November 2024 - ruby-dev - ml.ruby-lang.org

[ruby-dev:52086] [Ruby master Bug#20489] Ractor behavior strange in ruby master
by nekoyama32767 (Jinsong Yu) 10 Jan '25

10 Jan '25

Issue #20489 has been reported by nekoyama32767 (Jinsong Yu). ---------------------------------------- Bug #20489: Ractor behavior strange in ruby master https://bugs.ruby-lang.org/issues/20489 * Author: nekoyama32767 (Jinsong Yu) * Status: Open * ruby -v: ruby 3.4.0dev (2024-05-14T01:58:31Z master 9d01f657b3) [x86_64-linux] * Backport: 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- This is a tarai program Run`./ruby tarai_ractor.rb 2 8` is to use 2 thread to run 8 times tarai function total, that means 4 times tarai for each ractor(thread). ``` GC.disable def split_len(len, split) ret = [] mod = len % split head = 0 tail = 0 split.times do |i| if head >= len break end k = 0 if i < mod then k = 1 end tail = tail + (len/split) + k ret.append(head...tail) head = tail end return ret end def ary_split(ary, split) return split_len(ary.length,split) end def item_check(item) if item[0] != nil 1 + item_check(item[0]) + item_check(item[1]) else 1 end end def tarai(x, y, z) = x <= y ? y : tarai(tarai(x-1, y, z), tarai(y-1, z, x), tarai(z-1, x, y)) times = ARGV[0].to_i split = ARGV[1].to_i p split_len(times, split) split_len(times, split).each.map do |sp| Ractor.new (sp) { s = _1 s.each do tarai(13, 7, 0) end } end.each(&:take) ``` The problem is in ruby 3.1.2 and ruby 3.3 `./ruby tarai_ractor.rb 1 1` has simiular execute time with `./ruby tarai_ractor.rb 8 8` because each thread only run 1 time of tarai function, like follow: ruby 3.1.2p20 (2022-04-12 revision 4491bb740a) [x86_64-linux]: ``` time ruby exp_ractor_tarai.rb 1 1 [0...1] <internal:ractor>:267: warning: Ractor is experimental, and the behavior may change in future versions of Ruby! Also there are many implementation issues. real 0m1.442s user 0m1.429s sys 0m0.014s time ruby exp_ractor_tarai.rb 8 8 [0...1, 1...2, 2...3, 3...4, 4...5, 5...6, 6...7, 7...8] <internal:ractor>:267: warning: Ractor is experimental, and the behavior may change in future versions of Ruby! Also there are many implementation issues. real 0m1.857s user 0m13.817s sys 0m0.041s ``` But in ruby master(ruby 3.4.0dev) ruby 3.4.0dev (2024-05-14T01:58:31Z master 9d01f657b3) [x86_64-linux] 1 ractor 1 tarai: ``` time ../ruby exp_ractor_tarai.rb 1 1 `RubyGems' were not loaded. `error_highlight' was not loaded. `did_you_mean' was not loaded. `syntax_suggest' was not loaded. [0...1] exp_ractor_tarai.rb:47: warning: Ractor is experimental, and the behavior may change in future versions of Ruby! Also there are many implementation issues. real 0m1.671s user 0m1.666s sys 0m0.005s ``` 8 ractor 8 tarai: ``` time ../ruby exp_ractor_tarai.rb 8 8 `RubyGems' were not loaded. `error_highlight' was not loaded. `did_you_mean' was not loaded. `syntax_suggest' was not loaded. [0...1, 1...2, 2...3, 3...4, 4...5, 5...6, 6...7, 7...8] exp_ractor_tarai.rb:47: warning: Ractor is experimental, and the behavior may change in future versions of Ruby! Also there are many implementation issues. real 0m18.408s user 1m58.659s sys 0m0.021s ``` And in ruby 3.4.0dev when run `time ../ruby exp_ractor_tarai.rb 16 16` 16 thread should be used in system monitoring while only 8 threads are used. Ruby 3.3 and Ruby 3.1.2 do not have this problem. ---Files-------------------------------- thead16_16.png (168 KB) thread16_8.png (165 KB) -- https://bugs.ruby-lang.org/

