perf: reduce hot-path allocations in resolve pipeline#615
Conversation
🦋 Changeset detectedLatest commit: 31643b6 The changes in this PR will be included in the next version bump. This PR includes changesets to release 1 package
Not sure what this means? Click here to learn what changesets are. Click here if you're a maintainer who wants to add another changeset to this PR |
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #615 +/- ##
=======================================
Coverage 98.09% 98.09%
=======================================
Files 49 49
Lines 9740 9757 +17
=======================================
+ Hits 9554 9571 +17
Misses 186 186
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
Merging this PR will improve performance by 78.07%
|
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ❌ | Memory | pathological-deep-stack: alias chain of 50 (warm) |
14.8 KB | 24.9 KB | -40.53% |
| ❌ | Simulation | realistic-midsize: mixed batch (cold cache) |
7 ms | 8 ms | -12.6% |
| ⚡ | Memory | cache-predicate: mixed cached/uncached requests (warm) |
96.7 KB | 1.5 KB | ×64 |
| ⚡ | Simulation | cache-predicate: mixed cached/uncached requests (warm) |
2.3 ms | 2 ms | +15.65% |
| ⚡ | Memory | symlinks: follow symlinks=true (warm) |
491.6 KB | 425.4 KB | +15.58% |
| ⚡ | Memory | node-compare: enhanced-resolve promise x 1000 (no cache) |
118 KB | 103 KB | +14.55% |
| ⚡ | Simulation | alias-field: browser field (warm) |
1.5 ms | 1.3 ms | +11.13% |
Tip
Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.
Comparing perf/reduce-hot-path-allocations (31643b6) with main (e56eb3f)
- ParsePlugin: replace triple object spread with single spread + direct field assignments, eliminating two property enumerations per resolve (~5-7% throughput improvement on cache-hit benchmarks) - Resolver.resolve(): skip redundant resolveContext wrapper object when no yield callback is present - Resolver.resolve(): defer message template literal construction to the failure-only retry path - Resolver.doResolve(): mutate-and-restore stack on internally-owned context instead of allocating via createInnerContext, saving 3-4 object allocations per resolve on the no-log path - Add Resolver.parse() output shape test to guard against ParsePlugin field drift
xiaoxiaojx
force-pushed
the
perf/reduce-hot-path-allocations
branch
from
ae33455 to
31643b6
Compare
2 participants
Summary
Reduce per-resolve object allocations on the hot path by eliminating redundant spreads, deferring string construction, and reusing context objects. Motivated by profiling the cache-hit resolve path which showed ~15-20 short-lived allocations per resolve. Benchmarks show 5-7% throughput improvement on unsafe-cache-hit workloads with no regressions.
Related: general resolve performance improvement, no linked issue.
What kind of change does this PR introduce?
Performance (perf).
Did you add tests for your changes?
Yes —
test/identifier.test.jsgains aResolver.parse()output shape assertion that guards againstParsePluginfield drift.Does this PR introduce a breaking change?
No. All changes are internal allocation optimizations with identical observable behavior.
If relevant, what needs to be documented once your changes are merged?
n/a
Use of AI
Used Claude Code for analysis, implementation, and benchmarking.
🤖 Generated with Claude Code