TRAC D.A.P. 0.4.6

Starting with TRAC D.A.P. 0.4.6, binary distributions of the platform components are available with each release on the GitHub release page. The tracdap-platform package includes distributions of all the service components and supporting tools, suitable for enterprise deployment using containers or physical or virtual hardware. The tracdap-sandbox package includes all the services and tools in a single distribution, for easy setup in development, testing or demo situations.

For more information on deploying the platform, refer to our deployment documentation.

Version 0.4.6 is metadata-compatible with versions 0.4.0 through 0.4.5.

TRAC D.A.P. 0.4.5

TRAC D.A.P. 0.4.5 adds minor fixes and updates for the model runtime:

• Improved null handling in dev mode using the lenient CSV parser
• Performance fix for loading very wide datasets
• Update Apache Arrow to version 9

Version 0.4.5 is metadata-compatible with versions 0.4.0 through 0.4.4.

TRAC D.A.P. 0.4.4

TRAC D.A.P. 0.4.4 adds a few small but helpful featuress to the 0.4 release series:

• use_temporal_objects - flag available in the runtime API, to request temporal fields as Python objects instead of NumPy datetime64 type
• csv.date_format / csv.datetime_format - storage options to allow reading / writing CSV data with non-ISO date/time formats
• Additional validation on the model runtime API
• Fix path translation in the gateway for HTTP routes
• Update web API to use Protobuf JS version 7
• Documentation updates

Version 0.4.4 is metadata-compatible with versions 0.4.0 through 0.4.3.

TRAC D.A.P. 0.4.3

Starting with TRAC D.A.P. 0.4.3, platform JARs are available in Maven Central including JARs for the platform services as well as libraries needed for building back-end plugins. The following features and fixes are also added since 0.4.2:

• Define data schemas using CSV files in the model repository (these can be used for model input/output schemas)
• Model loading now supports nested package structures and resources across multiple model versions (improved shim loader mechanism)
• Controlled tags for timestamp and user name/ID are added to every object
• Improved filesystem handling for local execution, with options to save temp files for debugging
• Fix filesystem permissions issues using the TRAC runtime on Windows, due to recent Windows updates
• Fix handling of Git checkout conflicts

Version 0.4.3 is metadata-compatible with versions 0.4.0 through 0.4.2.

TRAC D.A.P. 0.4.2

TRAC D.A.P. 0.4.2 adds data conformance for primary model data and a preview of the run flow capability.

Version 0.4.2 is metadata-compatible with versions 0.4.1 and 0.4.0.

TRAC D.A.P. 0.4.1

TRAC D.A.P. 0.4.1 adds full validation of metadata and platform APIs.

Version 0.4.1 is metadata-compatible with version 0.4.0.

TRAC D.A.P. 0.4.0

TRAC D.A.P. 0.4.0 sees TRAC become officially part of the FINOS foundation! Everyone involved in TRAC is delighted by this move and we look forward to working as part of the FINOS community.

For version 0.4.0, documenation and packages for TRAC D.A.P. have moved to new locations:

• Documentation: https://tracdap.finos.org/
• Model runtime for Python: https://pypi.org/project/tracdap-runtime/
• Web API: https://www.npmjs.com/package/@finos/tracdap-web-api

TRAC D.A.P. 0.4.0 also includes a preview of the orchestration capability, as well as several minor fixes and enhancements.

The 0.4 release series is not metadata-compatible with the 0.3 series.

TRAC 0.4.0 beta 1

Release series 0.4 signals the move of TRAC D.A.P. into the FINOS foundation. This beta release is to test the packaging and publishing automation ahead of release 0.4.0.

TRAC 0.3.1

This is a security patch release to address vulnerabilities in the Log4j library:
https://www.ncsc.gov.uk/news/apache-log4j-vulnerability

• Update Log4j to version 2.16.0
• Update all other Java dependencies to latest stable versions

Netty and many other Java libraries have released their own fixes to address this issue, which are included in their latest stable versions.

TRAC 0.3.0

TRAC 0.3 adds core data capabilities to the platform. The data engine is based on Apache Arrow, it provides storage abstraction, format translation and data transfer using gRPC streaming methods. Data updates and versioning are supported. Client apps can store, retrieve and manipulate data using standard data formats. This release also includes automated checks for security vulnerabilities and licensing, to let the platform pass typical enterprise software compliance scans.

The 0.3 release series is not metadata-compatible with the 0.2 series.

• TRAC data API definition and service implementation
• Core data engine based on Apache Arrow
• Supported data formats: Arrow streams, Arrow files, CSV, JSON
• Supported storage engines: Local storage
• TRAC plugin API, for adding data formats and storage engines
• Data capabilities added to the web API package
• Documentation, examples and tutorials for application development
• Automated compliance scans for security vulnerabilities and licensing
• Add support for Python 3.10 in the model runtime