Remove infill 0 buckets from custom distributions

Author: badboy

CHANGELOG.md

@@ -2,6 +2,8 @@
 
 [Full changelog](https://github.com/mozilla/glean/compare/v65.2.1...main)
 
+* General
+  * BREAKING: Remove infill 0 buckets from custom distributions ([#3246](https://github.com/mozilla/glean/pull/3246))
 * Swift
   * Make `EventMetricType`, `ObjectMetricType`, `URLMetricType` and `Ping` `Sendable`  ([#3255](https://github.com/mozilla/glean/pull/3255))
   * Glean for iOS is now being built with Xcode 16.4 ([#3270](https://github.com/mozilla/glean/pull/3270))

docs/dev/core/internal/payload.md

@@ -228,8 +228,8 @@ A [Custom distribution](../../../book/reference/metrics/custom_distribution.md)
 | `sum` | Integer | The sum of all recorded values. |
 | `values` | Map<String, Integer> | The values in each bucket. The key is the minimum value for the range of that bucket. |
 
-A contiguous range of buckets is always sent, so that the server can aggregate and visualize distributions, without knowing anything about the specific bucketing function used.
-This range starts with the first bucket (as specified in the `range_min` parameter), and ends at one bucket beyond the last bucket with a non-zero accumulation (so that the upper bound on the last bucket is retained).
+From Glean v66.0.0 on the `values` only includes filled buckets.
+Empty buckets are not sent.
 
 For example, suppose you had a custom distribution defined by the following parameters:
 
@@ -241,23 +241,21 @@ For example, suppose you had a custom distribution defined by the following para
 The following shows the recorded values vs. what is sent in the payload.
 
 ```
-recorded:        12: 2,                      22: 1
-sent:     10: 0, 12: 2, 14: 0, 17: 0, 19: 0, 22: 1, 24: 0
+recorded: 12: 2, 22: 1
+sent:     12: 2, 22: 1
 ```
 
+Up until Glean v65.0.3 a contiguous range of buckets was always sent, so that the server can aggregate and visualize distributions, without knowing anything about the specific bucketing function used.
+This range started with the first bucket (as specified in the `range_min` parameter), and ends at one bucket beyond the last bucket with a non-zero accumulation (so that the upper bound on the last bucket is retained).
+
 #### Example:
 
 ```json
 {
     "sum": 3,
     "values": {
-        "10": 0,
         "12": 2,
-        "14": 0,
-        "17": 0,
-        "19": 0,
         "22": 1,
-        "24": 0
     }
 }
 ```

glean-core/src/histogram/exponential.rs

@@ -57,7 +57,7 @@ fn exponential_range(min: u64, max: u64, bucket_count: usize) -> Vec<u64> {
 ///
 /// Buckets are pre-computed at instantiation with an exponential distribution from `min` to `max`
 /// and `bucket_count` buckets.
-#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq, MallocSizeOf)]
+#[derive(Debug, Clone, Serialize, Deserialize, Eq, MallocSizeOf)]
 pub struct PrecomputedExponential {
     // Don't serialize the (potentially large) array of ranges, instead compute them on first
     // access.
@@ -68,6 +68,14 @@ pub struct PrecomputedExponential {
     pub(crate) bucket_count: usize,
 }
 
+impl PartialEq for PrecomputedExponential {
+    fn eq(&self, other: &Self) -> bool {
+        // `bucket_ranges` are computed on-demand based on `min`, `max` and `bucket_count`,
+        // so we don't need to compare it (and thus forcing it to be computed)
+        self.min == other.min && self.max == other.max && self.bucket_count == other.bucket_count
+    }
+}
+
 impl Bucketing for PrecomputedExponential {
     /// Get the bucket for the sample.
     ///

glean-core/src/histogram/linear.rs

@@ -37,7 +37,7 @@ fn linear_range(min: u64, max: u64, count: usize) -> Vec<u64> {
 ///
 /// Buckets are pre-computed at instantiation with a linear  distribution from `min` to `max`
 /// and `bucket_count` buckets.
-#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq, MallocSizeOf)]
+#[derive(Debug, Clone, Serialize, Deserialize, Eq, MallocSizeOf)]
 pub struct PrecomputedLinear {
     // Don't serialize the (potentially large) array of ranges, instead compute them on first
     // access.
@@ -48,6 +48,12 @@ pub struct PrecomputedLinear {
     pub(crate) bucket_count: usize,
 }
 
+impl PartialEq for PrecomputedLinear {
+    fn eq(&self, other: &Self) -> bool {
+        self.min == other.min && self.max == other.max && self.bucket_count == other.bucket_count
+    }
+}
+
 impl Bucketing for PrecomputedLinear {
     /// Get the bucket for the sample.
     ///

glean-core/src/histogram/mod.rs

@@ -124,19 +124,7 @@ impl<B: Bucketing> Histogram<B> {
     /// Gets a snapshot of all values from the first bucket until one past the last filled bucket,
     /// filling in empty buckets with 0.
     pub fn snapshot_values(&self) -> HashMap<u64, u64> {
-        let mut res = self.values.clone();
-
-        let max_bucket = self.values.keys().max().cloned().unwrap_or(0);
-
-        for &min_bucket in self.bucketing.ranges() {
-            // Fill in missing entries.
-            let _ = res.entry(min_bucket).or_insert(0);
-            // stop one after the last filled bucket
-            if min_bucket > max_bucket {
-                break;
-            }
-        }
-        res
+        self.values.clone()
     }
 
     /// Clear this histogram.