perf(builder): fast path for uniform-shape InsertRecords - #60
Open
VojtechVitek wants to merge 2 commits into
Open
perf(builder): fast path for uniform-shape InsertRecords#60VojtechVitek wants to merge 2 commits into
VojtechVitek wants to merge 2 commits into
Conversation
InsertRecords supports mixed-shape batches by unioning columns across rows, which costs a map[string]any plus a padded []any per row. A uniform batch (every record has the same columns — the common case) pays that for nothing. This benchmarks the uniform case as the baseline for the fast path in the next commit. Pure, no database. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
InsertRecords supports mixed-shape batches by unioning columns across
rows, allocating a map[string]any and a padded []any per row. When every
record has the same columns (the common case) that work is wasted.
Map every record once, then compare column slices: if all rows share
row 0's columns, take a fast path that feeds each row's values straight
to the builder — no per-row map, no padding. Mixed-shape batches fall
back to the existing union+DEFAULT path, reusing the already-mapped
columns (no second Map pass). Generated SQL and args are identical; the
shared no-columns error is factored into noColumnsErr.
BenchmarkInsertRecords, count=8, p=0.000:
rows=100 time 117.5µs -> 102.0µs (-13%)
B/op 167.1Ki -> 132.2Ki (-21%)
allocs 3007 -> 2702 (-10%)
Behavior pinned by the existing TestInsertRecords_* suite (uniform,
mixed-shape union/DEFAULT, omitzero/omitempty, all-empty, maps).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
InsertRecordssupports mixed-shape batches (rows with different columns) by unioning columns and emittingDEFAULTfor gaps. To do that it allocates amap[string]anyand a padded[]anyper row. When every record has the same columns — the common case — that work is wasted.Now: Map every record once, then compare column slices. If all rows share row 0's columns, take a fast path that feeds each row's values straight to the builder (no per-row map, no padding). Mixed-shape batches fall back to the existing union+DEFAULT path, reusing the already-mapped columns (no second
Mappass). Generated SQL and args are byte-for-byte identical.Commits
test(builder)—BenchmarkInsertRecordson a uniform batch (10 and 100 rows), incl.ToSql. Baseline. Pure, no DB.perf(builder)— the fast path.Result
BenchmarkInsertRecords, count=8, allp=0.000:Smallest of the batch — squirrel's
ToSqldominates the absolute cost — but removes N map + N slice allocations per batch, scaling with row count.Testing
./tests/...passes.go build ./...,go vet ./clean.🤖 Generated with Claude Code