• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

supabase / etl / 43827717546
79%

Build:
DEFAULT BRANCH: main
Ran 16 Nov 2026 08:06AM UTC
Jobs 1
Files 267
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

16 Nov 2026 07:59AM UTC coverage: 79.053% (+0.03%) from 79.023%
43827717546

push

github

web-flow
perf(clickhouse): stream INSERT bodies in chunks (#1259)

* bench: add hot-path instruction benchmark

Add a `hot_paths` binary that counts instructions retired and
allocations per row or cell for code that runs once per row, cell, or
replication message. It needs no Postgres or destination, so two builds
can be compared in seconds.

Instructions are the primary metric (MEAS-2): across runs the median
moves by about 0.1%, where wall time moves by several percent. Each
scenario prebuilds fixtures from a seed, discards a warm-up pass, and
keeps its output so a checksum taken outside the measured window proves
the work ran and matches between builds (MEAS-4).

The binary uses jemalloc and installs the Prometheus recorder, as the
replicator does. Instruction counts come from `proc_pid_rusage`, so
they are reported on macOS only; other platforms show `n/a`.

Internal decoders and metric recording are reached through a new
doc-hidden `etl::benchmarking` module behind a `benchmarking` feature,
following the existing `fuzzing` module.

Baseline on this commit (Apple M-series, --repeat 9, 20,000 units):

scenario             instr/unit  spread  allocs/unit
numeric_display          2624.0   0.11%        3.302
date_display             2232.0   0.05%        2.000
time_display             2222.8   0.13%        1.500
timetz_display           2808.8   0.10%        1.498
timestamp_display        4472.2   0.10%        3.502
timestamptz_display      5632.8   0.93%        5.497
timetz_parse             1714.8   1.96%        0.000
copy_row_plain           4442.6   0.39%        5.000
copy_row_escaped        11806.1   2.15%       16.000
copy_row_metrics         1027.5   0.31%        2.016
cdc_insert_decode        6831.8   4.79%        5.752
cdc_update_decode        7224.1   0.55%        5.753
cdc_insert_metrics       3833.4   0.05%        8.016

The in-process counter was cross-checked against `/usr/bin/time -l`:
200 extra passes of `timetz_parse` added 65.9M instructions ... (continued)

494 of 558 new or added lines in 5 files covered. (88.53%)

27 existing lines in 4 files now uncovered.

62246 of 78740 relevant lines covered (79.05%)

58248.65 hits per line

Uncovered Changes

Lines Coverage ∆ File
25
0.0
crates/etl-benchmarks/src/hot_paths/clickhouse.rs
21
95.67
-1.34% crates/etl-destinations/src/clickhouse/encoding.rs
17
0.0
crates/etl-destinations/src/clickhouse/benchmarking.rs
1
93.57
1.19% crates/etl-destinations/src/clickhouse/client.rs

Coverage Regressions

Lines Coverage ∆ File
15
86.66
-0.29% crates/etl-destinations/src/ducklake/core.rs
10
92.14
1.07% crates/etl-destinations/src/ducklake/encoding.rs
1
87.27
-0.21% crates/etl-postgres/src/tokio/test_utils.rs
1
83.95
-0.14% crates/etl/src/error.rs
Jobs
ID Job ID Ran Files Coverage
1 43827717546.1 16 Nov 2026 08:06AM UTC 267
79.05
GitHub Action Run
Source Files on build 43827717546
  • Tree
  • List 267
  • Changed 11
  • Source Changed 5
  • Coverage Changed 10
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #43827717546
  • b8bbac99 on github
  • Prev Build on main (#43824053770)
  • Next Build on main (#43876282423)
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE TRIAL · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc