What's Changed

• Feature #18980: Re-reconsider numbered parameters: it as a default block parameter
• Feature #19370: Anonymous parameters for blocks?
• Fix Window private file mapping unlink EACCES issue. by ioquatix · Pull Request #9358
• Feature #19982: Bump required Visual Studio version to 2015 after 3.3
• Bug #20087: Uninitialized instance variable warning - obsolete documentation
• Test_SyncDefaultGems: Fail when editor run by sorah · Pull Request #9365
• Bug #20094: Inline while loop behavior changed unexpectedly in 3.3.0
• Bug #20088: Ruby 3.3.0 does not cross-complie on arm64-darwin
• Bug #20085: Fiber.new{ }.resume causes Segmentation fault for Ruby 3.3.0 on aarch64-linux
• Bug #20096: Ruby 3.2.2 win32/registry: Junk appended to Windows Registry String Value
• Bug #20083: String#match? behaving inconsistently with Ruby 3.3.0
• [DOC] Add What's Here for Complex by BurdetteLamar · Pull Request #9366
• Bug #14607: Fix use of the rb_profile_frames start parameter
• Bug #20098: Wrong regexp match in ruby 3.2 and 3.3
• Reduce if for decreasing counter on OP_REPEAT_INC by makenowjust · Pull Request #9393
• [DOC] Fix wrong description about Process.waitpid2 by takaram · Pull Request #9401
• Roll net-* bundled gems from git to fix CI by sorah · Pull Request #9403
• Bug #20104: Regexp#match returns nil but allocates T_MATCH objects
• Bug #19409: Object's shape is reset after a ractor move
• Bug #20145: Memory leak when duplicating identhash
• Bug #19916: URI#to_s can serialize to a value that doesn't deserialize to the original
• Bug #11526: Streaming HTTP requests are not idempotent and should not be retried
• YJIT: Let RubyVM::YJIT.enable respect --yjit-stats by k0kubun · Pull Request #9415
• Bug #20157: Regression in GC.measure_total_time
• Bug #20150: Memory leak in grapheme clusters
• YJIT: Fallback Integer#<< if a shift amount varies by k0kubun · Pull Request #9426
• Bug #20161: Memory leak in regexp grapheme clusters
• Fix to work match cache with peek next optimization by makenowjust · Pull Request #9459
• Fix test case for test_match_cache_with_peek_optimization by makenowjust · Pull Request #9466
• Bug #20101: rb_file_open and rb_io_fdopen don't perform CRLF -> LF conversion when encoding is set
• Bug #20162: Memory leak when duplicating too complex object
• Bug #20111: defined? returns expression for assignment operators combined with fully qualified constants
• Remove unnecessary semicolons by S-H-GAMELINKS · Pull Request #9469
• Bug #20172: Socket.addrinfo failing randomly
• Remove a unused variable in i_print_name_entry by makenowjust · Pull Request #9468
• Bug #19990: Could we reconsider the second argument to Kernel#load?
• Bug #20178: Out of bounds stack read on Array#first when built with -O0
• Bug #20001: Make Ruby work properly with ASAN enabled
• YJIT: expandarray for non-arrays by ywenc · Pull Request #9495
• Bug #20064: Inconsistent behavior between array splat *nil and hash splat **nil
• Improve behavioural consistency of unallocated (zero length) IO::Buffer. by ioquatix · Pull Request #9532
• YJIT: Optimize Integer#succ by k0kubun · Pull Request #9519
• YJIT: optimized codegen for String#length by maximecb · Pull Request #9543
• Bug #20184: Ruby segfaults on Fly.io with 256 MB RAM
• Bump the required BASERUBY version to 2.7 by k0kubun · Pull Request #9566
• YJIT: specialized codegen for integer right shift by maximecb · Pull Request #9564
• YJIT: Support empty splat and some block_arg calls to ivar getters by XrXr · Pull Request #9567
• [DOC] RDoc for ARGF by BurdetteLamar · Pull Request #9558
• YJIT: Stop incrementing chain_depth on defer_compilation by k0kubun · Pull Request #9597
• Bug #20192: YJIT in 3.3.0 miscompiles yield with keyword splats
• YJIT: Optimize defined?(yield) by k0kubun · Pull Request #9599
• YJIT: Avoid doubly splitting Opnd::Value on CSel by k0kubun · Pull Request #9617
• Bug #20198: Threaded DNS resolver does not propagate errno to the calling thread
