Revert part of the markdown cleanup: the two TrainingAreaBase plugin
ICD docs and jenkins-ci-rme/README.md stay in the tree; the root
migration/validation reports remain removed. README.md pointers to
jenkins-ci-rme/README.md are restored.
Drop the 13 leftover documents (12 migration/validation reports at the
tree root, 2 plugin ICD docs under Lts/Proxies/TrainingAreaBase, and
jenkins-ci-rme/README.md); README.md is the only .md in the tree. The
two README pointers at jenkins-ci-rme/README.md now describe the
pipeline files inline.
The pruned tree no longer registers the out-of-scope ctest entries; the
expected baseline is now 7 in-scope tests passed + 3 Not Run
(nglcomp_test, GrantleeParserSmoke, utils_test).
Verified with a full pipeline run on the pruned tree: 46/46 targets
built, ctest 7/7 in-scope passed, 5 DEBs produced.
MockCoordTrans is a Release 2 GROUP3 target: a header-only static
library defined in Lts/Common/CommonUnitTestsPlaceholder/mocks. As a
header-only lib it has no source file as a ninja edge input, so the
build-closure keep set did not pull its dir in and the prune removed
its CMakeLists (ninja: unknown target MockCoordTrans). Restore the dir
and its add_subdirectory; the smoke tests inside stay behind their
default-OFF options.
include/grantlee/context.h (the only grantlee header referenced by kept
code) includes its sibling stubs (abstractlocalizer.h, ...); the
file-level include/ prune left those siblings behind, breaking
codegen_grantlee compilation (fatal error: abstractlocalizer.h). The
grantlee engine stubs are a self-contained header set — keep the whole
dir, same treatment as include/GeographicLib.
FindArcGIS.cmake resolves the ArcGIS SDK via hardcoded hints rooted at
${CMAKE_SOURCE_DIR}/external/ArcGIS/MapsSDK_Qt_300.0.0/... ; in the
container that only resolves through this symlink to the pipeline-root
external/ mount. Its removal broke configure (Could NOT find ArcGIS,
missing ARCGIS_LIBRARY). No kept source references ArcGIS headers — the
include/ArcGIS tree stays removed; only the symlink is restored.
new_docker_build.sh and rpmbuild/make_rpm_ubuntu26.py (Release 1
validation-scope driver/producer) did not belong to this branch; their
functional callers are adapted to the Release 2 files:
- rpmbuild/make_rpm_ubuntu26_release2.py: PIPELINE_FILES now packages
new_docker_build_release2.sh and make_rpm_ubuntu26_release2.py.
- jenkins-ci-rme/jenkins_builder_ubuntu26.sh: runs
make_rpm_ubuntu26_release2.py with -DBUILD_RTI=ON and the Release 2
target list (--build_targets; a full-tree build is not possible on
this tree); staging tar also excludes ./build4*.
- jenkins-ci-rme/release-16-10-2026.sh (local dry run): calls
new_docker_build_release2.sh; derives the pipeline root from the tree
layout and exports MIGRATE_PIPELINE_ROOT; uses the prebuilt image when
present, otherwise the driver installs deps in-container (the R1
--build-prebuilt-image option does not exist in the R2 driver).
- jenkins-ci-rme/jenkins/Jenkinsfile: coverage stage runs the R2 driver
with MIGRATE_PIPELINE_ROOT exported (wins over the checked-in .envrc,
which carries a host-specific path).
- Doc/comment updates: Dockerfile header, install_arcgis.sh, publisher,
Jenkinsfile descriptions, jenkins-ci-rme/README.md, README.md.
Remaining references to the old names are only in historical migration
reports, which are left as-is.
Verified every claim against new_docker_build_release2.sh, the DEB
producer and a real pipeline run, then removed/fixed what does not
belong to this branch:
- Removed Step 1 (--build-prebuilt-image/--delete-prebuilt-image): those
options do not exist in the driver.
- Replaced the Release 1 bookkeeping (R1 packages table, R1 modules
table, 38-target count) with the actual Release 2 scope: 46 targets
(GROUP1 6, GROUP3 36, GROUP5 3, GROUP6 1).
- Fixed the build description: Ninja (not make -j8), unique
build4_<timestamp>_<pid> dirs (not a shared build4/), the real cmake
flags (BUILD_RTI, BUILD_GROUP*, RTI defines), no RELEASE_BUILD flag.
- Fixed image/mount facts: default image qt6-migrated-legatus-release2
(not guez-release2:2.0), /repo mount root (not /workspace), no
WORKSPACE env var, no build4/release_build.log.
- Fixed DEB names/contents: qt6-release-*_1.0-1_amd64.deb (not
guez-*_2.0), real PIPELINE_FILES list, honest note that the
ctest-runtime DEB is currently empty (build4 vs build4_<ts> mismatch).
- ctest baseline: 7 in-scope tests passed + the 9 expected Not Run
entries.
- Options reference: real options/defaults (lts-version 1.0), env
override list matching the driver; troubleshooting updated
(offscreen Qt, RTI license, prebuilt image).
