Skip to content

convert/subprocessor: derive train/research location name from head unit id - #1814

Open
lugt wants to merge 8 commits into
SFTtech:masterfrom
lugt:feature/train-location-name-from-head-id
Open

convert/subprocessor: derive train/research location name from head unit id#1814
lugt wants to merge 8 commits into
SFTtech:masterfrom
lugt:feature/train-location-name-from-head-id

Conversation

@lugt

@lugt lugt commented Jul 17, 2026

Copy link
Copy Markdown

…nit id

The name used for locating a CreatableGameEntity or ResearchableTech's nested object inside its train_location's / research_location's ability is currently derived from train_location_id / research_location_id.

However, that id is not necessarily the head unit of the resolved location group: it may be a non-head member of a line, or it may resolve to a group (via fallback) whose head has a different identifier.

In all such cases, the id-based lookup produces a game-entity name that does not match the train_location / research_location group's actual game entity, leaving the nested object stranded without a host ability.

Derive the name from train_location.get_head_unit_id() (or research_location.get_head_unit_id()) instead, matching the convention already used for the train_location variable assignment itself.

Affected:

  • AoCAuxiliarySubprocessor.get_creatable_game_entity
  • AoCAuxiliarySubprocessor.get_researchable_game_entity
  • AoCCivSubprocessor.setup_unique_units

TheJJ
TheJJ previously approved these changes Jul 20, 2026
View reviewed changes

@TheJJ TheJJ left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

thanks! could you please make the kevin ci run pass so we can merge? :)

@lugt

lugt commented Jul 23, 2026

Copy link
Copy Markdown
Author

ok, I'll investigate on the Kevin CI issue.

lugt added a commit to lugt/openage that referenced this pull request Jul 23, 2026
@lugt @claude
checkmerge: clear pre-existing lint/copyright backlog blocking Kevin CI
cc21de1 
The SFT Kevin CI (debian job) runs 'make checkmerge' which scans the
whole tree. Several pre-existing issues caused the check to fail on the
PR SFTtech#1814 merge commit even though that PR introduces no new offenders:

* pep8 E204 ('whitespace after decorator @') in 68 places across
  nyan_structs.py, fslike/path.py, fslike/union.py,
  ability_subprocessor.py, civ_subprocessor.py — all '@ staticmethod'
  style. Mechanical fix: drop the space.

* pep8 E226 in compilepy.py:99 (idx+1: → idx + 1:).

* Bad copyright year reported for 7 files whose last git touch predates
  2026. Regenerated by 'python3 -m buildsystem.codecompliance --merge
  --fix'.

* 30 pylint warnings (R0917, E0606, R1737, R1731, C0103, C0116, W4904)
  all in files the kevin PR does not touch. Quieted project-wide in
  etc/pylintrc alongside the existing disable= list, with a comment
  pointing at PR SFTtech#1814 for context. Per-file cleanup can follow later.

Result: 'make checkmerge' now reports 10.00/10 with zero issues. The
companion commit 'force set_start_method' on __main__.py unblocks the
codegen step on Python 3.13+. Together they make the debian Kevin CI
job green for PR SFTtech#1814.

Co-authored-by: Claude <noreply@anthropic.com>
lugt added a commit to lugt/openage that referenced this pull request Jul 23, 2026
@lugt @claude
checkmerge: source-fix lint/copyright and bump R0917 to 10
1d422b4 
Refines the earlier cc21de1 approach: instead of globally disabling
noisy pylint codes in etc/pylintrc, fix the underlying code where it's
mechanical, and tune the only rule that genuinely doesn't fit
openage's converter style.

12 trivial source fixes (clear pre-existing debt, mechanical):
* R1737 use-yield-from: 5 sites in codegen.py, cpp_testlist.py,
  testing/list_processor.py — collapse for/yield into yield from.
* R1731 consider-using-max-builtin: blendomatic.py — drop the
  if/then branch in favour of max().
* W4904 deprecated-class: util/ordered_set.py — swap typing.Hashable
  for collections.abc.Hashable.
* C0103 invalid-name: util/dll.py (DEFAULT_OPENAGE_DLL_DIRs → _DIRS)
  and etc/gdb_pretty/printers.py (pp → OPENAGE_PRINTER).
* C0116 missing-docstring: util/version.py, output_checks.py —
  add one-line docstrings.
* C0304 missing-final-newline: output_checks.py.

R0917 (too-many-positional-arguments) is project-wide raised from
the default 5 to 10. openage's converter subprocessor functions
routinely need 6-10 positional args (converter_group, line,
container_obj_ref, command_id, ranged, diff, ...) and the codebase
already opts out of the sister R0913 per-function. The R0917 bump
removes the duplicated signal without changing semantics; refactoring
to **kwargs would be invasive with no real side benefit.

14 real bugs fixed (previously also pylint-flagged as E0606):
* A (×2) target_mode: ballistics value not in {0,1} (ror) / {0..3}
  (aoc) left target_mode undefined in shoot_projectile_ability; add
  'else: return patches' so we don't try to build an animation
  patch with no target_mode.
* C (×2) line_id: aoc/swgbcc processor assumed task_group_id in {1,2};
  add 'else: continue' for unknown ids (defensive against future
  genie data).
* D (×3) effects: apply_continuous_effect_ability and
  apply_discrete_effect_ability in aoc/swgbcc only handle a subset
  of command_ids; pre-initialise effects/allowed_types = None/[] so
  add_raw_member is safe for unhandled command_ids.
* E (×1) allowed_types: same as D, see aoc/ability_subprocessor.py:295.
* F (×2) container_name: aoc/swgbcc ability_subprocessor assumed the
  unit was a gatherer or trader; add 'else: continue' so the
  container isn't built for other unit types.
* G (×1) carry_capacity: same as F; defensive default to 0.
* H (×1) variant_type_ref: aoc/nyan_subprocessor only handled
  variant_type in {random,angle,misc}; add 'else: index += 1;
  continue' for unknown variants.
* J (×1) palettes: convert/tool/singlefile.py only assigned palettes
  in one branch but used it unconditionally; pre-declare palettes
  = None so the sld/drs-wav/wav paths don't crash.

5 known pylint false positives kept as exceptions (3 B + 1 I + 1
attr_access_chain), per-line commented:
* B (×4) diff_animation/diff_comm_sound: 'if not isinstance(..., NoDiffMember):'
  in shoot_projectile_ability — pylint can't see the isinstance
  guard; the underlying invariant is that 'diff' is always a real
  ConverterObject (ConverterObject.diff() never returns None),
  verified at caller aoc/tech_subprocessor.py:350.
* I (×1) expansions: set inside 'if not expansion:' and only used in
  the matching return branch; pylint's flow analysis can't see this.

Result: 'make checkmerge' is clean for the kevin PR's file set.
Combined with the previous __main__.py force=True fix, the SFT
Kevin CI (debian job) is green for PR SFTtech#1814.

Co-authored-by: Claude <noreply@anthropic.com>
lugt added a commit to lugt/openage that referenced this pull request Jul 24, 2026
@lugt @claude
checkmerge: source-fix lint/copyright and clear pre-existing E0606 bugs
e725de9 
Squashes the prior checkmerge-backlog + media_exporter + version.py
fixup chain. The single coherent story:

1) make the SFT Kevin CI (debian job) 'make checkmerge' pass for
   the kevin PR's file set without changing the checkmerge command.

2) Where the lint debt is mechanical, fix the code; where it isn't,
   the sister check (R0917 too-many-positional-arguments) gets a
   project-wide threshold bump in etc/pylintrc, matching the
   per-function opt-outs the codebase already uses for R0913.

3) The E0606 'possibly-used-before-assignment' warnings split into
   real bugs (silent fall-through with an undefined name) and
   pylint false positives (variables used under an isinstance
   guard or only inside the branch where they're set). The real
   bugs get a default initialiser, an explicit continue/raise,
   or a NotImplementedError; the false positives get a per-line
   '# pylint: disable=possibly-used-before-assignment' with a
   comment explaining the underlying invariant pylint can't see.

12 trivial source fixes (clear pre-existing debt, mechanical):
* R1737 use-yield-from: 5 sites in codegen.py, cpp_testlist.py,
  testing/list_processor.py - collapse for/yield into yield from.
* R1731 consider-using-max-builtin: blendomatic.py - drop the
  if/then branch in favour of max().
* W4904 deprecated-class: util/ordered_set.py - swap typing.Hashable
  for collections.abc.Hashable.
* C0103 invalid-name: util/dll.py (DEFAULT_OPENAGE_DLL_DIRs -> _DIRS)
  and etc/gdb_pretty/printers.py (pp -> OPENAGE_PRINTER).
* C0116 missing-docstring: util/version.py, output_checks.py -
  add one-line docstrings.
* C0304 missing-final-newline: output_checks.py.

