Test and fix Trial

experitur/core/experiment.py

@@ -506,9 +506,7 @@ def _trial_generator(self):
                 ).as_dict(),
             )
 
-            trial = self.ctx.trials.create(
-                trial_configuration, record_used_parameters=True
-            )
+            trial = self.ctx.trials.create(trial_configuration)
             os.makedirs(trial.wdir, exist_ok=True)
 
             yield trial
@@ -653,7 +651,7 @@ def run_trial(self, trial: Trial):
         self._handle_event("on_pre_run", trial)
 
         try:
-            with self.ctx.set_current_trial(trial):
+            with self.ctx.set_current_trial(trial), trial.record_used_parameters():
                 result = self.func(trial, *args, **kwargs)
 
             # Merge returned result into existing result

experitur/core/trial.py

@@ -1,4 +1,5 @@
 import collections.abc
+import contextlib
 import copy
 import datetime
 import glob
@@ -85,7 +86,11 @@ def __contains__(self, x: object) -> bool:
         return x in self.data()
 
     def __iter__(self):
-        return _filter_prefixed(self.trial._prefix, self.data())
+        return iter(set(_filter_prefixed(self.trial._prefix, self.data())))
+
+    def __repr__(self):
+        items = ", ".join(repr(o) for o in self)
+        return f"<_PrefixedTrialDataSetView {{{items}}}>"
 
     def __len__(self):
         return sum(1 for _ in self)
@@ -219,9 +224,12 @@ def __getitem__(self, name):
 
         return self.resolved_parameters[name]
 
+    def __eq__(self, o: object) -> bool:
+        return self is o
+
     @property
-    def unused_parameters(self):
-        return sorted(set(self.resolved_parameters) - set(self.used_parameters))
+    def unused_parameters(self) -> Set:
+        return set(self.resolved_parameters) - set(self.used_parameters)
 
     def __setitem__(self, name, value):
         """Set the value of a parameter."""
@@ -234,6 +242,17 @@ def __delitem__(self, name):
     def __iter__(self):
         return iter(self.resolved_parameters)
 
+    def items(self):
+        with self.record_used_parameters(False):
+            return super().items()
+
+    @contextlib.contextmanager
+    def record_used_parameters(self, record=True):
+        old_value = self._record_used_parameters
+        self._record_used_parameters = record
+        yield
+        self._record_used_parameters = old_value
+
     def todict(self, with_prefix=False):
         """
         Convert trial data to dictionary.
@@ -324,7 +343,8 @@ def get(self, key, default=None, setdefault=True):
         return super().get(key, default)
 
     def save(self):
-        # Compact used parameters
+        # Store and compact used and unused parameters
+        self._data["used_parameters"] = sorted(self.used_parameters)
         self._data["unused_parameters"] = sorted(self.unused_parameters)
 
         # Write to the store

tests/core/test_trial.py

@@ -52,14 +52,18 @@ def test_trial_parameters(tmp_path):
 
         @Experiment(configurator={"a": [1], "b": [2], "c": ["{a}"]})
         def experiment(trial: Trial):
-            assert ctx.current_trial == trial
+            assert ctx.current_trial is trial
+
+            assert trial.used_parameters == set()
+            assert trial.unused_parameters == {"a", "b", "c"}
 
             assert trial["a"] == 1
             assert trial["b"] == 2
             assert trial["c"] == trial["a"]
             assert len(trial) == 3
 
-            print(trial["a"], trial["c"])
+            assert trial.used_parameters == {"a", "b", "c"}
+            assert trial.unused_parameters == set()
 
             for k, v in trial.items():
                 pass
@@ -85,7 +89,7 @@ def experiment(trial: Trial):
                 trial["prefix__" + k] = v
                 trial["prefix1__" + k] = v
 
-            assert trial.prefixed("prefix__") == seed
+            assert trial.prefixed("prefix__").todict() == seed
 
             # test call
             def identity(a, b, c=4, d=5):
@@ -104,7 +108,7 @@ def identity(a, b, c=4, d=5):
 
             # test record_defaults
             trial.prefixed("prefix2__").record_defaults(identity)
-            assert trial.prefixed("prefix2__") == {"c": 4, "d": 5}
+            assert trial.prefixed("prefix2__").todict() == {"c": 4, "d": 5}
 
             # test call: functools.partial
 
@@ -125,7 +129,7 @@ def identity(a, b, c=4, d=5):
             # Keyword arguments will *not* be recorded and can *not* be overwritten
             identity_a8_kwd = functools.partial(identity, a=8)
             trial.prefixed("prefix4_").record_defaults(identity_a8_kwd)
-            assert trial.prefixed("prefix4_") == {"c": 4, "d": 5}
+            assert trial.prefixed("prefix4_").todict() == {"c": 4, "d": 5}
 
             with pytest.raises(TypeError):
                 trial.prefixed("prefix5_").record_defaults(identity_a8_kwd, a=9)
@@ -146,7 +150,7 @@ def identity(a, b, c=4, d=5):
                 4,
                 9,
             )
-            assert trial.prefixed("prefix7__") == {"c": 4}
+            assert trial.prefixed("prefix7__").todict() == {"c": 4}
 
             with pytest.raises(TypeError):
                 trial.prefixed("prefix7__").call(identity_d9_kwd, 1, 2, 5, 10)
@@ -165,7 +169,7 @@ def identity(a, b, c=4, d=5):
             # setdefaults
             assert trial.prefixed("prefix8__").setdefaults(
                 dict(a=1, b=2, c=3, d=4), e=10
-            ) == dict(a=1, b=2, c=3, d=4, e=10)
+            ).todict() == dict(a=1, b=2, c=3, d=4, e=10)
             assert dict(trial.prefixed("prefix8__")) == dict(a=1, b=2, c=3, d=4, e=10)
 
             # Make sure that the default value is recorded when using .get