Experiment.summary() should show all columns when summary_columns=None

expt/data.py

@@ -886,18 +886,24 @@ def _df(self) -> pd.DataFrame:
         },
     })
 
-    if self._summary_columns:
-      # TODO: h.summary is expensive and slow, cache it
+    def _append_summary(summary_columns):
+      nonlocal df
       df = pd.concat([
           df,
           pd.DataFrame({
               k: [
-                  h.summary(columns=self._summary_columns).loc[0, k]
+                  # TODO: h.summary is expensive and slow, cache it
+                  h.summary(columns=summary_columns).loc[0, k]
                   for h in self._hypotheses.values()
-              ] for k in self._summary_columns
+              ] for k in summary_columns
           }),
       ], axis=1)  # yapf: disable
 
+    if self._summary_columns is not None:
+      _append_summary(summary_columns=self._summary_columns)
+    else:
+      _append_summary(summary_columns=self.columns)
+
     # Need to sort index w.r.t the multi-index level hierarchy, because
     # the order of hypotheses being added is not guaranteed
     df = df.set_index([*self._config_keys, 'name']).sort_index()

expt/data_test.py

@@ -566,6 +566,9 @@ def test_create_simple(self):
     assert len(ex.hypotheses) == 1
     assert ex.name == "one_hypo"
 
+    assert ex._summary_columns is None
+    assert list(ex._df.columns) == ["hypothesis", "x", "y", "z"]
+
   def test_create_from_dataframe_run(self, runs_gridsearch: RunList):
     """Tests Experiment.from_dataframe with the minimal defaults."""
 
@@ -601,6 +604,9 @@ def test_create_from_dataframe_run_multicolumns(self,
     assert list(ex._df.index.names) == ['algo', 'name']  # Note the order
     # yapf: enable
 
+    # All other columns than by = "algo"
+    assert ex._summary_columns == ("env_id", "seed")
+
     # by: not exists?
     with pytest.raises(KeyError):  # TODO: Improve exception
       ex = Experiment.from_dataframe(df, by="unknown", name="Exp.foobar")