Files
gitlore/src/core/backoff.rs
Taylor Eernisse 72f1cafdcf perf: Optimize SQL queries and reduce allocations in hot paths
Change detection queries (embedding/change_detector.rs):
- Replace triple-EXISTS subquery pattern with LEFT JOIN + NULL check
- SQLite now scans embedding_metadata once instead of three times
- Semantically identical: returns docs needing embedding when no
  embedding exists, hash changed, or config mismatch

Count queries (cli/commands/count.rs):
- Consolidate 3 separate COUNT queries for issues into single query
  using conditional aggregation (CASE WHEN state = 'x' THEN 1)
- Same optimization for MRs: 5 queries reduced to 1

Search filter queries (search/filters.rs):
- Replace N separate EXISTS clauses for label filtering with single
  IN() clause with COUNT/GROUP BY HAVING pattern
- For multi-label AND queries, this reduces N subqueries to 1

FTS tokenization (search/fts.rs):
- Replace collect-into-Vec-then-join pattern with direct String building
- Pre-allocate capacity hint for result string

Discussion truncation (documents/truncation.rs):
- Calculate total length without allocating concatenated string first
- Only allocate full string when we know it fits within limit

Embedding pipeline (embedding/pipeline.rs):
- Add Vec::with_capacity hints for chunk work and cleared_docs hashset
- Reduces reallocations during embedding batch processing

Backoff calculation (core/backoff.rs):
- Replace unchecked addition with saturating_add to prevent overflow
- Add test case verifying overflow protection

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-02-05 11:21:28 -05:00

93 lines
2.8 KiB
Rust

use rand::Rng;
pub fn compute_next_attempt_at(now: i64, attempt_count: i64) -> i64 {
let capped_attempts = attempt_count.min(30) as u32;
let base_delay_ms = 1000_i64.saturating_mul(1 << capped_attempts);
let capped_delay_ms = base_delay_ms.min(3_600_000);
let jitter_factor = rand::thread_rng().gen_range(0.9..=1.1);
let delay_with_jitter = (capped_delay_ms as f64 * jitter_factor) as i64;
now.saturating_add(delay_with_jitter)
}
#[cfg(test)]
mod tests {
use super::*;
const MAX_DELAY_MS: i64 = 3_600_000;
#[test]
fn test_exponential_curve() {
let now = 1_000_000_000_i64;
for attempt in 1..=10 {
let result = compute_next_attempt_at(now, attempt);
let delay = result - now;
let expected_base = 1000_i64 * (1 << attempt);
let min_expected = (expected_base as f64 * 0.89) as i64;
let max_expected = (expected_base as f64 * 1.11) as i64;
assert!(
delay >= min_expected && delay <= max_expected,
"attempt {attempt}: delay {delay} not in [{min_expected}, {max_expected}]"
);
}
}
#[test]
fn test_cap_at_one_hour() {
let now = 1_000_000_000_i64;
for attempt in [20, 25, 30, 50, 100] {
let result = compute_next_attempt_at(now, attempt);
let delay = result - now;
let max_with_jitter = (MAX_DELAY_MS as f64 * 1.11) as i64;
assert!(
delay <= max_with_jitter,
"attempt {attempt}: delay {delay} exceeds cap {max_with_jitter}"
);
}
}
#[test]
fn test_jitter_range() {
let now = 1_000_000_000_i64;
let attempt = 5;
let base = 1000_i64 * (1 << attempt);
let min_delay = (base as f64 * 0.89) as i64;
let max_delay = (base as f64 * 1.11) as i64;
for _ in 0..100 {
let result = compute_next_attempt_at(now, attempt);
let delay = result - now;
assert!(
delay >= min_delay && delay <= max_delay,
"delay {delay} not in jitter range [{min_delay}, {max_delay}]"
);
}
}
#[test]
fn test_first_retry_is_about_two_seconds() {
let now = 1_000_000_000_i64;
let result = compute_next_attempt_at(now, 1);
let delay = result - now;
assert!(
(1800..=2200).contains(&delay),
"first retry delay: {delay}ms"
);
}
#[test]
fn test_overflow_safety() {
let now = i64::MAX / 2;
let result = compute_next_attempt_at(now, i64::MAX);
assert!(result > now);
}
#[test]
fn test_saturating_add_prevents_overflow() {
let now = i64::MAX - 10;
let result = compute_next_attempt_at(now, 30);
assert_eq!(result, i64::MAX);
}
}