Improve logging behaviour

.github/workflows/deploy_to_pypi.yml

@@ -20,6 +20,7 @@ jobs:
         env:
           TWINE_USERNAME: "__token__"
           TWINE_PASSWORD: ${{ secrets.PYPI_TOKEN }}
+          TORCHQUAD_RELEASE_BUILD: "True"
         run: |
           python setup.py sdist bdist_wheel
           twine upload dist/*

.github/workflows/deploy_to_test_pypi.yml

@@ -23,6 +23,7 @@ jobs:
         env:
           TWINE_USERNAME: "__token__"
           TWINE_PASSWORD: ${{ secrets.TEST_PYPI_TOKEN }}
+          TORCHQUAD_RELEASE_BUILD: "True"
         run: |
           python setup.py sdist bdist_wheel
           twine upload -r testpypi dist/*

README.md

@@ -193,8 +193,24 @@ integral_value = mc.integrate(
     backend="torch",
 )
 ```
-To change the logger verbosity, set the `TORCHQUAD_LOG_LEVEL` environment
-variable; for example `export TORCHQUAD_LOG_LEVEL=DEBUG`.
+## Logging Configuration
+
+By default, torchquad disables its internal logging when installed from PyPI to avoid interfering with other loggers in your application. To enable logging:
+
+1. **Set the log level**: Use the `TORCHQUAD_LOG_LEVEL` environment variable:
+   ```bash
+   export TORCHQUAD_LOG_LEVEL=DEBUG   # For detailed debugging
+   export TORCHQUAD_LOG_LEVEL=INFO    # For general information  
+   export TORCHQUAD_LOG_LEVEL=WARNING # For warnings only (default when enabled)
+   ```
+
+2. **Enable logging programmatically**:
+   ```python
+   import torchquad
+   torchquad.set_log_level("DEBUG")  # This will enable and configure logging
+   ```
+
+Note: When developing from a git clone, logging is enabled by default. The `TORCHQUAD_RELEASE_BUILD` environment variable controls this behavior.
 
 You can find all available integrators [here](https://torchquad.readthedocs.io/en/main/integration_methods.html).
 

torchquad/__init__.py

@@ -1,5 +1,4 @@
 import os
-from loguru import logger
 
 # TODO: Currently this is the way to expose to the docs
 # hopefully changes with setup.py
@@ -42,5 +41,11 @@
     "_deployment_test",
 ]
 
-set_log_level(os.environ.get("TORCHQUAD_LOG_LEVEL", "WARNING"))
-logger.info("Initializing torchquad.")
+# Check for release build flag to avoid interfering with other loggers
+TORCHQUAD_RELEASE_BUILD = os.environ.get("TORCHQUAD_RELEASE_BUILD", "False").lower() == "true"
+
+if not TORCHQUAD_RELEASE_BUILD:
+    from loguru import logger
+
+    set_log_level(os.environ.get("TORCHQUAD_LOG_LEVEL", "WARNING"))
+    logger.info("Initializing torchquad.")

torchquad/integration/utils.py

@@ -123,7 +123,9 @@ def _setup_integration_domain(dim, integration_domain, backend):
     domain_arg_backend = infer_backend(integration_domain)
     convert_to_tensor = domain_arg_backend == "builtins"
     if not convert_to_tensor and backend is not None and domain_arg_backend != backend:
-        logger.warning("integration_domain should be a list when the backend argument is set.")
+        warning_msg = "integration_domain should be a list when the backend argument is set."
+        logger.warning(warning_msg)
+        warnings.warn(warning_msg, RuntimeWarning)
         convert_to_tensor = True
 
     # Convert integration_domain to a tensor if needed

torchquad/integration/vegas_map.py

@@ -1,6 +1,7 @@
 from autoray import numpy as anp
 from autoray import astype, to_backend_dtype
 from loguru import logger
+import warnings
 
 from .utils import _add_at_indices
 
@@ -185,10 +186,12 @@ def update_map(self):
         """Update the adaptive map, Section II C."""
         smoothed_weights = self._smooth_map(self.weights, self.counts, self.alpha)
         if smoothed_weights is None:
-            logger.warning(
+            warning_msg = (
                 "Cannot update the VEGASMap. This can happen with an integrand "
                 "which evaluates to zero everywhere."
             )
+            logger.warning(warning_msg)
+            warnings.warn(warning_msg, RuntimeWarning)
             self._reset_weight()
             return
 
@@ -242,9 +245,9 @@ def update_map(self):
                 # smoothed_weights[i][indices] can be zero, which leads to
                 # invalid edges.
                 num_edges = self.x_edges.shape[1]
-                logger.warning(
-                    f"{num_edges - anp.sum(finite_edges)} out of {num_edges} calculated VEGASMap edges were infinite"
-                )
+                warning_msg = f"{num_edges - anp.sum(finite_edges)} out of {num_edges} calculated VEGASMap edges were infinite"
+                logger.warning(warning_msg)
+                warnings.warn(warning_msg, RuntimeWarning)
                 # Replace inf edges with the average of their two neighbours
                 middle_edges = 0.5 * (self.x_edges[i][:-2] + self.x_edges[i][2:])
                 self.x_edges[i][1:-1] = anp.where(

torchquad/utils/enable_cuda.py

@@ -1,4 +1,5 @@
 from loguru import logger
+import warnings
 
 from .set_precision import set_precision
 
@@ -20,4 +21,6 @@ def enable_cuda(data_type="float32"):
         if data_type is not None:
             set_precision(data_type)
     else:
-        logger.warning("Error enabling CUDA. cuda.is_available() returned False. CPU will be used.")
+        warning_msg = "Error enabling CUDA. cuda.is_available() returned False. CPU will be used."
+        logger.warning(warning_msg)
+        warnings.warn(warning_msg, RuntimeWarning)

torchquad/utils/set_precision.py

@@ -1,5 +1,6 @@
 from loguru import logger
 import os
+import sys
 
 
 def _get_precision(backend):
@@ -31,9 +32,9 @@ def set_precision(data_type="float32", backend="torch"):
     # upper-case letters
     data_type = {"float": "float32", "double": "float64"}.get(data_type.lower(), data_type)
     if data_type not in ["float32", "float64"]:
-        logger.error(
-            f'Invalid data type "{data_type}". Only float32 and float64 are supported. Setting the data type to float32.'
-        )
+        error_msg = f'Invalid data type "{data_type}". Only float32 and float64 are supported. Setting the data type to float32.'
+        logger.error(error_msg)
+        print(f"ERROR: {error_msg}", file=sys.stderr)
         data_type = "float32"
 
     if backend == "torch":
@@ -67,4 +68,6 @@ def set_precision(data_type="float32", backend="torch"):
         os.environ["TORCHQUAD_DTYPE_NUMPY"] = data_type
         logger.info(f"NumPy default dtype set to {_get_precision('numpy')}")
     else:
-        logger.error(f"Changing the data type is not supported for backend {backend}")
+        error_msg = f"Changing the data type is not supported for backend {backend}"
+        logger.error(error_msg)
+        print(f"ERROR: {error_msg}", file=sys.stderr)