A full Ninja build directory (613 files, ~1GB of build output) was
committed to the tree. Build artifacts do not belong in the index; the
build4_* gitignore rule added in 06209a496 keeps regenerated copies
out of status.
The driver builds into build4_$(date +%Y%m%d_%H%M%S)_$; the old
build4_2* pattern only covered current-year timestamps. Ignore the
whole family so build artifacts stop showing up as untracked.
A full-tree build (ninja all) is not possible on the Release 2 tree:
out-of-scope modules reference modules excluded from the scope
(e.g. TopicEditorBase -> LtsScriptTimeline; 100+ CMakeLists references),
and cmake --install hard-fails on targets that were never built. Align
the DEB pipeline with its own description (binaries and libraries from
release scope S):
- make_rpm_ubuntu26_release2.py: new --build_targets option
(space-separated CMake/Ninja targets; empty = legacy full tree).
Step 1 builds only those targets; the runtime DEB is staged from the
scoped build artifacts (bin/ + lib/) and packaged with build_deb
nobuild, replacing the full-tree debuild/cmake-buildsystem path.
- new_docker_build_release2.sh: pass the Release 2 group targets via
--build_targets; exclude the driver build4* dirs from the staging tar
(the DEB step builds its own tree; nothing consumes the driver build
output). Fix two comments inside the single-quoted docker bash -c
string that contained apostrophes, which broke word construction and
failed bash -n.
Run 3 exposed two driver bugs in the DEB step:
- A '#' comment line placed inside the backslash-continued python3
invocation terminated the logical command early: python ran without
--cmake_build_flags (hence the empty cmake flags and the recurring
pdm_enums.hpp failure) and bash then tried to execute
'--cmake_build_flags' as a command (exit 127). Moved the comment
above the statement.
- GUEZ_MODE was defaulted on the host but never passed to the container
(-e GUEZ_MODE missing), so the DEB cmake flags expanded -DGUEZ_MODE=
(empty -> false in CMake).
- The DEB cmake flags lacked BUILD_RTI: the full-tree DEB build compiles
RTI codegen consumers (svcPdmCommon -> ProjectDataManagement/pdm_enums.hpp)
whose BUILD_RTI=OFF stubs were removed as 'forbidden OFF variants', so the
deb_stage build failed with 'pdm_enums.hpp: No such file or directory'.
RTI_INSTALL_DIR still resolves in-container via DOCKER_MOUNT_ROOT.
- ctest: GUI test executables (coprtcnv/prtcnv unit tests, test_nglpdfwidget)
aborted in the headless container ('no Qt platform plugin could be
initialized' - xcb needs a display); export QT_QPA_PLATFORM=offscreen
(the offscreen plugin is present in the image).
- Host-side check now also accepts the license at the source tree root
(checked-in location) instead of only MIGRATE_PIPELINE_ROOT, where it
does not exist in this checkout
- In-container: export RTI_LICENSE pointing at the mounted license file so
RTI Connext DDS processes (build tools, ctest) can find it at runtime
The full-module import merge (097e5b488) replaced the validated Release 2
test wiring with the full-tree CMake files, leaving 5 targets referenced by
new_docker_build_release2.sh / README undefined:
- Restore LogViewLogger library in TetraIF/LogView/CMakeLists.txt
(pre-merge file defined both LogViewLogger lib and LogView exe)
- Re-add ADD_SUBDIRECTORY(unit-test) in TetraIF/CMakeLists.txt
(coprtcnv_unit_test, logview_unit_test, prtcnv_unit_test)
- Add CommonUnitTestsPlaceholder to Lts/Common/CMakeLists.txt (MockCoordTrans)
Driver (new_docker_build_release2.sh):
- Pass -DBUILD_TESTING=ON: TetraIF/unit-test is gated on BUILD_TESTING
- Default GUEZ_MODE=ON: the DEB step expanded ${GUEZ_MODE} in-container
where it was unset, producing -DGUEZ_MODE= (false in CMake)
rpmbuild/make_rpm_ubuntu26_release2.py:
- Use the Ninja generator (cmake -G Ninja / dh_auto_configure -GNinja):
with CMake 4.2.3 in the legatus-qt6:1.0 image the Unix Makefiles
generator emits no rule for AUTOMOC <target>_autogen/mocs_compilation.cpp
("No rule to make target") and every Qt target fails; Ninja handles
automoc correctly (verified: ninja -n of all 39 Release 2 targets passes)
- Add ninja-build to debian Build-Depends
Verified inside legatus-qt6:1.0 (mount /repo, BUILD_RTI=ON, BUILD_TESTING=ON,
Ninja): cmake configure exit 0 and ninja -n of the full Release 2 target
list (GROUP1/3/5/6, 39 targets) exit 0.
new_docker_build_release2.sh and the Release 2 README reference
rpmbuild/make_rpm_ubuntu26_release2.py, but only make_rpm_ubuntu26.py
existed (used by the jenkins-ci-rme builder). The Release 2 producer is
the same general 5-DEB builder (runtime, sources, ctest-runtime,
ctest-sources, build-pipeline), so add it under the expected name.
make_rpm_ubuntu26.py is kept for the jenkins-ci-rme flow.
- Update scope section with Release 2 target breakdown (38 targets)
- Update DEB package names from qt6-release-* to guez-*
- Update test suite expectations (13 tests including TetraIF)
- Update build instructions with guez-release2:2.0 Docker image
- Update Steps 4-6 for Release 2 DEB packaging and CI/CD
- Update options reference table with version 2.0
- Add troubleshooting note for Qt modules and build script
Co-authored-by: Copilot <2235566219+Copilot@users.noreply.github.com>
- Removed duplicate GROUP1_TARGETS definition (lines 306-316)
- Removed duplicate GROUP3_TARGETS definition with 100+ svc* services (lines 318-499)
- Set BUILD_GROUP1 default to OFF (not in Release 2 scope)
- Removed svcCoreClient from GROUP1_TARGETS
- Script reduced from 851 to 669 lines
- Only in-scope targets remain: 24 in GROUP3, 6 in GROUP1, 3 in GROUP5, 1 in GROUP6
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Replace all /workspace references with ${DOCKER_MOUNT_ROOT} in DEB packaging block
- Add WORKSPACE_DIR variable pointing to qt-migration-legatus subdirectory
- Use WORKSPACE_DIR as tar extraction source to correctly stage workspace files
- Fix output directory paths for RPM and packages-out locations
This fixes the staging error where tar couldn't find /workspace inside the container.
- Changed single quotes to double quotes in printf statement
- Prevents bash syntax error when inline script is passed to docker run -c
- Fixes DEB packaging stage execution failure
- Fix 'log_error: command not found' in DEB packaging stage
- Docker run uses inline bash script with separate function definitions
- Added log_error() after log_info() in Docker inline script (line 712)
- Function executes inside container context where DEB packaging runs
- log_error() function was used in DEB packaging block but not defined
- Added function definition after log_info() at script top
- Fixes 'log_error: command not found' error in DEB packaging stage
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Enable LtsMultiMonitor and PluginInterfaces in Lts/Common/CMakeLists.txt
- Add LtsMultiMonitor, DataPluginInterface, GuiPluginInterface to GROUP3_TARGETS
- Required by LtsSession which is used by LtsWidgets
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Removed pdmTests, eemTests, flmTests, srpFtTests from GROUP3_TARGETS
as they are test executables, not library targets.
Release 2a now contains only 18 library targets:
- 14 NGL/Qt single application libraries
- 2 wms_json_ds, tcpserver
- 2 LtsCoordinate, LtsCoordTrans
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Exclude LtsElevation library (depends on AssetManagement, needed only for Release 2b services)
- Update GROUP3_TARGETS to remove LtsElevation
- Comment out LtsElevation subdirectory in CMakeLists.txt
Co-authored-by: Copilot <2235566219+Copilot@users.noreply.github.com>
Add Jenkins builder and coverage options:
- --run-jenkins-builder: Enable DEB package production mode
- --package-output-dir: Output directory for .deb packages
- --lts-version: LTS version for package metadata
- --project: Project name (guez)
- --build-number: Build number for package versioning
- --build-tag: Build tag/identifier
- --run-coverage: Enable lcov code coverage
- --coverage-output-dir: Output directory for coverage reports
Add MIGRATE_PIPELINE_ROOT handling with precedence:
1. Exported variable (Jenkinsfile)
2. .envrc file
3. Script location default
Add validation for mutually exclusive options:
- --run-jenkins-builder and --run-coverage cannot be used together
Export all new variables for Docker container and pass to CMake.
Co-authored-by: Copilot <223556619+Copilot@users.noreply.github.com>
- LtsElevation required by Trajectory generator used in svcAesCommon
- Part of core libraries needed for AreaEffectSimulation
- Fixes build error: missing ltselevation_export.h
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Use TARGET_EXISTS generator expressions for external dependencies
- Allows building without Viewer/SituationMap module in Release 2
- Maintains compatibility when those modules are available
- Required by svcAesClient/svcAesCommon for coordinate transformation
- Fixes missing include directories for coordinate transformation library
- Both libraries are core dependencies for AreaEffectSimulation services
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- QT5_ADD_RESOURCES was a no-op stub under Qt6, silently dropping .qrc
files: LogView's :/image/* icons were never compiled into the binary.
Map the stub onto qt6_add_resources() (outfiles-variable form, which
appends the generated qrc_*.cpp to the source list). LogView is the
only caller of the stub.
- nglwidgets' Qt6 branch only processed rme_light_style.qrc;
nglresources.qrc (which provides :icons/16x16/* toolbar/window icons)
was left as a raw file in the source list and never compiled
(CMAKE_AUTORCC is off). Process both qrcs.
- LogView never loads libnglwidgets.so at runtime (--as-needed drops it:
the code uses no nglwidgets symbols), so its :icons/16x16 resources
were never registered for the app. Embed the 20 icons LogView actually
uses directly into LogViewResources.qrc with explicit aliases (rcc
rejects XML comments in .qrc and, without an alias, registers the full
relative path as the resource name).
berrux-dev
torstenandCopilot