkmf
--enable-mkmf-verbose enable verbose in mkmf
```
Run the following command to enable the mkmf verbose mode.
```
$ ./configure --enable-mkmf-verbose
$ grep MKMF_VERBOSE config.status
S["MKMF_VERBOSE"]="1"
```
In this mkmf verbose mode, when compiling a native extension, the
`rake compile` prints the compiling commands such as
"gcc -I. <...> path/to/file" instead of "compiling path/to/file".
```
$ git clone https://github.com/deivid-rodriguez/byebug.git
$ cd byebug
$ bundle install --standalone
$ bundle exec rake compile
...
gcc -I. <...> path/to/file
...
```
This method is at least 7 years old and is widely used in the wild.
Since we need to support it, let's document it to make it discoverable.
Add docs and move it out of the `# :stopdoc:` zone.
