Skip to main content

riffdb-mlflow

riffdb-mlflow is a compiled RiffDB tracking-store plugin for MLflow. It is an out-of-tree integration: no MLflow-specific code or vocabulary belongs in the RiffDB repository.

The project targets MLflow 3.15.1 and implements the complete AbstractStore tracking, tracing, evaluation assessment, trace analytics, logged-model, and run-lineage surface:

  • Experiments & Runs: atomic experiment creation with tags, lifecycle and tag updates, exact/LIKE/ILIKE experiment search, atomic run creation with tags, run get/update/delete/restore, bounded atomic log_batch (metrics, parameters, tags), get_metric_history, log_inputs, log_outputs, and record_logged_model.
  • Traces & Out-of-Order Spans (TraceInfo, Span): start_trace, get_trace_info, batch_get_traces, log_spans, log_spans_async, ingest_spans_out_of_order, search_traces, delete_traces, set_trace_tag, delete_trace_tag, link_traces_to_run, unlink_traces_from_run, and link_prompts_to_trace. IngestTraceSpans uses init_or_mutate across Trace and TraceSpan in one partition commit, allowing child spans to arrive before parent/root spans and even before start_trace() without foreign-key failures or read-time FINAL merges.
  • Assessments (Feedback & Expectation): create_assessment, get_assessment, update_assessment, and delete_assessment.
  • Trace Metrics & Analytical Rollups (query_trace_metrics, ADR-0315): backed by contract-declared scalar columnar projections (TraceAnalytics, SpanAnalytics, AssessmentAnalytics) and event-driven rollups (SpanModelRollup with min/max/mean/sum/count and hour(...) temporal bucket keys) queried via source projected in .riffq.
  • Logged Models (LoggedModel): create_logged_model, get_logged_model, delete_logged_model, finalize_logged_model, set_logged_model_tags, delete_logged_model_tag, log_logged_model_params, and search_logged_models.

The intended server configuration is:

mlflow server \
  --backend-store-uri riffdb://mlflow-development \
  --default-artifact-root s3://example/mlflow-artifacts

RiffDB TLS and credential material is supplied through protected environment configuration described in DESIGN.md; it is never embedded in the tracking URI.

Development and verification:

uv sync
make check
make conformance

RiffDB compatibility CI

This adapter owns its RiffDB compatibility checks. Pull requests compile the application sources and verify existing locks and generated artifacts against the full commit in .github/riffdb-revision. Scheduled checks use RiffDB main; manual dispatch accepts a full commit or main. Failures are reported here and do not block RiffDB itself. Framework conformance remains separate.

To upgrade, update the pin, migrate sources as needed, review the result of riffdb application lock --write, and verify with riffdb application check. Commit the pin and migration together in this repository.

Metadata

Release files for riffdb-mlflow 0.2.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for riffdb-mlflow 0.2.0
File Size Uploaded Provenance Actions
riffdb_mlflow-0.2.0.tar.gz 278.4 kB Without provenance Details for riffdb_mlflow-0.2.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for riffdb-mlflow 0.2.0
File Interpreter ABI Platform Actions
riffdb_mlflow-0.2.0-py3-none-any.whl Python 3 none any Details for riffdb_mlflow-0.2.0-py3-none-any.whl

Total release size: 350.4 kB

Release files / riffdb_mlflow-0.2.0.tar.gz

Download URL riffdb_mlflow-0.2.0.tar.gz
Size 278.4 kB
Tags Source
SHA-256 checksum
How to use checksums
ea3e6c7120f2f86a5356e245eef04ab3bbde8c3a12990d9d5b5ec8ced9f691f4
BLAKE2b-256 checksum
How to use checksums
b6eba6d0d4f6f992ed3f21f8856122ef6f597ea152d431f06d6de2f4d83440ff
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.12.3

Release files / riffdb_mlflow-0.2.0-py3-none-any.whl

Download URL riffdb_mlflow-0.2.0-py3-none-any.whl
Size 72.0 kB
Tags Python 3
SHA-256 checksum
How to use checksums
897a81b75cff98619f9aba0b4b73b3c57a8f49d937f5e2ba9e5481955225604b
BLAKE2b-256 checksum
How to use checksums
f779c61aeaebe5c49480895d1965f41435858d289e9e878af826ae4ec6733307
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.12.3

Security for riffdb-mlflow 0.2.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page