5 13

[ruby-dev:52125] [Ruby master Bug#20888] TestGemRemoteFetcherLocalSSLServer tests or so began to fail
by mtasaka (Mamoru TASAKA) 06 Dec '24

06 Dec '24

Issue #20888 has been reported by mtasaka (Mamoru TASAKA). ---------------------------------------- Bug #20888: TestGemRemoteFetcherLocalSSLServer tests or so began to fail https://bugs.ruby-lang.org/issues/20888 * Author: mtasaka (Mamoru TASAKA) * Status: Open * ruby -v: ruby 3.4.0dev (2024-11-12 master a6fdd8883c) +PRISM [x86_64-linux] * Backport: 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- With repeatedly trying to rebuild ruby master on Fedora copr build server, recently TestGemRemoteFetcherLocalSSLServer tests or so tests began to fail on x86_64 as: ``` 183) Failure: TestSocket_TCPSocket#test_initialize_with_connection_failure_after_hostname_resolution_failure [/builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/socket/test_tcp.rb:318]: [Errno::ECONNREFUSED] exception expected, not #<Socket::ResolutionError: getaddrinfo(3): Non-recoverable failure in name resolution>. 184) Failure: TestSocket_TCPSocket#test_initialize_resolv_timeout_with_connection_failure [/builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/socket/test_tcp.rb:284]: [Errno::ETIMEDOUT] exception expected, not #<Errno::ECONNREFUSED: Connection refused - connect(2) for "localhost" port 39177>. 185) Failure: TestSocket_TCPSocket#test_initialize_with_hostname_resolution_failure_after_connection_failure [/builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/socket/test_tcp.rb:303]: [Socket::ResolutionError] exception expected, not #<Errno::ECONNREFUSED: Connection refused - connect(2) for "localhost" port 40983>. 186) Error: TestGemRemoteFetcherLocalSSLServer#test_ssl_connection_allow_verify_none: Gem::RemoteFetcher::FetchError: Errno::ECONNREFUSED: Failed to open TCP connection to localhost:34229 (Connection refused - connect(2) for "localhost" port 34229) (https://localhost:34229/yaml) /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/lib/rubygems/remote_fetcher.rb:265:in 'Gem::RemoteFetcher#fetch_path' /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/rubygems/test_gem_remote_fetcher_local_ssl_server.rb:103:in 'block in TestGemRemoteFetcherLocalSSLServer#test_ssl_connection_allow_verify_none' /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/rubygems/test_gem_remote_fetcher_local_ssl_server.rb:145:in 'TestGemRemoteFetcherLocalSSLServer#with_configured_fetcher' /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/rubygems/test_gem_remote_fetcher_local_ssl_server.rb:102:in 'TestGemRemoteFetcherLocalSSLServer#test_ssl_connection_allow_verify_none' 187) Error: TestSocket_TCPSocket#test_initialize_v6_hostname_resolution_failed_and_v4_hostname_resolution_is_success: Socket::ResolutionError: getaddrinfo(3): Non-recoverable failure in name resolution /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/socket/test_tcp.rb:266:in 'TCPSocket#initialize' /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/socket/test_tcp.rb:266:in 'IO.new' /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/socket/test_tcp.rb:266:in '<main>' /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/socket/test_tcp.rb:259:in 'TestSocket_TCPSocket#test_initialize_v6_hostname_resolution_failed_and_v4_hostname_resolution_is_success' 188) Error: TestSocket_TCPSocket#test_initialize_v6_hostname_resolved_earlier_and_v6_server_is_not_listening: Errno::ECONNREFUSED: Connection refused - connect(2) for "localhost" port 47167 /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/socket/test_tcp.rb:222:in 'TCPSocket#initialize' /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/socket/test_tcp.rb:222:in 'IO.new' /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/socket/test_tcp.rb:222:in '<main>' /builddir/build/BUILD/ruby-3.4.0_20241112.1157gitfee706d9dd-build/ruby-3.4.0-fee706d9dd/test/socket/test_tcp.rb:213:in 'TestSocket_TCPSocket#test_initialize_v6_hostname_resolved_earlier_and_v6_server_is_not_listening' Finished tests in 692.000057s, 46.3801 tests/s, 9452.0484 assertions/s. 32095 tests, 6540818 assertions, 3 failures, 3 errors, 182 skips ``` ( From 1) to 182) are skipped tests ) Looks like * commit:821a5b966fbc2926dc3bf88b6ba09879fa35318e tests are successful * commit:a6fdd8883cab6b31357552db8a64cfb335678172 sees the above error * commit:fee706d9dd52d7f0444357259f33d335f21c980b sees the above error So looks like the culprit is between the above 2 commits. ( I suspect commit:4c270200dbc2a3a4511e8b793a033078afd6fb31 but not sure ) -- https://bugs.ruby-lang.org/

3 6

[ruby-dev:52133] [Ruby master Feature#11177] DATAでEOF文字以降が読めない
by YO4 (Yoshinao Muramatsu) 22 Nov '24

22 Nov '24

Issue #11177 has been updated by YO4 (Yoshinao Muramatsu). rubyの-xオプション付与でソースファイルをバイナリオープンさせることでも```\x1A```以降の読み込みが可能でした。こちらは ruby 3.3またはそれ以前でも可能のようです。 ``` >less -F data.rb #!ruby DATA.pos # posなしだと [] しか得られない p IO.open(DATA.fileno, "rt") {|f| f.readlines } __END__ textpart1 textpart1^Z binpart1 binpart2 >ruby -x data.rb ["textpart1\n", "textpart1\u001A\n", "binpart1\n", "binpart2\n"] ``` 同様の動作を仕込むことで EOF以降も DATAでアクセス可能にできそうですが、DATA.class が Fileでなく IOになっちゃいますね… ---------------------------------------- Feature #11177: DATAでEOF文字以降が読めない https://bugs.ruby-lang.org/issues/11177#change-110732 * Author: mame (Yusuke Endoh) * Status: Open ---------------------------------------- 遠藤です。 Windows で `__END__` 以降に EOF 文字 (`\x1A`) があったとき、それより先が読めないのは仕様でしょうか。 gen.rb: puts "p DATA.read" puts "__END__" puts "foo\x1Abar" 以下のように実行すると再現します。 > ruby gen.rb > t.rb > ruby t.rb "foo" `DATA.binmode.read` などとしてみても同じです。もちろん、Linux では先まで読めます。Windows でも、ソースコードをパイプで流し込んだ場合はなぜか読めます。 > ruby < t.rb "foo\x1Abar\n" さらに、EOF 以降に文字がいっぱいあった場合、EOF 以降の一部の文字が抜け落ちるような挙動になるようです。 gen2.rb: puts "p DATA.read" puts "__END__" puts "foo\x1A" + "X" * 8192 + "Z" > ruby gen2.rb > t.rb > ruby t.rb "fooXXXXXXXXXXXXXXXXXXXXXXXXXXXZ\r\n" バッファリングのバグっぽい挙動ですが、バグでしょうか。 -- Yusuke Endoh <mame(a)ruby-lang.org> -- https://bugs.ruby-lang.org/

1 0

[ruby-dev:52132] [Ruby master Feature#11177] DATAでEOF文字以降が読めない
by YO4 (Yoshinao Muramatsu) 21 Nov '24

