vortex-data/vortex - CodSpeed

vortex

Blog Docs Changelog

Performance History

Latest Results

Fix struct binding Signed-off-by: Nicholas Gates <nick@nickgates.com>

ngates/layout27

1 hour ago

Merge develop into layout27 Signed-off-by: "Nicholas Gates" <nick@nickgates.com>

ngates/layout27

2 hours ago

rename Signed-off-by: Robert Kruszewski <github@robertk.io>

rk/skipnans

4 hours ago

Fix bounded max partial state (#8515) This is a breaking change to the serialized form of BoundedMax. The ZonedLayout change to persist these types hasn't been released yet though, so it is still safe to change for now. Signed-off-by: "Nicholas Gates" <nick@nickgates.com>

develop

5 hours ago

fixes Signed-off-by: Robert Kruszewski <github@robertk.io>

rk/skipnans

5 hours ago

ci: install only the uv binary for v4 ingest (drop the workspace sync) Supersedes the earlier reorder. The v4 ingest runs `uv run --no-project --with psycopg[binary] --with boto3 --with xxhash`, which needs only the uv binary and an ephemeral env -- never the workspace. But the "Install uv" step used spiraldb/actions/setup-uv, which runs a full `uv sync --all-extras --dev`: it rebuilds vortex-python via maturin -> sccache -> S3, and on the s3-cache bench runners that fails `s3:GetObject AccessDenied` under the just-assumed ingest role (rds-db:connect only). Production proved the sync is dead weight -- the "Install uv" step went red while the ingest still succeeded, because the binary installs before the sync and the ingest doesn't need it. Switch the three v4 "Install uv" steps to astral-sh/setup-uv (the same v7.6.0 binary spiraldb's action vendors) with no sync. This removes the wasteful workspace build, eliminates the sccache->S3 dependency entirely (so step order no longer matters -- the reorder is reverted), and is much faster. The benchmark job's own non-v4 setup-uv is left as-is (it legitimately needs the workspace). Signed-off-by: Connor Tsui <connor@spiraldb.com>

ct/bench-v4-emitter-uv-binary

7 hours ago

ci: fix v4 ingest uv/sccache ordering; rename commit-metadata workflow (#8516) Two related cleanups to the v4 emitter dual-write workflows. **Fix the v4 ingest sccache failure.** The v4 Postgres dual-write step assumes `GitHubBenchmarkIngestRole` via `configure-aws-credentials`, which persists those credentials as the job's ambient AWS credentials. The ingest role only has `rds-db:connect`, so the immediately-following "Install uv" step — which runs `uv sync` and compiles `vortex-python` through sccache (S3-backed) — intermittently failed with `s3:GetObject AccessDenied` on the benchmark runners (the ones with `extras=s3-cache`). The step is `continue-on-error`, so CI stayed green, but the affected jobs' v4 writes silently dropped. This reorders the v4 block so "Install uv" runs before "Configure AWS credentials": sccache keeps the job's original S3-capable credentials, and the role is assumed immediately before the ingest, which needs only `rds-db:connect`. **Rename the commit-metadata workflow.** `v3-commit-metadata.yml` uploads the `commits` dimension to both the v3 server and the v4 Postgres, so naming it "v3" is misleading — renamed to `commit-metadata.yml` / `Commit metadata`. The per-step `v3 server` / `v4 Postgres` labels stay, since they target specific backends. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Signed-off-by: Connor Tsui <connor@spiraldb.com>

develop

8 hours ago

ci: rename the v3-commit-metadata workflow to commit-metadata The commit-metadata upload feeds BOTH the v3 server and the v4 Postgres (the `commits` dim), so naming it "v3" is misleading. Rename the file and the workflow `name:` to drop "v3"; the per-step `v3 server` / `v4 Postgres` labels stay (they target specific backends). Not a required check and referenced nowhere else, so the rename is self-contained. Signed-off-by: Connor Tsui <connor@spiraldb.com>

ct/bench-v4-emitter-uv-fix

8 hours ago

Latest Branches

-7%

[experiment] Layout Reader#8518

1 hour ago

e0d77d0

ngates/layout27

+1%

Numerical aggregate functions have an option to skip or include nans in calculation, skip by default#8457

4 hours ago

dbd26f8

rk/skipnans

+18%

ci: install only the uv binary for v4 ingest (drop the workspace sync)#8517

8 hours ago

1b7139c

ct/bench-v4-emitter-uv-binary

© 2026 CodSpeed Technology

Home Terms Privacy Docs