Cascading series of CI dependency problems (#746)

A new git version outputs a different date format that older versions of
python's datetime module don't recognize.
pypa/setuptools-scm#1038

setuptools-scm cut a new version to address that, but it's not available
in conda main channel yet, which is required, since without that the
conda pip phase can't execute in a single transaction

For now, we install things via conda-forge and adjust the channel
priority order so that the full set of dependencies could be resolved.

That of course brought in additional changes (e.g., `python=3.12` by
default, new `pylint`, `pycodestyle`, etc.), so this now also includes
some additional linting changes.

However, longer term, we need to switch to a pyproject.toml file to fix
this properly.
There we can specify prereqs for even loading the setup.py as well as
fix some other config complexities, though that is a broader change.

---------

Co-authored-by: Sergiy Matusevych <sergiym@microsoft.com>

.devcontainer/Dockerfile

@@ -47,16 +47,18 @@ RUN umask 0002 \
     && /opt/conda/bin/conda install -v -y -n base conda-libmamba-solver \
     && /opt/conda/bin/conda config --set solver libmamba \
     && /opt/conda/bin/conda list -n base \
-    && /opt/conda/bin/conda clean -v -y -a
-
-# Update the base. This helps save space by making sure the same version
-# python is used for both the base env and mlos env.
-RUN umask 0002 \
-    && /opt/conda/bin/conda update -v -y -n base -c defaults --all \
-    && /opt/conda/bin/conda update -v -y -n base -c defaults conda python \
     && /opt/conda/bin/conda clean -v -y -a \
     && /opt/conda/bin/conda run -n base pip cache purge
 
+# No longer relevant since we're using conda-forge in the environment files by default now.
+## Update the base. This helps save space by making sure the same version
+## python is used for both the base env and mlos env.
+#RUN umask 0002 \
+#    && /opt/conda/bin/conda update -v -y -n base -c defaults --all \
+#    && /opt/conda/bin/conda update -v -y -n base -c defaults conda python \
+#    && /opt/conda/bin/conda clean -v -y -a \
+#    && /opt/conda/bin/conda run -n base pip cache purge
+
 # Install some additional editor packages for the base environment.
 RUN umask 0002 \
     && /opt/conda/bin/conda run -n base pip install --no-cache-dir -U pynvim

conda-envs/mlos-3.10.yml

@@ -1,5 +1,6 @@
 name: mlos-3.10
 channels:
+  - conda-forge
   - defaults
 dependencies:
   # Basic dev environment packages.
@@ -10,7 +11,7 @@ dependencies:
   - pydocstyle
   - flake8
   - setuptools
-  - setuptools-scm
+  - setuptools-scm>=8.1.0
   - jupyter
   - ipykernel
   - nb_conda_kernels

conda-envs/mlos-3.11.yml

@@ -1,5 +1,6 @@
 name: mlos-3.11
 channels:
+  - conda-forge
   - defaults
 dependencies:
   # Basic dev environment packages.
@@ -10,7 +11,7 @@ dependencies:
   - pydocstyle
   - flake8
   - setuptools
-  - setuptools-scm
+  - setuptools-scm>=8.1.0
   - jupyter
   - ipykernel
   - nb_conda_kernels

conda-envs/mlos-3.8.yml

@@ -1,5 +1,6 @@
 name: mlos-3.8
 channels:
+  - conda-forge
   - defaults
 dependencies:
   # Basic dev environment packages.
@@ -10,7 +11,7 @@ dependencies:
   - pydocstyle
   - flake8
   - setuptools
-  - setuptools-scm
+  - setuptools-scm>=8.1.0
   - jupyter
   - ipykernel
   - nb_conda_kernels

conda-envs/mlos-3.9.yml

@@ -1,5 +1,6 @@
 name: mlos-3.9
 channels:
+  - conda-forge
   - defaults
 dependencies:
   # Basic dev environment packages.
@@ -10,7 +11,7 @@ dependencies:
   - pydocstyle
   - flake8
   - setuptools
-  - setuptools-scm
+  - setuptools-scm>=8.1.0
   - jupyter
   - ipykernel
   - nb_conda_kernels

conda-envs/mlos-windows.yml

@@ -1,7 +1,6 @@
 name: mlos
 channels:
   # Note: we have to reverse the channel priority for Windows to accomodate strict channel_priority setups.
-  # Hence, additional dependencies may differ from the Linux setup.
   - conda-forge
   - defaults
 dependencies:
@@ -13,7 +12,7 @@ dependencies:
   - pydocstyle
   - flake8
   - setuptools
-  - setuptools-scm
+  - setuptools-scm>=8.1.0
   - jupyter
   - ipykernel
   - nb_conda_kernels
@@ -23,10 +22,11 @@ dependencies:
   - pyarrow
   - swig
   - libpq
-  - python=3.11
+  # FIXME: https://github.com/microsoft/MLOS/issues/727
+  - python<3.12
   # Install an SMAC requirement pre-compiled from conda-forge.
   # This also requires a more recent vs2015_runtime from conda-forge.
-  - conda-forge::pyrfr>=0.9.0
+  - pyrfr>=0.9.0
   - pip:
     - autopep8>=1.7.0
     - bump2version

conda-envs/mlos.yml

@@ -1,5 +1,6 @@
 name: mlos
 channels:
+  - conda-forge
   - defaults
 dependencies:
   # Basic dev environment packages.
@@ -10,7 +11,7 @@ dependencies:
   - pydocstyle
   - flake8
   - setuptools
-  - setuptools-scm
+  - setuptools-scm>=8.1.0
   - jupyter
   - ipykernel
   - nb_conda_kernels
@@ -20,7 +21,8 @@ dependencies:
   - pyarrow
   - swig
   - libpq
-  - python
+  # FIXME: https://github.com/microsoft/MLOS/issues/727
+  - python<3.12
   - pip:
     - autopep8>=1.7.0
     - bump2version

mlos_core/mlos_core/optimizers/bayesian_optimizers/smac_optimizer.py

@@ -27,7 +27,7 @@ class SmacOptimizer(BaseBayesianOptimizer):
     Wrapper class for SMAC based Bayesian optimization.
     """
 