21 Nov '24

Issue #11177 has been updated by YO4 (Yoshinao Muramatsu). 本件についてDATA.pos が__END__直後を示さないこと、Cのfseek(0, SEEK_CUR)っぽく DATA.seek(0, IO:SEEK_CUR)を行っても改善しないことから単純なワークアラウンドは存在しないと思っていましたが、 3.4-head で試したところ prismでの仕様の違いか別の理由からかは確認できていませんが DATA.pos が __END__直後を指していたので ``` IO.open(DATA.fileno, "rt") { _1.readlines } # 明示的にuniversal newlineを使用してopen ``` で```\x1A```込みで読み込み可能でした。 ``` >less -F data.rb p IO.open(DATA.fileno, "rt") { _1.readlines } __END__ textpart1 textpart1^Z binpart1 binpart2 >ruby data.rb ["textpart1\n", "textpart1\u001A\n", "binpart1\n", "binpart2\n"] ``` 上記にて、data.rb の改行は CR+LF です。 ---------------------------------------- Feature #11177: DATAでEOF文字以降が読めない https://bugs.ruby-lang.org/issues/11177#change-110720 * Author: mame (Yusuke Endoh) * Status: Open ---------------------------------------- 遠藤です。 Windows で `__END__` 以降に EOF 文字 (`\x1A`) があったとき、それより先が読めないのは仕様でしょうか。 gen.rb: puts "p DATA.read" puts "__END__" puts "foo\x1Abar" 以下のように実行すると再現します。 > ruby gen.rb > t.rb > ruby t.rb "foo" `DATA.binmode.read` などとしてみても同じです。もちろん、Linux では先まで読めます。Windows でも、ソースコードをパイプで流し込んだ場合はなぜか読めます。 > ruby < t.rb "foo\x1Abar\n" さらに、EOF 以降に文字がいっぱいあった場合、EOF 以降の一部の文字が抜け落ちるような挙動になるようです。 gen2.rb: puts "p DATA.read" puts "__END__" puts "foo\x1A" + "X" * 8192 + "Z" > ruby gen2.rb > t.rb > ruby t.rb "fooXXXXXXXXXXXXXXXXXXXXXXXXXXXZ\r\n" バッファリングのバグっぽい挙動ですが、バグでしょうか。 -- Yusuke Endoh <mame(a)ruby-lang.org> -- https://bugs.ruby-lang.org/

1 0

[ruby-dev:52120] [Ruby master Bug#20862] gem install --document=rdoc,ri fails with some gem
by mtasaka (Mamoru TASAKA) 18 Nov '24

18 Nov '24

Issue #20862 has been reported by mtasaka (Mamoru TASAKA). ---------------------------------------- Bug #20862: gem install --document=rdoc,ri fails with some gem https://bugs.ruby-lang.org/issues/20862 * Author: mtasaka (Mamoru TASAKA) * Status: Open * ruby -v: ruby 3.4.0dev (2024-11-02 master 24f7829abf) +PRISM [x86_64-linux] * Backport: 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- With using commit:24f7829abfd29fc226e38829caaecaa76885afbf , trying `$ gem install --document=rdoc,ri foo.gem` fails with some (not one) gem like: ``` $ gem install -V --document=rdoc,ri tk ... ... ERROR: While executing gem ... (RDoc::Error) error generating Busy.html: Error while evaluating /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/template/darkfish/class.rhtml: undefined method 'match' for an instance of RDoc::Markup::Document (RDoc::Error) /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/darkfish.rb:789:in 'RDoc::Generator::Darkfish#excerpt' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/template/darkfish/class.rhtml:16:in 'block in RDoc::Generator::Darkfish#generate_class' /usr/share/ruby/erb.rb:429:in 'Kernel#eval' /usr/share/ruby/erb.rb:429:in 'ERB#result' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/darkfish.rb:750:in 'RDoc::Generator::Darkfish#template_result' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/darkfish.rb:729:in 'block in RDoc::Generator::Darkfish#render_template' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/darkfish.rb:724:in 'IO.open' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/darkfish.rb:724:in 'Pathname#open' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/darkfish.rb:724:in 'RDoc::Generator::Darkfish#render_template' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/darkfish.rb:356:in 'RDoc::Generator::Darkfish#generate_class' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/darkfish.rb:382:in 'block in RDoc::Generator::Darkfish#generate_class_files' <internal:array>:42:in 'Array#each' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/darkfish.rb:379:in 'RDoc::Generator::Darkfish#generate_class_files' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/generator/darkfish.rb:246:in 'RDoc::Generator::Darkfish#generate' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/rubygems_hook.rb:139:in 'block in RDoc::RubygemsHook#document' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/rubygems_hook.rb:136:in 'Dir.chdir' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/rubygems_hook.rb:136:in 'RDoc::RubygemsHook#document' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/rubygems_hook.rb:198:in 'RDoc::RubygemsHook#generate' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/rubygems_hook.rb:56:in 'block in RDoc::RubygemsHook.generation_hook' <internal:array>:42:in 'Array#each' /usr/share/gems/gems/rdoc-6.7.0/lib/rdoc/rubygems_hook.rb:55:in 'RDoc::RubygemsHook.generation_hook' /usr/share/rubygems/rubygems/request_set.rb:313:in 'block in Gem::RequestSet#install_hooks' <internal:array>:42:in 'Array#each' /usr/share/rubygems/rubygems/request_set.rb:312:in 'Gem::RequestSet#install_hooks' /usr/share/rubygems/rubygems/request_set.rb:210:in 'Gem::RequestSet#install' /usr/share/rubygems/rubygems/commands/install_command.rb:207:in 'Gem::Commands::InstallCommand#install_gem' /usr/share/rubygems/rubygems/commands/install_command.rb:223:in 'block in Gem::Commands::InstallCommand#install_gems' <internal:array>:42:in 'Array#each' /usr/share/rubygems/rubygems/commands/install_command.rb:216:in 'Gem::Commands::InstallCommand#install_gems' /usr/share/rubygems/rubygems/commands/install_command.rb:162:in 'Gem::Commands::InstallCommand#execute' /usr/share/rubygems/rubygems/command.rb:326:in 'Gem::Command#invoke_with_build_args' /usr/share/rubygems/rubygems/command_manager.rb:253:in 'Gem::CommandManager#invoke_command' /usr/share/rubygems/rubygems/command_manager.rb:194:in 'Gem::CommandManager#process_args' /usr/share/rubygems/rubygems/command_manager.rb:152:in 'Gem::CommandManager#run' /usr/share/rubygems/rubygems/gem_runner.rb:56:in 'Gem::GemRunner#run' /usr/bin/gem:12:in '<main>' Installing ri documentation for tk-0.5.0 Installing darkfish documentation for tk-0.5.0 ``` I guess this is related to commit:ed993b5bcc4fcae661dd022d3211dcc770425218 , but not sure. Note that "vanilla" rdoc 6.7.0 does not seem to have the above change, the change on rdoc was installed on https://github.com/ruby/rdoc/commit/716bc16a7dbcac937d38c328d2d89a2f7cff2fa9 , and "vanilla" rdoc 6.7.0 seems NOT to reproduce this issue. -- https://bugs.ruby-lang.org/

3 3

[ruby-dev:52124] 'help'
by たぬき 12 Nov '24

12 Nov '24

1 0

[ruby-dev:52116] [Ruby master Bug#20767] VM_ASSERT fails on clear_method_entry_by_id_in_class() under a specific situation
by tagomoris (Satoshi Tagomori) 05 Nov '24

05 Nov '24

Issue #20767 has been reported by tagomoris (Satoshi Tagomori). ---------------------------------------- Bug #20767: VM_ASSERT fails on clear_method_entry_by_id_in_class() under a specific situation https://bugs.ruby-lang.org/issues/20767 * Author: tagomoris (Satoshi Tagomori) * Status: Open * Backport: 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- VM_ASSERT in `clear_method_entry_by_id_in_class()` (vm_method.c) fails when: * Kernel#require is refined * Kernel#require is replaced by `alias_method` and `define_method` * User code calls `require` * The aliased method entry is undefined Under this situation, `VM_ASSERT(cme->owner == T_CLASS)` fails in `clear_method_entry_by_id_in_class`. This CI failure shows it: https://github.com/ruby/ruby/actions/runs/11081754435/job/30793895707?pr=11… In this case, the cme had: * def->type: CFUNC * defined_class: Kernel * owner: Kernel I couldn't find the root cause of why Kernel is set on cme->owner, and why the owner is checked here. How should we fix this problem? * Just delete the VM_ASSERT() (because the cme is just to be invalidated) * Fix the root cause of the invalid&unexpected cme->owner -- https://bugs.ruby-lang.org/

4 4