• YJIT: Drop extra arguments passed by yield by XrXr · Pull Request #9596
• Set a cached Ruby path using setup-ruby by k0kubun · Pull Request #9645
• Bump shlex from 1.1.0 to 1.3.0 in /yjit/bindgen by dependabot[bot] · Pull Request #9652
• Bump capstone from 0.10.0 to 0.11.0 in /yjit by dependabot[bot] · Pull Request #9653
• Bug #20194: Memory leak with TracePoint on bmethod
• Bug #20204: 3.3.0 YJIT rises TypeError instead of ArgumentError with some incorrect calls
• YJIT: Allow inlining ISEQ calls with a block by k0kubun · Pull Request #9622
• Rewrite Array#each in Ruby using Primitive by k0kubun · Pull Request #9533
• Dump annotations on RubyVM::ISeq.disasm by k0kubun · Pull Request #9667
• Bug #20195: 3.3.0 YJIT mishandles ruby2_keywords splat into methods taking a rest parameter
• YJIT: fix small bug causing jit_rb_int_rshift to side-exit by maximecb · Pull Request #9684
• YJIT: reduce default exec mem size to 48MiB by maximecb · Pull Request #9685
• YJIT: Update yjit.md about mem size by k0kubun · Pull Request #9687
• Bug #20209: YJIT can leak memory by retaining objects with singleton class
• Omit TestCompilePrism on s390x for now by k0kubun · Pull Request #9694
• Bug #20197: Postponed job invocations are significantly reduced in Ruby 3.3
• YJIT: Add a counter for invokebuiltin exits by k0kubun · Pull Request #9696
• YJIT: Support concattoarray and pushtoarray by k0kubun · Pull Request #9708
• Bundle rbs-3.4.3 by soutaro · Pull Request #9702
• YJIT: Fix exits on splatkw instruction by k0kubun · Pull Request #9711
• Optimize compilation of large literal arrays by jeremyevans · Pull Request #9721
• Bug #20217: void value not checked for begin/ensure/rescue
• Bug #20219: Segfault with circular parameter
• Bug #20207: Segmentation fault for a regexp containing positive and negative lookaheads
• YJIT: add asm comment when we clear local types by maximecb · Pull Request #9713
• YJIT: Fix tailcall and JIT entry eating up FINISH frames by XrXr · Pull Request #9729
• YJIT: print warning when disasm options used without a dev build by maximecb · Pull Request #9744
• Bug #20227: Unexpected "internal:array" appears since c84237f9531aed3b204d3fdacc2dd9d2bd4c7d81
• YJIT: add specialized codegen for fixnum XOR by maximecb · Pull Request #9763
• YJIT: Specialize splatkw on T_HASH by k0kubun · Pull Request #9764
• Annotate Symbol#to_s as leaf by k0kubun · Pull Request #9769
• YJIT: Add jit_prepare_call_with_gc function by k0kubun · Pull Request #9775
• YJIT: Add codegen for Float arithmetics by k0kubun · Pull Request #9774
• Don't wait in io_binwrite_string if not necessary. by ioquatix · Pull Request #9792
• [DOC] Pretty tables for Kernel#test doc by BurdetteLamar · Pull Request #9751
• Bug #20228: Memory leak in Regexp timeout
• YJIT: add missing jge comparison instruction by maximecb · Pull Request #9819
• Bug #20169: GC.compact can raises EFAULT on IO
• Bug #20220: M:N threading needs to tell ASAN about stack switches
• Fix test session reuse but expire by nurse · Pull Request #9824
• Bug #20239: Segmentation fault when using Regex on a large String
• Bug #20221: ASAN: make test-basic: un-prefixed symbol leakage
• Bundle rbs-3.4.4 by soutaro · Pull Request #9883
• Bug #20245: Crash when checking symbol encoding
• YJIT: Skip pushing a frame for Hash#empty? by k0kubun · Pull Request #9875
• YJIT: Use jit_prepare_call_with_gc as much as possible by k0kubun · Pull Request #9874
• YJIT: Report invalidation counts in non-stats mode by jhawthorn · Pull Request #9878
• YJIT: Maintain MapToLocal that is just upgraded by k0kubun · Pull Request #9876
• YJIT: Allow tracing a counted exit by k0kubun · Pull Request #9890
• YJIT: Remove unnecessary casts for chain_depth by k0kubun · Pull Request #9893
• YJIT: Refactor recv_known_class to Option by k0kubun · Pull Request #9895
• Bug #20250: Crash with "Object ID seen, but not in mapping table: proc" error
• More on IO doc by BurdetteLamar · Pull Request #9842
• YJIT: Fallback megamorphic opt_case_dispatch by k0kubun · Pull Request #9894
• YJIT: Add top ISEQ call counts to --yjit-stats by k0kubun · Pull Request #9906
• Bug #20256: Make required ASAN settings "built in" without the need for ASAN_OPTIONS env var
• Bug #20248: Ruby does not build with ASAN support with clang
• [DOC] :stopdoc: directive must be on its own line by nobu · Pull Request #9916
• YJIT: Fix insufficient type guards by k0kubun · Pull Request #9911
• YJIT: Adjust the padding size of counts automatically by k0kubun · Pull Request #9912
• Bug #20253: Proc.dup and Proc#clone don't preserve finalizers
• YJIT: Prefer an overloaded cme if available by k0kubun · Pull Request #9913
• More on auto-link by BurdetteLamar · Pull Request #9907
• Bug #20246: Unexpected behavior for Regexp in Subexpression Calls on Ruby 3.3.0
• Specialize String#byteslice(a, b) by tenderlove · Pull Request #9939
• [DOC] Doc compliance by BurdetteLamar · Pull Request #9944
• [DOC] Doc compliance by BurdetteLamar · Pull Request #9955
• YJIT: Simplify Kernel#send guards and admit more cases by XrXr · Pull Request #9956
• YJIT: Add --yjit-perf=codegen option by k0kubun · Pull Request #9957
• YJIT: Use i32 over isize for ctx.sp_opnd() by XrXr · Pull Request #9968
• YJIT: Replace perf_fn! with a simpler macro by k0kubun · Pull Request #9971
• Feature #16495: Inconsistent quotes in error messages
• Feature #19117: Include the method owner in backtraces, not just the method name
• YJIT: Pass nil to anonymous kwrest when empty by XrXr · Pull Request #9972
• Let annocheck workflow run only test-annocheck by k0kubun · Pull Request #9990
• Bump the required BASERUBY version to 3.0 by k0kubun · Pull Request #9976
• Bug #20280: Error at invalid encoding symbols
• Feature #20257: Rearchitect Ripper
• [DOC] Fix links in doc by BurdetteLamar · Pull Request #9970
• YJIT: Verify the assumption of leaf C calls by k0kubun · Pull Request #10002
• YJIT: rb_str_concat_literals is not leaf by k0kubun · Pull Request #10035
• Ensure that exiting thread invokes end-of-life behaviour. by ioquatix · Pull Request #10039
• Fix exception handling in rb_fiber_scheduler_set. by ioquatix · Pull Request #10042
• YJIT: Allow non-leaf calls on opt_* insns by k0kubun · Pull Request #10033
• [DOC] Doc for field processing by BurdetteLamar · Pull Request #10008
• YJIT: Grab stack operands after SP change in String#byteslice by XrXr · Pull Request #10060
• Fix the insn position shown by rb_vm_check_canary by k0kubun · Pull Request #10062
• YJIT: Assert Opnd::Stack's SP expectation by k0kubun · Pull Request #10061
• Remove SAVE_ROOT_JMPBUF as it no longer has any effect. by ioquatix · Pull Request #10066
• Bug #20292: Abort ruby by String#initialize
• YJIT: Optimize attr_writer by k0kubun · Pull Request #9986
• Skip under_gc_compact_stress on s390x by k0kubun · Pull Request #10073
• Fallback every macos-arm-oss usage to macos-14 by k0kubun · Pull Request #10074
• Fix -Wmaybe-uninitialized on Complex#** by k0kubun · Pull Request #10075
• Bug #20295: Segmentation fault when parsing /invalid_regexp/ =~ s
• Bug #20258: Socket::IPV6_RECVERR for IPv6
• Bug #20234: Segfault parsing begin statement inside method definition receiver
• YJIT: Lazily push a frame for specialized C funcs by k0kubun · Pull Request #10080
• [DOC] Doc for command line options by BurdetteLamar · Pull Request #10059
• Bug #20296: Complex(:sym, exception: false) generate exception with weird timing
• Bump capstone from 0.11.0 to 0.12.0 in /yjit by dependabot[bot] · Pull Request #10094
• Introduction of Happy Eyeballs Version 2 (RFC8305) in Socket.tcp by shioimm · Pull Request #9374
• Bug #5179: Complex#rationalize and to_r with approximate zeros
• Bug #20305: commit 1d2d25dcadda0764f303183ac091d0c87b432566 breaks grapheme_clusters
• Bug #20302: Multiple refinements cannot be applied to the same module
• [DOC] Stop discouraging the use of Array#each by k0kubun · Pull Request #10119
• Feature #20274: Add RubyVM::ASAN.enabled?
