JuliaLang / julia

78%

LAST BUILD ON BRANCH master

branch: SELECT

CHANGE BRANCH

x

• No branch selected
• 92524901548ddad4a7529f00be47d2e1d9d80bae
• HEAD
• IanButterworth:ib/base_cov_fix
• IanButterworth:ib/coverage_fix
• IanButterworth:ib/coverage_timeout_bump
• IanButterworth:ib/force_color
• amitm/topodoc
• anj/factorize
• anj/triangle
• backports-release-1.8
• backports-release-julia-1.10
• backports-release-julia-1.12
• bigint
• cjh/linalg-refactor
• db/trsv
• dpa/1.8-html-docs-suitesparse
• dpa/buildkite_tester_linux64
• dpa/ci-buildkite-coverage-worker-timeout
• dpa/coverage
• dpa/coverage-red-flag
• dpa/coverage-timeout
• dpa/coverage-update-rootfs-image
• dpa/fix-coverage
• dpa/fix-windows-docs---julia-1.11
• dpa/julia_test_verbose_logs_dir
• dpa/juliac-github-status-on-prs
• dpa/no-bb
• dpa/revert-trimming-test-julia-1.11
• dpa/schedule
• dpa/scheduled-pipeline
• dpa/signature-coverage-buildkite
• dpa/trigger-ci---release-julia-1.10
• dpa/update-rootfs-images
• dpa/upload-linux-assert-builds-to-s3
• ericphanson:eph/no-waiting-on-tests
• fe/r2
• gba/test-juliac
• giordano:mg/codecov
• heads/v1.5.0
• heads/v1.5.0-beta1
• heads/v1.5.0-rc1
• heads/v1.5.0-rc2
• heads/v1.5.1
• heads/v1.5.2
• heads/v1.5.3
• heads/v1.6.0-rc1
• ib/coverage_mark_broken
• ib/fix_coverage
• jb/checked_int_trunc
• jb/functions
• jb/tupleoverhaul
• jcb/docfixes
• jn/callmore
• jn/over-coverage-fix
• jn/stream.open
• kf/libunwind-gcc-fix
• kf/usemanifest
• ksh/lapack
• kshyatt:ib/base_cov_fix
• main
• master
• master2
• master3
• mb/abstractsmarts
• multithreading
• patch-1
• refs/pull/36464/merge
• release-0.3
• release-0.4
• release-1.3
• release-1.4
• release-1.5
• release-1.6
• release-1.8
• sf/build_sysimg3.0
• sf/cpuset_limited_filter
• sf/optimization
• sf/perfcodespeed
• showundef
• sjk/bigfloat-linspace
• sjk/faster-sumabs2
• sjk/isnan
• sjk/sparse-reductions
• sk/mathconst
• sk/utf16helpers
• teh/cartesian_iteration2
• teh/malloclog
• teh/nosys
• teh/pkg_test
• tk/backports-0.4.2
• tk/git-advice
• tk/travis-fastfail
• topic/linspace
• topic/logspace
• topic/matdiv
• topic/parallelmake
• topolarity:ct/debug-juliac
• v1.10.0
• v1.10.0-alpha1
• v1.10.0-beta2
• v1.10.0-beta3
• v1.10.0-rc1
• v1.10.0-rc2
• v1.10.0-rc3
• v1.10.1
• v1.10.10
• v1.10.11
• v1.10.2
• v1.10.3
• v1.10.4
• v1.10.5
• v1.10.6
• v1.10.7
• v1.10.9
• v1.11.0
• v1.11.0-alpha1
• v1.11.0-alpha2
• v1.11.0-beta1
• v1.11.0-beta2
• v1.11.0-rc1
• v1.11.0-rc2
• v1.11.0-rc3
• v1.11.0-rc4
• v1.11.1
• v1.11.2
• v1.11.3
• v1.11.4
• v1.11.5
• v1.11.6
• v1.11.7
• v1.11.8
• v1.11.9
• v1.12.0
• v1.12.0-beta3
• v1.12.0-rc2
• v1.12.0-rc3
• v1.12.1
• v1.12.2
• v1.12.3
• v1.12.4
• v1.12.5
• v1.12.6
• v1.8.5
• v1.9.0
• v1.9.0-beta1
• v1.9.0-beta2
• v1.9.0-beta3
• v1.9.0-beta4
• v1.9.0-rc1
• v1.9.0-rc2
• v1.9.1
• v1.9.2
• v1.9.3
• v1.9.4
• vs/sparsemodule
• wordcount-example
• yyc/genstdlib-module
• yyc/misc

Build # 1562

Build Type

push

buildkite

Committed by web-flow

Commit Message

staticdata: fix serialization ordering for cyclic datatype supertypes (#62050)

Change the deferred enqueue for `super` types to utilize a worklist
`deferred_supers` that `jl_serialize_reachable` drains only once the
pre-order recursion has fully unwound. This allows us to properly defer
exploration of the supertype and all of its type parameters. This
extends the previous fix in JuliaLang/julia#52752, which did not handle
the case that recursion occurs through supertype parameters.

These kinds of "special case" / "defer sometimes" behaviors are a great
place for bugs to hide anyway, so it is better to have one uniform
behavior to handle `super` in the serializer IMO. Case in point: Also
fix (1) a bug with `must_be_new_dt` not accounting for `super` deferral
and (2) a corner case where items with identical supertypes would be
deserialized incorrectly due to how we iterate `uniquing_super`, spotted
by inspection by Claude Fable 5 and a more serious problem now that this
makes deferrals more common.

Fixes #62047

Bug analyzed and fix implemented by Claude Fable 5
<noreply@anthropic.com> 🤖 🪄

I did check its work / revise the changes and I believe its analysis /
fix is correct.

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

Coverage Stats

65957 of 84857 relevant lines covered (77.73%)

22815628.9 hits per line