Introduce JRL CMake Modules v2 by ahoarau · Pull Request #798 · jrl-umi3218/jrl-cmakemodules

ahoarau · 2025-12-10T16:05:56Z

In this PR we introduce the JRL CMake Modules v2, extracted from the repo: https://github.com/ahoarau/jrl-cmakemodules-v2

The integration on the user-side can be done as follows https://github.com/ahoarau/nanoeigenpy/blob/db692b1dbe3fff76688550048fd3004ecbd1a2d7/cmake/get-jrl-cmakemodules.cmake

find_package(jrl-cmakemodules 1.2.0 CONFIG REQUIRED)

v1.2.0 will integrate the v2 folder, containing the new API.
Either set an env variable JRL_CMAKEMODULES_SOURCE_DIR pointing to the jrl repo source (example $env:JRL_CMAKEMODULES_SOURCE_DIR="C:/a/jrl-cmakemodules"
We do not intend to use both v1 and v2 in projects.

TODO:

Port the main modules
Test the integration in client repo (proxsuite, eigenpy, nanoeigenpy, coal)
Figure out why gersemi still needs to re-format the new modules
Add COMPATIBLITY_OPTION for jrl_cmake_dependent_option()
Port tests from original repo
Port CI from original repo
Rewrite release.cmake in python

jorisv · 2025-12-30T11:15:18Z

+#  * Otherwise, if running inside a Conda environment on Windows, an
+#    absolute path to `sysconfig.get_path('purelib')` is returned.


As we already talked, I'm not a huge fan of handling this special case. The ${PROJECT_NAME}_PYTHON_INSTALL_DIR variable is made for this purpose.

I tried both options to see which one feels more natural. In the end, this function is here to handle the different cases of a python install dir, so it makes sense to take the conda+windows special/unique/annoying case here. All the other cases are fine with the default behaviour, so I think it's wise to keep it.

The ${PROJECT_NAME}_PYTHON_INSTALL_DIR is still useful as it can be used to override this behavior.

I still have some concern.
This mean on Windows we will have an absolute path, ignoring the CMAKE_INSTALL_PREFIX.

Can we at least make the install path relative to CMAKE_INSTALL_PREFIX ? This will avoid user be surprised to install the binding in their conda environment when they specified a CMAKE_INSTALL_PREFIX != CONDA_PREFIX

arntanguy

Just gave it a try with SpaceVecAlg: jrl-umi3218/SpaceVecAlg#67
There seems to be an issue with BUILDSYSTEM_TARGETS, see my comment inline.

Besides that it worked just fine. I'll keep exploring during the week.

nim65s

That's a lot of work 😅

I did not look into external-modules yet, but here are some comments as a starting point.

jorisv · 2026-01-22T09:27:34Z

+#]============================================================================]
+macro(jrl_find_nanobind)
+    string(REPLACE ";" " " args_pp "${ARGN}")
+    _jrl_check_var_defined(Python_EXECUTABLE "Python executable not found (variable Python_EXECUTABLE).


Use Python::Interpreter instead

jorisv · 2026-01-22T09:50:16Z

+#  * Otherwise, if running inside a Conda environment on Windows, an
+#    absolute path to `sysconfig.get_path('purelib')` is returned.


I still have some concern.
This mean on Windows we will have an absolute path, ignoring the CMAKE_INSTALL_PREFIX.

Can we at least make the install path relative to CMAKE_INSTALL_PREFIX ? This will avoid user be surprised to install the binding in their conda environment when they specified a CMAKE_INSTALL_PREFIX != CONDA_PREFIX

jorisv · 2026-01-22T09:56:09Z

+   absolute path to `sysconfig.get_path('purelib')` is returned.
+ * In all other cases, the relative path to site-packages is returned.
+
+### Windows Layout


Suggested change

### Windows Layout

### Conda Windows Layout

…xtractor

which has been removed because unused

ahoarau marked this pull request as draft December 11, 2025 16:41

jorisv requested changes Dec 12, 2025

View reviewed changes

nim65s self-requested a review December 15, 2025 10:24

ahoarau force-pushed the jrl-next branch from 81d0ea4 to 8148456 Compare December 19, 2025 15:21

nim65s mentioned this pull request Dec 20, 2025

Revert "Refactor CMake with JRL CMake Modules v2" Simple-Robotics/nanoeigenpy#36

Merged

jorisv requested changes Dec 30, 2025

View reviewed changes

ahoarau mentioned this pull request Jan 5, 2026

Refactor CMake with JRL CMake Modules v2 (v2) Simple-Robotics/nanoeigenpy#40

Open

ahoarau force-pushed the jrl-next branch from aba7a86 to bc63bf3 Compare January 6, 2026 12:37

