Ruby Issue Tracking System

Today

04:59 PM Ruby Bug #21958 (Closed): Snapshot CI is breaking now

Fixed it by https://github.com/ruby/ruby/pull/16482 and https://github.com/ruby/ruby/pull/16483. k0kubun (Takashi Kokubun)

04:44 PM Ruby Revision 2ca2865a (git): ZJIT: Remove eager nil-fill of locals in JIT-to-JIT calls (#16544)

The caller in gen_send_iseq_direct was eagerly writing Qnil to all
non-parameter local slots of the callee's frame before every JIT-to-JIT
call. This is unnecessary because compile_jit_entry_state already
initializes non-parameter locals... k0kubun (Takashi Kokubun)

04:44 PM Ruby Revision ff89d56b (git): Pass reg_cfp instead of iseq to vm_search_method and related functions (#16541)

Change vm_search_method_fastpath, vm_search_method, vm_method_cfunc_is,
opt_equality, vm_opt_neq, vm_opt_nil_p, vm_opt_not, and vm_objtostring
to take a CFP instead of an iseq. The iseq is now read lazily from
reg_cfp->iseq only when the... k0kubun (Takashi Kokubun)

05:07 AM Ruby Revision 48f86bfa (git): [ruby/resolv] Increase DNS resolver timeouts in

test_multiple_servers_with_timeout_and_truncated_tcp_fallback
(https://github.com/ruby/resolv/pull/125)
The test uses intentionally tight timeouts to trigger TCP fallback, but
0.1s/0.2s can be too tight on loaded CI machines, causing th... k0kubun (Takashi Kokubun)

04:53 AM Ruby Revision 0bebc77a (git): Retry IO.select with a longer timeout on flaky platforms (#16542)

On MinGW (Windows), pipe readiness after syswrite may not be
immediately visible to IO.select with timeout 0, causing a
spurious nil return. Retry with a longer timeout to handle this.
https://github.com/ruby/ruby/actions/runs/235192695... k0kubun (Takashi Kokubun)

03/24/2026

11:31 PM Ruby Revision 9a857c54 (git): ZJIT: Generate polymorphic branches for skewed polymorphic getivar (#16533)

This PR lets skewed polymorphic callsites to generate polymorphic code. Currently, skewed polymorphic callsites are considered monomorphic and don't use https://github.com/ruby/ruby/pull/16476, but doing so seems to help lobsters. k0kubun (Takashi Kokubun)

09:47 PM Ruby Revision 4383b371 (git): ZJIT: Update trace-exits docs for Fuchsia trace format

The Stackprof-based documentation was outdated after the switch to
Fuchsia Trace Format (.fxt) output. Update with Perfetto UI and
trace_processor_shell workflow, including a working example and
SQL query for finding hottest exit locations. k0kubun (Takashi Kokubun)

09:47 PM Ruby Revision 30dcc2a0 (git): ZJIT: Mention Perfetto UI for SQL queries in trace-exits docs

k0kubun (Takashi Kokubun)

09:35 PM Ruby Revision c819029d (git): Fix double-prefixed path in outdate-bundled-gems.rb (#16531)

When building with a relative srcdir (e.g., ../configure from a build
directory), default_gem? was called with an already-prefixed path on
line 149. Since default_gem? internally calls prefixed() again, this
resulted in a double prefix (... k0kubun (Takashi Kokubun)

04:47 PM Ruby Revision e74823a0 (git): ZJIT: Skip too-complex shapes in polymorphic getivar (#16526)

Too-complex shapes use hash tables for ivar storage, and
rb_shape_get_iv_index() doesn't work for them (it asserts in debug
builds). Without this check, the polymorphic getinstancevariable
optimization incorrectly returns nil for ivars o... k0kubun (Takashi Kokubun)