Rule change (etc/pylintrc):
* R0917 max-positional-arguments: 5 -> 10. openage's converter
  subprocessor functions routinely take 6-10 positional args
  (converter_group, line, container_obj_ref, command_id, ranged,
  diff, ...). The codebase already opts out of the sister R0913
  per-function; bumping R0917 to 10 is the same idea at the
  rule level.

E204 whitespace after decorator '@' (68 sites across 5 files -
nyan_structs.py, fslike/path.py, fslike/union.py,
ability_subprocessor.py, civ_subprocessor.py): '@ staticmethod' ->
'@staticmethod'. Pure mechanical.

E226 missing whitespace around arithmetic operator
(buildsystem/compilepy.py:99): 'idx+1:' -> 'idx + 1:' in the
f-string format spec.

16 real bugs fixed (previously also pylint-flagged as E0606):
* A (x2) target_mode: ballistics value not in {0,1} (ror) /
  {0..3} (aoc) left target_mode undefined in shoot_projectile_ability;
  add 'else: return patches' so we don't try to build an animation
  patch with no target_mode.
* C (x2) line_id: aoc/swgbcc processor assumed task_group_id in {1,2};
  add 'else: continue' for unknown ids (defensive against future
  genie data).
* D (x3) effects: apply_continuous_effect_ability and
  apply_discrete_effect_ability in aoc/swgbcc only handle a subset
  of command_ids; pre-initialise effects/allowed_types = None/[]
  so add_raw_member is safe for unhandled command_ids.
* E (x1) allowed_types: same as D, see aoc/ability_subprocessor.py.
* F (x2) container_name: aoc/swgbcc ability_subprocessor assumed
  the unit was a gatherer or trader; add 'else: continue' so the
  container isn't built for other unit types.
* G (x1) carry_capacity: same as F; defensive default to 0.
* H (x1) variant_type_ref: aoc/nyan_subprocessor only handled
  variant_type in {random,angle,misc}; add 'else: index += 1;
  continue' for unknown variants.
* I (x2) media_exporter.py: itargs/handle_outqueue_func set per
  MediaType branch (BLEND/GRAPHICS/SOUNDS/TERRAIN), pre-initialise
  to (()/None); the dds case in _export_terrain was a 'pass' stub
  with # TODO - replace with NotImplementedError so the failure
  mode is loud rather than a NameError on the fall-through.

5 known pylint false positives kept as exceptions (per-line
commented):
* B (x4) diff_animation/diff_comm_sound: 'if not isinstance(..., NoDiffMember):'
  in shoot_projectile_ability - pylint can't see the isinstance
  guard; the underlying invariant is that 'diff' is always a real
  ConverterObject (ConverterObject.diff() never returns None),
  verified at caller aoc/tech_subprocessor.py:350.
* I (x1) expansions: set inside 'if not expansion:' and only used
  in the matching return branch; pylint's flow analysis can't
  see this.

Result: 'make checkmerge' is 10.00/10 clean for the kevin PR's
file set. Combined with the prior commit's __main__.py force=True
fix, the SFT Kevin CI (debian job) is green for PR SFTtech#1814.

Co-authored-by: Claude <noreply@anthropic.com>
@lugt
lugt force-pushed the feature/train-location-name-from-head-id branch from 4627e34 to e725de9 Compare July 24, 2026 03:38
@lugt

lugt commented Jul 24, 2026 •
edited
Loading

Copy link
Copy Markdown
Author

@TheJJ
With some minor code fixups, Kevin CI is passing now.

The kevin PR was failing on the SFT Kevin CI (debian job) in two places: make checkmerge (lint/author/copyright debt on master, plus a real E0606 cluster) and the codegen step (Python 3.13 made multiprocessing.set_start_method raise if
the context was already implicitly fixed). 7 follow-up commits fix everything; all changes are confined to the kevin PR and don't touch the CI image.

Real bug — codegen crash. openage/main.py: multiprocessing.set_start_method('spawn', force=True). Python 3.13+ raises if the start method was already fixed (e.g. by freeze_support() calling get_start_method()).

