TRAC D.A.P. 0.7.0

TRAC D.A.P. 0.7 marks an important step in the journey to scale and amplify the TRAC platform. Having been incubated in Accenture - and Parker Fitzgerald before that - the TRAC open-source project is now under the stewardship of finTRAC. In addition to future releases of the open-source TRAC D.A.P. project, commercially supported versions of the platform will be launched across all the major cloud marketplaces over the next few months.

TRAC D.A.P. 0.7.0 is metadata-compatible with versions 0.6.0 and later. See the release notes for 0.6 regarding compatibility with the 0.5 series.

TRAC D.A.P. 0.7.0 RC 1

This is a release candidate for TRAC D.A.P. 0.7.0, to validate the release process and run regression testing ahead of the 0.7 release.

TRAC D.A.P. 0.6.6

This release adds incremental features and fixes to the 0.6 release series:

• Runtime data protocols - Support multiple data frameworks in a unified API with intelligent type hinting
• SQ: data import / export - For runtime import / export jobs, support SQL as a data source (MySQL, MariaDB, Postgres, SQL Server)
• Runtime job groups - Allow jobs to depend on child jobs, support for parallel and sequential job groups
• Fix for gRPC request queuing in platform components

Runtime data protocols and SQL import / export are considered experimental, to use these features in model code import from tracdap.rt.api.experimental. Features will be moved to the main API after feedback and evaluation.

Full Changelog: v0.6.5...v0.6.6

TRAC D.A.P. 0.6.6 is metadata-compatible with versions 0.6.0 and later. See the release notes for 0.6 regarding compatibility with the 0.5 series.

TRAC D.A.P. 0.6.5

This release adds incremental features and fixes to the 0.6 release series:

• Support for the Polars data framework in TRAC models
• Experimental API for custom import / export models to read and write data from external storage
• Metadata Resource API, allows clients to query public information about platform resources
• Performance and scheduling improvements in the data service
• Make pure-Python the default implementation for Git (native Git is still available as a config option)

Full Changelog: v0.6.4...v0.6.5

TRAC D.A.P. 0.6.5 is metadata-compatible with versions 0.6.0 and later. See the release notes for 0.6 regarding compatibility with the 0.5 series.

TRAC D.A.P. 0.6.4

This release adds incremental features and fixes to the 0.6 release series:

• Support dynamic schemas for model inputs and outputs (schemas are resolved at runtime)
• Fix for custom config loaders in a local development environment
• Documentation updates
• Updated service control scripts

Full Changelog: v0.6.3...v0.6.4

TRAC D.A.P. 0.6.4 is metadata-compatible with versions 0.6.0 and later. See the release notes for 0.6 regarding compatibility with the 0.5 series.

TRAC D.A.P. 0.6.3

This release is focused on internal improvements for the 0.6 series. There are no changes in the model API.

• Runtime API Server (gRPC)
• Executor framework updates
• Plugin interface for config loading in the runtime
• Add application-owned props at various points in the metadata model
• Security and compliance updates

Full Changelog: v0.6.2...v0.6.3

TRAC D.A.P. 0.6.3 is metadata-compatible with versions 0.6.0 and later. See the release notes for 0.6 regarding compatibility with the 0.5 series.

TRAC D.A.P. 0.6.2

This release adds incremental features and improvements to the 0.6 release series:

• Optional model inputs and outputs
• Structured error information available after failed API calls
• Improved error reporting for job consistency errors prior to execution
• Improved type script support in the web API bindings
• Updates to the modelling tutorials and reference documentation
• Internal improvements relating to gRPC support

Full Changelog: v0.6.1...v0.6.2

TRAC D.A.P. 0.6.2 is metadata-compatible with versions 0.6.0 and later. See the release notes for 0.6 regarding compatibility with the 0.5 series.

TRAC D.A.P. 0.6.1

This is a patch release to add a few minor capabilities and fixes onto 0.6.0.

• More SQL drivers are now available as part of the TRAC plugins package
• HTTP redirect support added to the gateway, configured in trac-platform.yaml
• Fix handling of paths longer than MAX_PATH length on Windows
• Fix handling of quoted arguments in the Windows service control scripts

Full Changelog: v0.6.0...v0.6.1

The TRAC 0.6 series is metadata-compatible with the 0.5 series.

TRAC D.A.P. 0.6.0

The 0.6 release series of TRAC D.A.P. adds a number of significant new capabilities to the platform, most notably storage plugins for the AWS, Azure and GCP cloud platforms. Support for all the major languages and 3rd party libraries is updated to the latest stable versions, with EOL versions being dropped. This release also includes a number of compliance updates and bug fixes, as well as requested minor features.

Changes in this version include:

• Storage plugins for AWS, Azure and GCP (AWS plugin is revised from the 0.5 series)
• Support for Python 3.12
• Support for Pandas 2.x
• Support for Java 21
• Revised storage layer implementation (platform and runtime)
• Improvements in model validation and error reporting
• Native HTTP download for FILE objects
• New job cache plugin
• Simplified platform configuration (gateway config is no longer separate)
• Deprecated APIs have been removed
• Support for Python 3.7 has been dropped (Python 3.7 is EOL)

Full Changelog: v0.5.28...v0.6.0

TRAC D.A.P. 0.6 is compatible with the 0.5.x series, including being metadata compatible, so long as no deprecated APIs or EOL language / library versions are in use. It is recommended to backup your data storage and metadata before upgrading.

TRAC D.A.P. 0.6.0 RC 3

This RC is provided to test the upcoming release of TRAC 0.6.0. The major updates since RC 2 are:

• Java 21 support
• Python 3.12 support
• Upgrade to Apache Arrow 16
• New job cache plugin (to support distributed deployment of the orchestrator)
• Config file changes (Separate trac-platform.yaml removed, better support for deployment layouts)
• All dependencies updated to latest stable
• Various compliance updates and small fixes

NumPy 2.0 is not supported at this time, since the ABI is incompatible with NumPy 1.0 and there is not yet significant uptake in the Python ecosystem.

Full Changelog: v0.6.0-rc2...v0.6.0-rc3