• Migrate auto_request_review back to CODEOWNERS by k0kubun · Pull Request #10133
• Update bundled_gems by soutaro · Pull Request #10121
• YJIT: Support inlining putself by k0kubun · Pull Request #10137
• Bug #20311: Struct.new("A") memory leak?
• YJIT: Support opt_invokebuiltin_delegate for leaf builtin by k0kubun · Pull Request #10152
• Optimize Hash methods with Kernel#hash by k0kubun · Pull Request #10160
• Bug #20322: rb_enc_interned_str_cstr doesn't accept null pointer for encoding
• Feature #20273: Disable callcc when compiled with ASAN
• Bug #20324: (1..).overlap?('foo'..) returns true
• [DOC] Fix invalid documentation for reachable_objects_from by iridakos · Pull Request #10172
• Bug #19975: ISeq#to_binary loses hidden local variable indices
• Hide public implementation of rb_io. by ioquatix · Pull Request #9568
• [DOC] Ruby command-line options by BurdetteLamar · Pull Request #10138
• YJIT: String#getbyte codegen by maximecb · Pull Request #10188
• Bug #20327: Time.new behaves differently when passing a zone as timezone object
• Bundle unmerged rbs for testing by soutaro · Pull Request #10211
• Pass RUN_OPTS in YJIT macos test by nobu · Pull Request #10214
• Bug #20277: Remove stale String test conditionals
• Revisions for help text by BurdetteLamar · Pull Request #10198
• Bug #20188: Module#const_source_location returns wrong information when real constant was defined but autoload is still ongoing
• YJIT: Fallback cfunc varg splat for ruby2_keywords by k0kubun · Pull Request #10226
• [DOC] Array doc by BurdetteLamar · Pull Request #10199
• Bug #20307: Hash#update from compare_by_identity hash can have unfrozen string keys
• Feature #20293: Add Warning.categories method that returns the warning category names
• Feature #20244: Show the conflicting another chdir block
• Feature #13557: there's no way to pass backtrace locations as a massaged backtrace
• Feature #20306: Add rb_free_at_exit_p
• Feature #20265: Deprecate and remove rb_newobj and rb_newobj_of
• YJIT: Let yjit_perf.py support perf with Python disabled by k0kubun · Pull Request #10246
• Bug #19918: Should a[&b]=c be syntax valid?
• Bug #20218: aset/masgn/op_asgn with keyword arguments
• Update unmerged RBS to debug EBADF error by soutaro · Pull Request #10275
• YJIT: Support arity=-2 cfuncs by XrXr · Pull Request #10268
• Feature #20205: Enable frozen_string_literal by default
• Misc #20279: Is the implementation of respond_to_missing? in BasicObject documentation correct?
• [DOC] Unify Doxygen formats by k0kubun · Pull Request #10285
• YJIT: Get rid of Type::TProc by k0kubun · Pull Request #10287
• Bug #20339: Parser segfault with ractor comment
• Declare all jobs in matrix.include by k0kubun · Pull Request #10327
• Feature #20275: Avoid extra backtrace entries for rescue and ensure
• Bug #20388: super + ruby2_keywords isn't working
• Bug #20389: Issue with chilled strings and mutability after 12be40ae
• Bug #20310: ASAN fake stacks need to be marked during GC for non-current execution context
• YJIT: Propagate Array, Hash, and String classes by k0kubun · Pull Request #10323
• Bug #15784: RUBY_DEBUG=gc_stress dumps core
• Bug #20392: Delegate super calls with a block
• Don't clear pending interrupts in the parent process. by ioquatix · Pull Request #10365
• Feature #20390: Issue with StringIO and chilled strings
• Bug #20398: heap-buffer-overflow in numeric literal parsing
• YJIT: add iseq_alloc_count to stats by maximecb · Pull Request #10398
• Bug #20402: Double-free in TestIseqLoad#test_stressful_roundtrip
• Bug #20403: TestResolvDNS#test_query_ipv4_address_truncated_tcp_fallback can flake if TCP port is in use
• Feature #20331: Should parser warn hash duplication and when clause?