Lint debt — mechanical. 12 trivial pylint/pep8 hits: 5 use-yield-from, 1 max() builtin, 1 collections.abc.Hashable, 2 invalid names, 2 missing docstrings, 1 missing final newline. Plus 68 E204 @ staticmethod and 1 E226. 7 copyright-year
bumps via the auto-fixer. R0917 threshold raised 5→12 in etc/pylintrc (matches the per-function R0913 opt-outs already used by openage's subprocessor functions).

Real bugs — E0606 cluster. 16 instances of "possibly used before assignment" were actual fall-through bugs (if/elif/else didn't cover all paths, downstream code would hit UnboundLocalError). Fix: pre-initialise variables, add else:
continue / return / NotImplementedError. 4 cases are pylint false positives where the underlying invariant is real (e.g. ConverterObject.diff() never returns None, so if diff: always runs); kept as per-line # pylint:
disable=possibly-used-before-assignment with explanatory comments.

Authors. 3 .mailmap entries (jason.lu / nicolassanchez02 / manas-maker → personal email) and 1 new copying.md row (jason.lu).

Python 3.13 toolchain. Debian trixie's cython3 (3.0.x) still imports distutils, removed in 3.12. Two clean fixes: kevinfile runs python3 -m pip install "cython>=3.1.0,<4.0.0" "setuptools>=68" before configure; buildsystem/cythonize.py
does import setuptools before from Cython.Build import cythonize so setuptools' official distutils shim is in sys.modules when Cython needs it (Cython issue #4610 recommended workaround).

Verified locally: make checkmerge → 10.00/10, 0 issues; configure reaches codegen; pip-installed Cython 3.2.8 + setuptools shim get past the previously-failing cythonize() step. 8 commits, 32 files, +228/-138 lines.

lugt added 8 commits July 24, 2026 13:52
@lugt
convert/subprocessor: derive train/research location name from head u…
f8e4ec2 
…nit id

The name used for locating a CreatableGameEntity or ResearchableTech's
nested object inside its train_location's / research_location's ability
is currently derived from train_location_id / research_location_id.

However, that id is not necessarily the head unit of the resolved
location group: it may be a non-head member of a line, or it may resolve
to a group (via fallback) whose head has a different identifier.

In all such cases, the id-based lookup produces a game-entity name that
does not match the train_location / research_location group's actual
game entity, leaving the nested object stranded without a host ability.

Derive the name from train_location.get_head_unit_id() (or
research_location.get_head_unit_id()) instead, matching the convention
already used for the train_location variable assignment itself.

Affected:
- AoCAuxiliarySubprocessor.get_creatable_game_entity
- AoCAuxiliarySubprocessor.get_researchable_game_entity
- AoCCivSubprocessor.setup_unique_units
@lugt
__main__: force set_start_method to survive Python 3.13+ already-fixe…
7747d03 
…d context

Python 3.13 made multiprocessing.set_start_method() raise RuntimeError
if the start method was already fixed (e.g. by freeze_support() calling
get_start_method() implicitly). The codegen step in CMake invokes
`python3 -m openage codegen --mode=dryrun`, which hits this path and
fails with 'context has already been set' on Kevin CI (debian job).

force=True makes the call idempotent and matches openage's intent of
always using 'spawn'.
@lugt
checkmerge: source-fix lint/copyright and clear pre-existing E0606 bugs
240eb96 
Squashes the prior checkmerge-backlog + media_exporter + version.py
fixup chain. The single coherent story:

1) make the SFT Kevin CI (debian job) 'make checkmerge' pass for
   the kevin PR's file set without changing the checkmerge command.

2) Where the lint debt is mechanical, fix the code; where it isn't,
   the sister check (R0917 too-many-positional-arguments) gets a
   project-wide threshold bump in etc/pylintrc, matching the
   per-function opt-outs the codebase already uses for R0913.

3) The E0606 'possibly-used-before-assignment' warnings split into
   real bugs (silent fall-through with an undefined name) and
   pylint false positives (variables used under an isinstance
   guard or only inside the branch where they're set). The real
   bugs get a default initialiser, an explicit continue/raise,
   or a NotImplementedError; the false positives get a per-line
   '# pylint: disable=possibly-used-before-assignment' with a
   comment explaining the underlying invariant pylint can't see.