-    def __init__(self, *,  # pylint: disable=too-many-locals
+    def __init__(self, *,  # pylint: disable=too-many-locals,too-many-arguments
                  parameter_space: ConfigSpace.ConfigurationSpace,
                  optimization_targets: List[str],
                  space_adapter: Optional[BaseSpaceAdapter] = None,
@@ -283,7 +283,7 @@ def _suggest(self, context: Optional[pd.DataFrame] = None) -> pd.DataFrame:
             Pandas dataframe with a single row. Column names are the parameter names.
         """
         if TYPE_CHECKING:
-            from smac.runhistory import TrialInfo  # pylint: disable=import-outside-toplevel
+            from smac.runhistory import TrialInfo  # pylint: disable=import-outside-toplevel,unused-import
 
         if context is not None:
             warn(f"Not Implemented: Ignoring context {list(context.columns)}", UserWarning)

mlos_core/mlos_core/optimizers/flaml_optimizer.py

@@ -30,7 +30,7 @@ class FlamlOptimizer(BaseOptimizer):
     Wrapper class for FLAML Optimizer: A fast library for AutoML and tuning.
     """
 
-    def __init__(self, *,
+    def __init__(self, *,   # pylint: disable=too-many-arguments
                  parameter_space: ConfigSpace.ConfigurationSpace,
                  optimization_targets: List[str],
                  space_adapter: Optional[BaseSpaceAdapter] = None,

mlos_core/mlos_core/spaces/adapters/llamatune.py

@@ -9,6 +9,7 @@
 from warnings import warn
 
 import ConfigSpace
+from ConfigSpace.hyperparameters import NumericalHyperparameter
 import numpy as np
 import numpy.typing as npt
 import pandas as pd
@@ -222,7 +223,7 @@ def _transform(self, configuration: dict) -> dict:
                 index = max(0, min(len(param.choices) - 1, index))
                 # NOTE: potential rounding here would be unfair to first & last values
                 orig_value = param.choices[index]
-            elif isinstance(param, ConfigSpace.hyperparameters.NumericalHyperparameter):
+            elif isinstance(param, NumericalHyperparameter):
                 if param.name in self._special_param_values_dict:
                     value = self._special_param_value_scaler(param, value)
 

mlos_core/mlos_core/spaces/converters/flaml.py

@@ -6,7 +6,7 @@
 Contains space converters for FLAML.
 """
 
-from typing import Dict
+from typing import Dict, TYPE_CHECKING
 
 import sys
 
@@ -16,6 +16,9 @@
 import flaml.tune
 import flaml.tune.sample
 
+if TYPE_CHECKING:
+    from ConfigSpace.hyperparameters import Hyperparameter
+
 if sys.version_info >= (3, 10):
     from typing import TypeAlias
 else:
@@ -46,7 +49,7 @@ def configspace_to_flaml_space(config_space: ConfigSpace.ConfigurationSpace) ->
         (ConfigSpace.UniformFloatHyperparameter, True): flaml.tune.loguniform,
     }
 
-    def _one_parameter_convert(parameter: ConfigSpace.hyperparameters.Hyperparameter) -> FlamlDomain:
+    def _one_parameter_convert(parameter: "Hyperparameter") -> FlamlDomain:
         if isinstance(parameter, ConfigSpace.UniformFloatHyperparameter):
             # FIXME: upper isn't included in the range
             return flaml_numeric_type[(type(parameter), parameter.log)](parameter.lower, parameter.upper)

mlos_core/mlos_core/tests/spaces/spaces_test.py

@@ -18,15 +18,15 @@
 import scipy
 
 import ConfigSpace as CS
-from ConfigSpace.hyperparameters import NormalIntegerHyperparameter
+from ConfigSpace.hyperparameters import Hyperparameter, NormalIntegerHyperparameter
 
 import flaml.tune.sample
 
 from mlos_core.spaces.converters.flaml import configspace_to_flaml_space, FlamlDomain, FlamlSpace
 
 
 OptimizerSpace = Union[FlamlSpace, CS.ConfigurationSpace]
-OptimizerParam = Union[FlamlDomain, CS.hyperparameters.Hyperparameter]
+OptimizerParam = Union[FlamlDomain, Hyperparameter]
 
 
 def assert_is_uniform(arr: npt.NDArray) -> None:

mlos_viz/mlos_viz/base.py

@@ -180,7 +180,7 @@ def compute_zscore_for_group_agg(
         compute_zscore_for_group_agg(results_groups_perf, stats_df, result_col, "var")
         quantiles = [0.50, 0.75, 0.90, 0.95, 0.99]
         for quantile in quantiles:     # TODO: can we do this in one pass?
-            quantile_col = result_col + f".p{int(quantile*100)}"
+            quantile_col = f"{result_col}.p{int(quantile * 100)}"
             stats_df[quantile_col] = results_groups_perf.transform("quantile", quantile)
         augmented_results_df = pandas.concat([augmented_results_df, stats_df], axis=1)
     return augmented_results_df