# # Copyright 2012-2016 Chef Software, Inc. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # name "ruby" license "BSD-2-Clause" license_file "BSDL" license_file "COPYING" license_file "LEGAL" skip_transitive_dependency_licensing true # - chef-client cannot use 2.2.x yet due to a bug in IRB that affects chef-shell on linux: # https://bugs.ruby-lang.org/issues/11869 # - the current status of 2.3.x is that it downloads but fails to compile. # - verify that all ffi libs are available for your version on all platforms. default_version "2.6.3" fips_enabled = (project.overrides[:fips] && project.overrides[:fips][:enabled]) || false dependency "ncurses" unless windows? || version.satisfies?(">= 2.1") dependency "zlib" dependency "openssl" dependency "libffi" dependency "libyaml" # Needed for chef_gem installs of (e.g.) nokogiri on upgrades - # they expect to see our libiconv instead of a system version. # Ignore on windows - TDM GCC comes with libiconv in the runtime # and that's the only one we will ever use. dependency "libiconv" version("2.6.3") { source sha256: "577fd3795f22b8d91c1d4e6733637b0394d4082db659fccf224c774a2b1c82fb" } source url: "https://cache.ruby-lang.org/pub/ruby/#{version.match(/^(\d+\.\d+)/)[0]}/ruby-#{version}.tar.gz" relative_path "ruby-#{version}" env = with_standard_compiler_flags(with_embedded_path) if mac_os_x? # -Qunused-arguments suppresses "argument unused during compilation" # warnings. These can be produced if you compile a program that doesn't # link to anything in a path given with -Lextra-libs. Normally these # would be harmless, except that autoconf treats any output to stderr as # a failure when it makes a test program to check your CFLAGS (regardless # of the actual exit code from the compiler). env["CFLAGS"] << " -I#{install_dir}/embedded/include/ncurses -arch x86_64 -m64 -O3 -g -pipe -Qunused-arguments" env["LDFLAGS"] << " -arch x86_64" elsif freebsd? # Stops "libtinfo.so.5.9: could not read symbols: Bad value" error when # compiling ext/readline. See the following for more info: # # https://lists.freebsd.org/pipermail/freebsd-current/2013-October/045425.html # http://mailing.freebsd.ports-bugs.narkive.com/kCgK8sNQ/ports-183106-patch-sysutils-libcdio-does-not-build-on-10-0-and-head # env["LDFLAGS"] << " -ltinfow" elsif aix? # this magic per IBM env["LDSHARED"] = "xlc -G" env["CFLAGS"] = "-I#{install_dir}/embedded/include/ncurses -I#{install_dir}/embedded/include" # this magic per IBM env["XCFLAGS"] = "-DRUBY_EXPORT" # need CPPFLAGS set so ruby doesn't try to be too clever env["CPPFLAGS"] = "-I#{install_dir}/embedded/include/ncurses -I#{install_dir}/embedded/include" env["SOLIBS"] = "-lm -lc" # need to use GNU m4, default m4 doesn't work env["M4"] = "/opt/freeware/bin/m4" elsif solaris_10? if sparc? # Known issue with rubby where too much GCC optimization blows up miniruby on sparc env["CFLAGS"] << " -std=c99 -g -pipe -mcpu=v9 -fms-extensions" env["LDFLAGS"] << " -mcpu=v9" else env["CFLAGS"] << " -std=c99 -O3 -g -pipe -fms-extensions" end elsif solaris_11? env["CFLAGS"] << " -std=c99" env["CPPFLAGS"] << " -D_XOPEN_SOURCE=600 -D_XPG6" elsif windows? env["CPPFLAGS"] << " -DFD_SETSIZE=2048" else # including linux if version.satisfies?(">= 2.3.0") && rhel? && platform_version.satisfies?("< 6.0") env["CFLAGS"] << " -O2 -g -pipe" else env["CFLAGS"] << " -O3 -g -pipe" end end build do # AIX needs /opt/freeware/bin only for patch patch_env = env.dup patch_env["PATH"] = "/opt/freeware/bin:#{env['PATH']}" if aix? if solaris_10? && version.satisfies?(">= 2.1") patch source: "ruby-no-stack-protector.patch", plevel: 1, env: patch_env if version.satisfies?("= 2.4.4") && sparc? patch source: "ruby-remove-headc.patch", plevel: 1, env: patch_env patch source: "ruby-no-m32-cflag.patch", plevel: 1, env: patch_env end elsif solaris_11? && version =~ /^2.1/ patch source: "ruby-solaris-linux-socket-compat.patch", plevel: 1, env: patch_env end # wrlinux7/ios_xr build boxes from Cisco include libssp and there is no way to # disable ruby from linking against it, but Cisco switches will not have the # library. Disabling it as we do for Solaris. if ios_xr? && version.satisfies?(">= 2.1") patch source: "ruby-no-stack-protector.patch", plevel: 1, env: patch_env end # disable libpath in mkmf across all platforms, it trolls omnibus and # breaks the postgresql cookbook. i'm not sure why ruby authors decided # this was a good idea, but it breaks our use case hard. AIX cannot even # compile without removing it, and it breaks some native gem installs on # other platforms. generally you need to have a condition where the # embedded and non-embedded libs get into a fight (libiconv, openssl, etc) # and ruby trying to set LD_LIBRARY_PATH itself gets it wrong. # # Also, fix paths emitted in the makefile on windows on both msys and msys2. if version.satisfies?(">= 2.1") patch source: "ruby-mkmf.patch", plevel: 1, env: patch_env # should intentionally break and fail to apply on 2.2, patch will need to # be fixed. end # Fix reserve stack segmentation fault when building on RHEL5 or below # Currently only affects 2.1.7 and 2.2.3. This patch taken from the fix # in Ruby trunk and expected to be included in future point releases. # https://redmine.ruby-lang.org/issues/11602 if rhel? && platform_version.satisfies?("< 6") && (version == "2.1.7" || version == "2.2.3") patch source: "ruby-fix-reserve-stack-segfault.patch", plevel: 1, env: patch_env end configure_command = ["--with-out-ext=dbm,readline", "--enable-shared", "--disable-install-doc", "--without-gmp", "--without-gdbm", "--without-tk", "--disable-dtrace"] configure_command << "--with-ext=psych" if version.satisfies?("< 2.3") configure_command << "--with-bundled-md5" if fips_enabled if aix? # need to patch ruby's configure file so it knows how to find shared libraries patch source: "ruby-aix-configure.patch", plevel: 1, env: patch_env # configure no longer seems to work under ksh, use bash instead patch source: "ruby-aix-configure-use-bash.patch", plevel: 1, env: patch_env # have ruby use zlib on AIX correctly patch source: "ruby_aix_openssl.patch", plevel: 1, env: patch_env # AIX has issues with ssl retries, need to patch to have it retry patch source: "ruby_aix_2_1_3_ssl_EAGAIN.patch", plevel: 1, env: patch_env # the next two patches are because xlc doesn't deal with long vs int types well patch source: "ruby-aix-atomic.patch", plevel: 1, env: patch_env patch source: "ruby-aix-vm-core.patch", plevel: 1, env: patch_env # per IBM, just help ruby along on what it's running on configure_command << "--host=powerpc-ibm-aix6.1.0.0 --target=powerpc-ibm-aix6.1.0.0 --build=powerpc-ibm-aix6.1.0.0 --enable-pthread" elsif freebsd? # Disable optional support C level backtrace support. This requires the # optional devel/libexecinfo port to be installed. configure_command << "ac_cv_header_execinfo_h=no" configure_command << "--with-opt-dir=#{install_dir}/embedded" elsif smartos? # Opscode patch - someara@opscode.com # GCC 4.7.0 chokes on mismatched function types between OpenSSL 1.0.1c and Ruby 1.9.3-p286 patch source: "ruby-openssl-1.0.1c.patch", plevel: 1, env: patch_env # Patches taken from RVM. # http://bugs.ruby-lang.org/issues/5384 # https://www.illumos.org/issues/1587 # https://github.com/wayneeseguin/rvm/issues/719 patch source: "rvm-cflags.patch", plevel: 1, env: patch_env # From RVM forum # https://github.com/wayneeseguin/rvm/commit/86766534fcc26f4582f23842a4d3789707ce6b96 configure_command << "ac_cv_func_dl_iterate_phdr=no" configure_command << "--with-opt-dir=#{install_dir}/embedded" elsif windows? if version.satisfies?(">= 2.3") # Windows Nano Server COM libraries do not support Apartment threading # instead COINIT_MULTITHREADED must be used patch source: "ruby_nano.patch", plevel: 1, env: patch_env end configure_command << " debugflags=-g" else configure_command << "--with-opt-dir=#{install_dir}/embedded" end # FFS: works around a bug that infects AIX when it picks up our pkg-config # AFAIK, ruby does not need or use this pkg-config it just causes the build to fail. # The alternative would be to patch configure to remove all the pkg-config garbage entirely env["PKG_CONFIG"] = "/bin/true" if aix? configure(*configure_command, env: env) make "-j #{workers}", env: env make "-j #{workers} install", env: env if windows? # Needed now that we switched to msys2 and have not figured out how to tell # it how to statically link yet dlls = ["libwinpthread-1"] if windows_arch_i386? dlls << "libgcc_s_dw2-1" else dlls << "libgcc_s_seh-1" end dlls.each do |dll| arch_suffix = windows_arch_i386? ? "32" : "64" windows_path = "C:/opscode/omnibus-toolchain/embedded/bin/#{dll}.dll" if File.exist?(windows_path) copy windows_path, "#{install_dir}/embedded/bin/#{dll}.dll" else raise "Cannot find required DLL needed for dynamic linking: #{windows_path}" end end else end end