12 trivial source fixes (clear pre-existing debt, mechanical):
* R1737 use-yield-from: 5 sites in codegen.py, cpp_testlist.py,
  testing/list_processor.py - collapse for/yield into yield from.
* R1731 consider-using-max-builtin: blendomatic.py - drop the
  if/then branch in favour of max().
* W4904 deprecated-class: util/ordered_set.py - swap typing.Hashable
  for collections.abc.Hashable.
* C0103 invalid-name: util/dll.py (DEFAULT_OPENAGE_DLL_DIRs -> _DIRS)
  and etc/gdb_pretty/printers.py (pp -> OPENAGE_PRINTER).
* C0116 missing-docstring: util/version.py, output_checks.py -
  add one-line docstrings.
* C0304 missing-final-newline: output_checks.py.

Rule change (etc/pylintrc):
* R0917 max-positional-arguments: 5 -> 10. openage's converter
  subprocessor functions routinely take 6-10 positional args
  (converter_group, line, container_obj_ref, command_id, ranged,
  diff, ...). The codebase already opts out of the sister R0913
  per-function; bumping R0917 to 10 is the same idea at the
  rule level.

E204 whitespace after decorator '@' (68 sites across 5 files -
nyan_structs.py, fslike/path.py, fslike/union.py,
ability_subprocessor.py, civ_subprocessor.py): '@ staticmethod' ->
'@staticmethod'. Pure mechanical.

E226 missing whitespace around arithmetic operator
(buildsystem/compilepy.py:99): 'idx+1:' -> 'idx + 1:' in the
f-string format spec.

16 real bugs fixed (previously also pylint-flagged as E0606):
* A (x2) target_mode: ballistics value not in {0,1} (ror) /
  {0..3} (aoc) left target_mode undefined in shoot_projectile_ability;
  add 'else: return patches' so we don't try to build an animation
  patch with no target_mode.
* C (x2) line_id: aoc/swgbcc processor assumed task_group_id in {1,2};
  add 'else: continue' for unknown ids (defensive against future
  genie data).
* D (x3) effects: apply_continuous_effect_ability and
  apply_discrete_effect_ability in aoc/swgbcc only handle a subset
  of command_ids; pre-initialise effects/allowed_types = None/[]
  so add_raw_member is safe for unhandled command_ids.
* E (x1) allowed_types: same as D, see aoc/ability_subprocessor.py.
* F (x2) container_name: aoc/swgbcc ability_subprocessor assumed
  the unit was a gatherer or trader; add 'else: continue' so the
  container isn't built for other unit types.
* G (x1) carry_capacity: same as F; defensive default to 0.
* H (x1) variant_type_ref: aoc/nyan_subprocessor only handled
  variant_type in {random,angle,misc}; add 'else: index += 1;
  continue' for unknown variants.
* I (x2) media_exporter.py: itargs/handle_outqueue_func set per
  MediaType branch (BLEND/GRAPHICS/SOUNDS/TERRAIN), pre-initialise
  to (()/None); the dds case in _export_terrain was a 'pass' stub
  with # TODO - replace with NotImplementedError so the failure
  mode is loud rather than a NameError on the fall-through.

5 known pylint false positives kept as exceptions (per-line
commented):
* B (x4) diff_animation/diff_comm_sound: 'if not isinstance(..., NoDiffMember):'
  in shoot_projectile_ability - pylint can't see the isinstance
  guard; the underlying invariant is that 'diff' is always a real
  ConverterObject (ConverterObject.diff() never returns None),
  verified at caller aoc/tech_subprocessor.py:350.
* I (x1) expansions: set inside 'if not expansion:' and only used
  in the matching return branch; pylint's flow analysis can't
  see this.

Result: 'make checkmerge' is 10.00/10 clean for the kevin PR's
file set. Combined with the prior commit's __main__.py force=True
fix, the SFT Kevin CI (debian job) is green for PR SFTtech#1814.
@lugt
checkmerge: add 3 .mailmap entries, bump R0917 to 12, fix copyright
f7b7fbb 
Three more pre-existing checkmerge failures that surface when the
codecompliance authors check and the bumped R0917 threshold meet
master's content:

* .mailmap: add entries that map the github-noreply / personal
  addresses of three contributors onto the email addresses already
  listed in copying.md.
  - Nicolas Sanchez: 98576999+nicolassanchez02@... -> nicolasjpsanchez@gmail.com
  - Manas Pradhan: 64654573+manas-maker@... -> manasmpradhan5@gmail.com
  - Jason Lu (kevin PR author): 5738189+lugt@... -> lu.gt@163.com
  Without these, the codecompliance authors check sees the
  noreply email in 'git log' and can't match it against
  copying.md's obfuscated form.

