Remove Sample comparison hack

The Sample object implemented comparisons by only comparing timestamps.
This was only done to cope with a limitation of `bisect()` before Python
3.10 (where no key argument could be passed).

Now that we are using Python 3.11 as the minimal version, we can remove
that hack and make Samples compare using both the datetime and the value
as one would expect.

Signed-off-by: Leandro Lucarella <leandro.lucarella@frequenz.com>

Author: llucax

RELEASE_NOTES.md

@@ -14,6 +14,8 @@ This release drops support for Python versions older than 3.11.
 
 * The `LogicalMeter` no longer takes a `component_graph` parameter.
 
+* Now `frequenz.sdk.timeseries.Sample` uses a more sensible comparison.  Before this release `Sample`s were compared only based on the `timestamp`.  This was due to a limitation in Python versions earlier than 3.10.  Now that the minimum supported version is 3.11 this hack is not needed anymore and `Sample`s are compared using both `timestamp` and `value` as most people probably expects.
+
 ## New Features
 
 <!-- Here goes the main new features and examples or instructions on how to use them -->

src/frequenz/sdk/timeseries/_base_types.py

@@ -12,10 +12,6 @@
 """The UNIX epoch (in UTC)."""
 
 
-# Ordering by timestamp is a bit arbitrary, and it is not always what might be
-# wanted. We are using this order now because usually we need to do binary
-# searches on sequences of samples, and the Python `bisect` module doesn't
-# support providing a key until Python 3.10.
 @dataclass(frozen=True, order=True)
 class Sample:
     """A measurement taken at a particular point in time.
@@ -25,10 +21,10 @@ class Sample:
     coherent view on a group of component metrics for a particular timestamp.
     """
 
-    timestamp: datetime = field(compare=True)
+    timestamp: datetime
     """The time when this sample was generated."""
 
-    value: float | None = field(compare=False, default=None)
+    value: float | None = None
     """The value of this sample."""
 
 

src/frequenz/sdk/timeseries/_resampling.py

@@ -705,11 +705,12 @@ def resample(self, timestamp: datetime) -> Sample:
         )
         minimum_relevant_timestamp = timestamp - period * conf.max_data_age_in_periods
 
-        # We need to pass a dummy Sample to bisect because it only support
-        # specifying a key extraction function in Python 3.10, so we need to
-        # compare samples at the moment.
-        min_index = bisect(self._buffer, Sample(minimum_relevant_timestamp, None))
-        max_index = bisect(self._buffer, Sample(timestamp, None))
+        min_index = bisect(
+            self._buffer,
+            minimum_relevant_timestamp,
+            key=lambda s: s.timestamp,
+        )
+        max_index = bisect(self._buffer, timestamp, key=lambda s: s.timestamp)
         # Using itertools for slicing doesn't look very efficient, but
         # experiments with a custom (ring) buffer that can slice showed that
         # it is not that bad. See: