Fix tautological assertions in test_each_value in test_hash.rb
Both assertions tested [].length == 0 rather than res.length, so the test passed even when each_value yielded nothing.
[].length == 0
res.length
each_value
Bump taiki-e/install-action
Bumps the github-actions group with 1 update in the / directory: taiki-e/install-action.
Updates taiki-e/install-action from 2.79.8 to 2.79.9
taiki-e/install-action
ZJIT: Replace rpo with reverse_post_order (#17122)
After a team discussion about naming, we decided that rpo should be expanded to reverse_post_order. This PR replaces the functions named rpo in HIR and LIR.
rpo
reverse_post_order
ZJIT: Add for_each_operand and for_each_operand_mut to LIR
Replace the operand iterator structs with the simpler, shorter visit-closure pattern that we also have in HIR.
ZJIT: Remove defunct LIR LiveReg instruction
We don't use this anymore since we switched to global linear scan register allocation.
[ruby/rubygems] Remove version constraints on dev dependencies
Bundler doesn't need to have an opinion on the current version of these tools. We can just include them without specific constraint, and if the user doesn't want the most recent version for some reason they...
Suppress -Warray-bounds false positive in RCLASS_WRITE_SUPER (#16990)
After 4bb1977b78 switched RCLASS_PRIME_CLASSEXT_WRITABLE_P to FL_TEST_RAW, GCC lost the RB_FL_ABLE-derived hint that klass is not a special const, and started warning that RCLASS_EXT_PRIME(klass) (used by the inlined...
ZJIT: LIR CCall survivors: Use one stack layout across arches
Previously, the last push, in case the total number of pushes was odd, ended up in a different location relative to the stack top on x64 versus A64. On A64, it ends up at sp[0], whereas on x64 it ends up at...
sp[0]
[Box] Invalidate callable me on per-box origin IClasses
[Box] Invalidate the method cache of the owner too
The owner could be different from the housing class when the housing module is prepended.
View all revisions | View revisions
Also available in: Atom