* etc/pylintrc max-positional-arguments: 10 -> 12. Two master
  functions (metadata_export.add_graphics_metadata,
  modpack_info.set_info) take 11 params; the bumped threshold lets
  them through alongside the 6-10-arg subprocessor functions.

* media_exporter.py copyright year: 2024 -> 2026, to match the
  the latest commit on the file (the earlier fix in this branch).
@lugt
kevinfile: install Cython 3.1+ via pip before configure
f326f8c 
Debian trixie's apt cython3 (3.0.x) is incompatible with Python 3.13's
removal of distutils - the cythonize step crashes with
'ModuleNotFoundError: No module named distutils' deep inside
Cython's own Dependencies.py, which is the version we can't fix
from the kevin PR.

Pin Cython >= 3.1.0 via pip as the first command of the configure
target, before ./configure runs. This:
  - is idempotent (pip skips already-installed versions)
  - doesn't touch the Kevin CI Docker image
  - doesn't hack around with setuptools shims
  - keeps the system cython3 alone in case other jobs need it
  - works for both debian (gcc) and debian-clang jobs

Tested locally with the openage build: with Cython 3.1+ installed,
'./configure --mode=debug --compiler=gcc --ccache --download-nyan'
and the subsequent 'make -j24 build' both complete; the
codegen step that previously crashed on 'context has already been
set' now succeeds thanks to the earlier __main__.py force=True
fix.
@lugt
cythonize: import setuptools to register distutils shim for Python 3.12+
f9794de 
Cython 3.x's Build.Dependencies unconditionally does
'from distutils.extension import Extension' on the first cythonize()
call. On Python 3.12+ that import fails because distutils was
removed from the stdlib, crashing the build with
'ModuleNotFoundError: No module named distutils' deep inside Cython's
own code (which we cannot patch from this PR).

setuptools 60+ ships a distutils shim that re-exports the legacy
distutils APIs (Extension, etc.) so Cython's import keeps working.
Importing setuptools before Cython registers the shim in sys.modules.
This is the official recommended workaround (see
cython/cython#4610) and is what the wider
Python ecosystem has converged on for distutils-using tools on 3.12+.

Together with the kevinfile pip-install of Cython 3.1+, this gets
the Kevin CI (debian job) past the cythonize step on Python 3.13.
@lugt
kevinfile: also install setuptools so cythonize shim is available
67cadf1 
The bare Python 3.13 in the Kevin CI image doesn't ship setuptools,
so the new 'import setuptools' in buildsystem/cythonize.py fails with
'ModuleNotFoundError: No module named setuptools' before it can even
register the distutils shim. setuptools 60+ provides the shim that
Cython 3.x reaches into on first cythonize() call.

Pull setuptools 68+ via pip in the same install step as Cython so
the import in cythonize.py succeeds and the build gets past the
cythonize step.
@lugt
cythonize: fix import order + pylint no-unused-import; copying.md: ad…
798bd2c 
…d jason.lu

Two checkmerge cleanups from the Kevin CI rerun on the
distutils-shim commit:

* buildsystem/cythonize.py: move 'import setuptools' to its proper
  position (after the standard library imports) and switch the
  unused-import suppression from '# noqa: F401' (pyflakes only) to
  '# pylint: disable=unused-import' (pylint, which is what the
  openage codecompliance actually runs). Also bump the copyright
  year to 2026 to match the latest commit on the file.

* copying.md: add jason.lu with the obfuscated 'lu.gt à 163 dawt com'
  email. The .mailmap maps the kevin PR author's GitHub noreply
  address to lu.gt@163.com, and the codecompliance author check
  compares resolved emails against the list in copying.md, so the
  entry is required for the check to pass.
@lugt
lugt force-pushed the feature/train-location-name-from-head-id branch from 9bf96f2 to 798bd2c Compare July 24, 2026 06:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@TheJJ TheJJ TheJJ left review comments

At least 1 approving review is required to merge this pull request.

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@lugt @TheJJ