Merge pull request #344 from ds4dm/gasse-patch-3

Author: AntoinePrv

CMakeLists.txt

@@ -1,5 +1,4 @@
-cmake_minimum_required(VERSION 3.14)
-cmake_policy(SET CMP0095 OLD)  # RPATH escaping
+cmake_minimum_required(VERSION 3.16)
 
 # Adapt compiler flags if using Conda compiler packages. Before project so they are not modified.
 include(cmake/Conda.cmake)

README.rst

@@ -27,7 +27,7 @@ in cooperation with a state-of-the-art Mixed Integer Linear Programming solver
 that acts as a controllable algorithm.
 
 The underlying solver used is `SCIP <https://scip.zib.de/>`_, and the user facing API is
-meant to mimic the `OpenAI Gym <https://gym.openai.com/>`_ API (as much as possible).
+meant to mimic the `OpenAI Gym <https://www.gymlibrary.dev/>`_ API (as much as possible).
 
 .. code-block:: python
 

cmake/DependenciesResolver.cmake

@@ -5,6 +5,10 @@
 # Based and fetch content, it avoids using `add_subdirectory` which exposes other project
 # targets and errors as part of this project.
 
+# Avoid warning about DOWNLOAD_EXTRACT_TIMESTAMP in CMake 3.24:
+if (CMAKE_VERSION VERSION_GREATER_EQUAL "3.24.0")
+	cmake_policy(SET CMP0135 NEW)
+endif()
 
 include(FetchContent)
 

docs/discussion/gym-differences.rst

@@ -3,7 +3,7 @@ Differences with OpenAI Gym
 
 Changing reward and observations
 --------------------------------
-Contrarily to `OpenAI Gym <https://gym.openai.com/>`_ where learning tasks are predefined,
+Contrarily to `OpenAI Gym <https://www.gymlibrary.dev/>`_ where learning tasks are predefined,
 Ecole gives the user the tools to easily extend and customize environments.
 This is because the objective with Ecole is not only to provide a collection of challenges
 for machine learning, but really to solve combinatorial optimization problems more

docs/index.rst

@@ -38,7 +38,7 @@ picking the next variable to branch on. Ecole aims at exposing these algorithmic
 standard reinforcement learning API (agent / environment loop), in order to ease the exploration
 of new machine learning models and algorithms for learning data-driven policies.
 
-Ecole's interface is inspired from `OpenAI Gym <https://gym.openai.com/>`_ and will look
+Ecole's interface is inspired from `OpenAI Gym <https://www.gymlibrary.dev/>`_ and will look
 familiar to reinforcement learning praticionners.
 The state-of-the-art Mixed Integer Linear Programming solver that acts as a controllable
 algorithm inside Ecole is `SCIP <https://scip.zib.de/>`_.

libecole/include/ecole/environment/environment.hpp

@@ -28,7 +28,7 @@ namespace ecole::environment {
  * Environments are the main abstraction exposed by Ecole.
  * They characterise the Markov Decision Process task to solve.
  * The interface to environments is meant to be close to that of
- * [OpenAi Gym](https://gym.openai.com/), with some differences nontheless due to the
+ * [OpenAi Gym](https://www.gymlibrary.dev/), with some differences nontheless due to the
  * requirements of Ecole.
  *
  * @tparam Dynamics The ecole::environment::EnvironmentDynamics driving the initial state and transition of the

setup.py

@@ -55,7 +55,7 @@ def get_cmake_in_package_install_args() -> List[str]:
     """Return default installation settings for installing libecole in the package."""
     system = platform.system()
     if system == "Linux":
-        origin = r"\${ORIGIN}"
+        origin = r"${ORIGIN}"
     elif system == "Darwin":
         origin = "@loader_path"
     else: