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

supabase / etl / 44265060665
79%

Build:
DEFAULT BRANCH: main
Ran 18 Nov 2026 08:21AM UTC
Jobs 1
Files 270
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

18 Nov 2026 08:12AM UTC coverage: 78.487% (-0.02%) from 78.504%
44265060665

push

github

web-flow
ci: run Kani proofs on parsers and encoders every week (#1261)

* ci: prove time parsing with Kani weekly

Add Kani proofs for the byte-level date, time, and timestamp fast
paths in the etl codec and for UTC offset parsing in etl-postgres.
Each proof covers every input of one length: no panic or overflow,
accepted values hold exactly the components their digits spell, and
an accepted date matches chrono. A wrong result here silently shifts
replicated temporal values.

The proofs live in `verification` child modules behind cfg(kani), so
normal builds do not compile them. The offset proofs stub core's
memchr, whose alignment-dependent loops Kani cannot bound.

A weekly workflow installs a pinned Kani, caches it per version, and
runs both crates' proofs.

* ci: prove event version and RowBinary encoding

Add Kani proofs for the values ClickHouse writes per row:

- EventPosition::as_u128 orders exactly like (commit_lsn, tx_ordinal)
  and splits back into both parts. ReplacingMergeTree keeps the row
  with the largest version, so a reordering keeps the wrong row.
- rb_varint emits minimal LEB128 that decodes to the original value
  and keeps earlier buffer contents.
- UUIDs encode as their high then low 64 bits, each little-endian.

RowBinary has no delimiters, so a wrong length or layout shifts every
later field. The UUID proof stubs backtrace capture and avoids
unwrapping the EtlResult: building or Debug-formatting an EtlError
reaches std code that Kani 0.68 cannot compile. The workflow runs the
etl-destinations proofs with the clickhouse feature.

* ci: prove bytea hex and array prefix parsing

Add Kani proofs for two text codec paths every row can reach:

- parse_bytea_hex_string succeeds exactly for `\x` followed by an
  even number of hex digits, and each byte is its digit pair's value.
- strip_array_dimensions_prefix returns input without `[` unchanged,
  removes exactly one well-formed `[lower:upper]=` group, and rejects
  every other prefix, inclu... (continued)

62317 of 79398 relevant lines covered (78.49%)

38591.22 hits per line

Coverage Regressions

Lines Coverage ∆ File
20
90.49
-0.58% crates/etl-destinations/src/ducklake/encoding.rs
6
92.29
-0.24% crates/etl-destinations/src/clickhouse/core.rs
5
77.01
-0.18% crates/etl-maintenance/src/ducklake/runner.rs
1
94.89
-0.12% crates/etl-destinations/src/bigquery/encoding.rs
1
98.26
0.0% crates/etl-destinations/src/ducklake/schema.rs
Jobs
ID Job ID Ran Files Coverage
1 44265060665.1 18 Nov 2026 08:21AM UTC 270
78.49
GitHub Action Run
Source Files on build 44265060665
  • Tree
  • List 270
  • Changed 13
  • Source Changed 6
  • Coverage Changed 13
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #44265060665
  • 586d6626 on github
  • Prev Build on main (#44213223634)
  • Next Build on main (#44277134146)
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