• Feature #20345: Add --target-rbconfig option to mkmf
• YJIT: A64: Use ADDS/SUBS/CMP (immediate) when possible by XrXr · Pull Request #10402
• Revert "Migrate auto_request_review back to CODEOWNERS" by k0kubun · Pull Request #10441
• YJIT: Suppress warn(static_mut_refs) by k0kubun · Pull Request #10440
• YJIT: Let sp_opnd take the number of slots by k0kubun · Pull Request #10442
• YJIT: update code optimization tips in yjit.md by maximecb · Pull Request #10445
• Bug #20342: Top level public, private and ruby2_keywords do not work in wrapped load
• Feature #20329: Clean up --dump sub-options
• Enumerator should use a non-blocking fiber. by ioquatix · Pull Request #10478
• Prefer to use Fiber#transfer in scheduler implementation. by ioquatix · Pull Request #10479
• Enumerator should use a non-blocking fiber, change rb_fiber_new to be non-blocking by default. by ioquatix · Pull Request #10481
• Bug #20417: RubyVM::AbstractSyntaxTree.parse warns about block local variables
• YJIT: x64: Remove register shuffle with opt_and and friends by XrXr · Pull Request #10498
• Bug #20423: Anonymous block forwarding shouldn't work with ... but it does
• Feature #15554: warn/error passing a block to a method which never use a block
• Feature #13077: [PATCH] introduce String#fstring method
• YJIT: A64: Avoid intermediate register in opt_and and friends by XrXr · Pull Request #10509
• Bug #20325: Enumerator.product.size bug with zero * infinite enumerators
• YJIT: End send fallback blocks by k0kubun · Pull Request #10539
• Bug #20418: StringIO#read(..., buffer) doesn't preserve buffer's encoding
• Feature #20429: Emit a performance warning when specially optimized core methods are redefined
• Feature #20335: Thread.each_caller_location should accept the same arguments as caller and caller_locations
• Feature #20347: Separate docs task from all
• Fiber#raise recursively raises on nested resuming_fiber. by ioquatix · Pull Request #10482
• YJIT: Optimize local variables when EP == BP by k0kubun · Pull Request #10487
• Feature #18576: Rename ASCII-8BIT encoding to BINARY
• YJIT: Fix canary crash with Array#<< by XrXr · Pull Request #10568
• chore: remove repetitive words by careworry · Pull Request #10573
• Refer autoconfigured endian macro by nobu · Pull Request #10572
• Revert "YJIT: Optimize local variables when EP == BP" by XrXr · Pull Request #10584
• YJIT: Fix shrinking block with assumption too much by XrXr · Pull Request #10585
• Sync IRB f9347b1 by st0012 · Pull Request #10611
• YJIT: Add a specialized codegen function for Class#superclass. by nirvdrum · Pull Request #10613
• Bug #20450: Ruby 3.3.1 broken with bootsnap
• YJIT: Optimize local variables when EP == BP (take 2) by k0kubun · Pull Request #10607
• Bug #20453: Pointer being freed was not allocated in Regexp timeout
• YJIT: Try splitting getlocal/setlocal blocks by k0kubun · Pull Request #10648
• Fixed a value specified for OPT_THREADED_CODE by jinroq · Pull Request #10657
• YJIT: Add specialized codegen function for TrueClass#=== by rwstauner · Pull Request #10640
• YJIT: Expand codegen for TrueClass#=== to FalseClass and NilClass by rwstauner · Pull Request #10679
• YJIT: Pass options to test-spec with -T by XrXr · Pull Request #10599
• YJIT: print msg to stderr when RubyVM::YJIT.disasm not available by maximecb · Pull Request #10688
• YJIT: Fix Struct accessors not firing tracing events by XrXr · Pull Request #10690
• Bug #19974: OpenSSL::PKCS7 generates SegFault when parsing invalid data
• Bug #20168: Process won't exit when Ractor.select waiting a Ractor
• [DOC] rewrite inject doc by pragdave · Pull Request #10009
• YJIT: Fix comment and counter in rb_yjit_invalidate_ep_is_bp() by XrXr · Pull Request #10722
• Bug #20474: Heredoc common leading whitespace calculation question
• Bug #20421: String#index and String#byteindex don't clear $~ when offset > size (or bytesize)
• Bug #7759: Marshal.load is not documented to be dangerous
• Bug #20481: Untrusted Marshal data can overwrite class/module instance variables
• Bug #20482: nil variables in a guard clause of a standalone => or in expression

Note: This list is automatically generated by tool/gen-github-release.rb. Because of this, some commits may be missing.

Full Changelog

https://github.com/ruby/ruby/compare/v3_3_0...v3_4_0_preview1