about summary refs log tree commit diff
path: root/test/python/test_groups.py
diff options
context:
space:
mode:
authorAlexander_Kabui <alexanderkabua@gmail.com>2026-10-07 06:13:50 +0300
committerAlexander_Kabui <alexanderkabua@gmail.com>2026-10-07 06:13:50 +0300
commit2a7895d3ba2e7fe4fa090473714fbb029e592d69 (patch)
tree42a621dc79f4d4e37b275422d82d9cad42437c25 /test/python/test_groups.py
parenta72ea07ea2e164cd5bbc0165e4bac95cbd7907b6 (diff)
downloadgenenetwork4-2a7895d3ba2e7fe4fa090473714fbb029e592d69.tar.gz
Cargo.lock: pin cxx-gen to 0.7.124, the cxx 1.0.122 family.
The C1.1 lock resolved cxx-gen to 0.7.198 (autocxx-build carries an
unbounded caret). cxx-gen 0.7.198 generates the newer cxxbridge symbol
scheme (Xapian$cxxbridge1$198$... with the per-crate disambiguator)
while cxxbridge-macro 1.0.122 — the pinned cxx 1.0.122 Rust side —
references the un-suffixed symbols: any binary that actually links the
xapian-rs bridge fails with undefined references. Nothing linked the
bridge until the native engine code, which is why the C1.1 gate
(build + full test, no xapian code) stayed green with 0.7.198.

0.7.124 is the resolution the fork itself ships in its own Cargo.lock
(xapian-rs @ 97fd343: cxx 1.0.122, cxxbridge-flags/macro 1.0.122,
cxx-gen 0.7.124) and the one the recipe lane proved hermetically.
The pin is exactly the watch item binding-report §6-2 carries
(unbounded carets drift off the 1.0.122 family).

Also resolves: codespan-reporting 0.13.1 -> 0.11.1, drops
unicode-width 0.2.2 (both were only pulled by 0.7.198). Link-failure
evidence: logs/c3-1.log (undefined Xapian$cxxbridge1$Enquire$set_query
vs nm on the built archive showing Xapian$cxxbridge1$198$Enquire$
set_query).
Diffstat (limited to 'test/python/test_groups.py')
0 files changed, 0 insertions, 0 deletions