Skip to content

Releases: antonbarr-data/dqt

v1.4.0

26 May 19:06
@antonbarr-data antonbarr-data
dqt-v1.4.0
77d2893

Choose a tag to compare

View all tags
v1.4.0 Latest
Latest

v1.4.0 - 2026-05-26

Metrics dashboard

  • Gmail-style row checkboxes with select-all (indeterminate state)
  • YAML import / export for metric definitions
  • Bulk delete with confirmation
  • Inline row editing (display_name, kind, dataset, description)
  • Excel-style multi-select column filter headers (matching Checks page)
  • Removed verdict dot badges from table rows

Checks page

  • Edit button replaced with pencil icon for visual consistency

Remove calculated-metric authoring (breaking)

The expression-builder path (SQL expression authoring, evaluate endpoints, MetricRun
table) has been removed. dqt is a data questioning tool that binds to existing warehouse
columns -- it does not compute metric values.

  • Removed dqt.subscriptions module entirely (was an in-memory stub, no real delivery)
  • Removed expr_type, expr_sql, numerator_sql, denominator_sql, filter_sql,
    time_column fields from MetricDefinition
  • Removed /metrics/evaluate-expression and /metrics/{fqn}/evaluate API endpoints
  • Dropped metric_runs table (migration runs DROP TABLE IF EXISTS metric_runs)
  • Removed ExpressionBuilder frontend component

Other

  • Fixed null crash in tasks page when API is unreachable during static prerender
  • Added /api/v1/trigger/history endpoint
  • README rewritten to ~130 lines with links to detailed docs
Assets 2
Loading

dqt v1.0.0

14 May 12:57
@antonbarr-data antonbarr-data
dqt-v1.0.0
1937ac4

Choose a tag to compare

View all tags
dqt v1.0.0

First stable release. The public library API is now frozen — no breaking changes
in any v1.x release. See STABILITY.md for the full compatibility commitment.

What's in v1.0.0

64 detectors — benchmarked and documented
Every detector in the registry has a results row in examples/benchmarks/results.csv
with precision, recall, and F1 across synthetic and real-world benchmark datasets.
Docs at docs/algorithms/.

Benchmark suite
python scripts/run_benchmark_suite.py --quick runs the full suite in under 60s.
8 Jupyter notebooks split by detector group in examples/benchmarks/.
CI runs on every push to main and weekly.

Public quality dashboard
https://dqt.dev/quality — per-detector F1, sortable and filterable by family.

Live adapter tests
Integration tests for ClickHouse, Snowflake, BigQuery, and Databricks run nightly
in CI when credentials are available. Fixtures for offline development via vcrpy.

API stability commitment
STABILITY.md documents the stable public surface and the v1.x compatibility policy.

Install

pip install dqtlib==1.0.0

Links

Loading