This was referenced Jan 6, 2026

Adapt to jrl-cmakemodules v2 and nix flake + nanobind bindings arntanguy/SpaceVecAlg#1

Closed

Adapt to jrl-cmakemodules v2 and nix flake + nanobind bindings jrl-umi3218/SpaceVecAlg#67

Draft

arntanguy reviewed Jan 6, 2026

View reviewed changes

Comment thread v2/modules/jrl.cmake Outdated

nim65s reviewed Jan 9, 2026

View reviewed changes

ahoarau force-pushed the jrl-next branch 3 times, most recently from 9b26ecf to 62e72ef Compare January 13, 2026 08:34

jorisv requested changes Jan 22, 2026

View reviewed changes

ahoarau force-pushed the jrl-next branch from 75c5b8b to af905fc Compare February 13, 2026 14:33

nim65s reviewed Mar 5, 2026

View reviewed changes

Comment thread v2/scripts/release.py Outdated

Comment thread v2/scripts/release.py Outdated

Comment thread v2/modules/jrl.cmake Outdated

ahoarau force-pushed the jrl-next branch from d4840a2 to 4304a5d Compare March 10, 2026 08:35

nim65s reviewed Mar 11, 2026

View reviewed changes

Comment thread CMakeLists.txt

ahoarau force-pushed the jrl-next branch from 588786d to 426cf3b Compare March 16, 2026 10:27

arntanguy reviewed Mar 16, 2026

View reviewed changes

Comment thread v2/modules/jrl.cmake Outdated

arntanguy reviewed Mar 16, 2026

View reviewed changes

Comment thread v2/modules/jrl.cmake Outdated

This was referenced Mar 16, 2026

Switch to jrl-cmakemodules v2 Hugo-L3174/DCM_VRPTask#1

Draft

Use standalone MuJoCo 3.3.6 and jrl-cmakemodules v2 rohanpsingh/mc_mujoco#98

Draft

ahoarau force-pushed the jrl-next branch from 7d0ad37 to c9f0204 Compare March 20, 2026 10:10

ahoarau added 29 commits April 19, 2026 16:15

pixi: add doc task to generate api doc

0ab2808

doc: update documentation

2c3da77

readme: update main readme for v2

2546c8d

remove unused generated readme

e3f4826

pixi: integrate python tests

36eb693

pixi: update lock file

2a51154

release.py: remove gitpython dep and cleanup comments

03823f7

release.py: update/shorten doc

dbde4a3

release.py: remove enumerate

b44c49c

release.py: change ChangelogVersionExtractor to inherit from VersionE…

aa15855

…xtractor

release.py: Add option to specify files to stage

9339527

release.py: display git tag and commit in dry run

be21f8d

refactor: simplify update_pixi_lock function params

95a1b7f

test: update assertion for previous_version in cli json output test

dd9fafd

tests: add release.py coverage

e4cce36

release.py: update dry run output message

b58b0fc

release.py: display the output of pixi list command

a996a85

release.py: uniformize the display output with the dry-run

d203d23

rename release.py to jrl_release.py

c5d89ad

jrl_release.py: update sheban according to ub doc and chmod +x

c002cf4

jrl_release.py: remove unused RegexVersionExtractor

1bc0198

tests: remove test on RegexVersionExtractor

cc6bacb

which has been removed because unused

jrl_target_set_default_compile_options: update doc

31454ef

jrl_target_enforce_msvc_conformance: update docs

6773ad6

ci: add boost-libs (required on boost 1.90)

1fb2d48

cmake: reformat with new gersemi

a6e9213

jrl: update error messages

e1d0307

pre-commit: update gersemi new repo

7a8afcb

gitignore: merge v2 with main

e37d49f

ahoarau force-pushed the jrl-next branch from 6d9f0f0 to e37d49f Compare April 19, 2026 14:15

		# * Otherwise, if running inside a Conda environment on Windows, an
		# absolute path to `sysconfig.get_path('purelib')` is returned.

Conversation

ahoarau commented Dec 10, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

jorisv Dec 30, 2025

Choose a reason for hiding this comment

Uh oh!

ahoarau Jan 5, 2026

Choose a reason for hiding this comment

Uh oh!

jorisv Jan 22, 2026

Choose a reason for hiding this comment

Uh oh!

arntanguy left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

nim65s left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

jorisv Jan 22, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

jorisv Jan 22, 2026

Choose a reason for hiding this comment

Uh oh!

jorisv Jan 22, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

ahoarau commented Dec 10, 2025 •

edited

Loading