fix: report versioned Fabric runtime metadata

[Rana Singh (ranadeepsingh)]

Summary

Update SynapseML Fabric LLM extended properties to report a versioned runtime:

{"feature":"synapseml","runtime":"fabric_spark_<spark_version>|fabric_python_<python_version>|fabric|unknown"}

Runtime selection matches SynapseML-Internal:

1. When Fabric detection is false, report unknown for local, Databricks, and other non-Fabric environments.
2. On Fabric, prefer fabric_spark_<spark_version>.
3. If Spark is unavailable, use fabric_python_<python_version>.
4. Otherwise use fabric.

Implementation

• Uses PlatformDetails.FabricRuntime as the shared runtime taxonomy owner.
• Safely treats unavailable Spark linkage as an absent Spark version, preserving the Python and generic Fabric fallbacks.
• Uses a Scala lazy val, so runtime detection runs once per JVM.
• Builds the complete immutable Fabric classification header map once per JVM, including ExtendedProperties JSON.
• Reuses that map across request rows; only request-specific custom-header sanitization and merging remains per request.
• Keeps existing Fabric-only and implicit-endpoint guards unchanged.
• SynapseML intentionally has no function field because this shared OpenAI transport represents generic SynapseML stages rather than an individual AI Functions operation.

Headers must still be attached to every outbound HTTP request, but their runtime detection and fixed value construction are not repeated.

Validation

• VerifyPlatformDetails: 11 passed.
• OpenAIFabricHeadersSuite: 7 passed.
• Main and test compilation passed with JDK 11.
• Scala and test scalastyle passed.
• Latest automated review covers the current head with no findings.
• Full Azure CI was run twice; relevant completed tests were green, but both runs were blocked by external dependency/feed infrastructure failures also reproduced on master.

References

• AB#5533937 - Add Fabric LLM taxonomy headers to SynapseML
• Prior header PR #2656
• Python AI Functions PR #2253680
• Fabric taxonomy and class-of-service guidance

[azure-pipelines]

Azure Pipelines:
There may be pipelines that require an authorized user to comment /azp run to run.

[github-actions]

Hey Rana Singh (@ranadeepsingh) 👋!
Thank you so much for contributing to our repository 🙌.
Someone from SynapseML Team will be reviewing this pull request soon.

We use semantic commit messages to streamline the release process.
Before your pull request can be merged, you should make sure your first commit and PR title start with a semantic prefix.
This helps us to create release messages and credit you for your hard work!

Examples of commit messages with semantic prefixes:

• fix: Fix LightGBM crashes with empty partitions
• feat: Make HTTP on Spark back-offs configurable
• docs: Update Spark Serving usage
• build: Add codecov support
• perf: improve LightGBM memory usage
• refactor: make python code generation rely on classes
• style: Remove nulls from CNTKModel
• test: Add test coverage for CNTKModel

To test your commit locally, please follow our guild on building from source.
Check out the developer guide for additional guidance on testing your change.

[Copilot]

Pull request overview

This PR updates Fabric LLM taxonomy metadata so SynapseML reports a versioned Fabric runtime string in X-Taxonomy-ExtendedProperties, enabling downstream workload classification to distinguish Spark vs Python runtimes (and include their versions).

Changes:

• Introduces PlatformDetails.FabricRuntime (lazy) and resolveFabricRuntime(...) to produce fabric_spark_<sparkVersion>, fabric_python_<pythonVersion>, or fabric.
• Switches OpenAI Fabric taxonomy header generation to use PlatformDetails.FabricRuntime instead of CurrentPlatform.
• Adds/updates targeted Scala tests for the new runtime resolution behavior and updated header expectation.

Show a summary per file

File	Description
core/src/main/scala/com/microsoft/azure/synapse/ml/logging/common/PlatformDetails.scala	Adds Fabric-specific runtime resolution and exposes FabricRuntime for shared use.
core/src/test/scala/com/microsoft/azure/synapse/ml/logging/common/VerifyPlatformDetails.scala	Adds unit tests validating FabricRuntime stability and fallback ordering.
cognitive/src/main/scala/com/microsoft/azure/synapse/ml/services/openai/OpenAI.scala	Updates Fabric header runtime attribution to use PlatformDetails.FabricRuntime.
cognitive/src/test/scala/com/microsoft/azure/synapse/ml/services/openai/OpenAIFabricHeadersSuite.scala	Updates expected runtime string in extended properties header assertions.

Review details

💡 Configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

• Files reviewed: 4/4 changed files
• Comments generated: 1
• Review effort level: Lite

[Copilot]

Review details

• Files reviewed: 4/4 changed files
• Comments generated: 0 new
• Review effort level: Lite

[Copilot]

Review details

Suppressed comments (1)

core/src/main/scala/com/microsoft/azure/synapse/ml/logging/common/PlatformDetails.scala:20

• FabricRuntime is currently always resolved from Option(org.apache.spark.SPARK_VERSION), which is effectively always Some(nonEmpty) when Spark is on the classpath. That means the advertised fallback to fabric_python_<python_version> / fabric when Spark is unavailable will never be taken, which conflicts with the PR description’s runtime selection rules.

  lazy val FabricRuntime: String = resolveFabricRuntime(
    runningOnFabric(),
    Option(org.apache.spark.SPARK_VERSION),
    sys.env.get("PYTHON_VERSION")
  )

• Files reviewed: 4/4 changed files
• Comments generated: 0 new
• Review effort level: Lite

[Rana Singh (ranadeepsingh)]

Addressed the current-head suppressed review finding in 9fbfbc5: Spark version lookup now safely handles unavailable Spark linkage, while retaining Spark-first behavior when packaged. Local Scala style and targeted suites pass.

[Rana Singh (ranadeepsingh)]

/azp run

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).

[Copilot]

Review details

• Files reviewed: 4/4 changed files
• Comments generated: 0 new
• Review effort level: Lite

[Rana Singh (ranadeepsingh)]

/azp run

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).

[Codecov Comments Bot (codecov-commenter)]

Codecov Report

❌ Patch coverage is 94.11765% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 87.19%. Comparing base (a6fd536) to head (9fbfbc5).

Files with missing lines	Patch %	Lines
...re/synapse/ml/logging/common/PlatformDetails.scala	80.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2665      +/-   ##
==========================================
- Coverage   87.19%   87.19%   -0.01%     
==========================================
  Files         337      337              
  Lines       19554    19559       +5     
  Branches     1894     1942      +48     
==========================================
+ Hits        17050    17054       +4     
- Misses       2504     2505       +1     

Files with missing lines	Coverage Δ
...soft/azure/synapse/ml/services/openai/OpenAI.scala	90.86% <100.00%> (ø)
...re/synapse/ml/logging/common/PlatformDetails.scala	69.23% <80.00%> (+2.56%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[Rana Singh (ranadeepsingh)]

PR-loop validation is complete for the current head (\9fbfbc5): latest automated review covers the head with no findings, all review threads are resolved, Scala/test scalastyle passed, \VerifyPlatformDetails\ passed 11/11, and \OpenAIFabricHeadersSuite\ passed 7/7.

Full Azure CI was run twice: 231733174 and 231746576. Both were blocked before relevant product tests by external infrastructure failures (unresolved Maven dependency; the rerun also hit conda-forge HTTP 403 and an 8.6-GB cache download timeout). The same Maven resolution failure is present on a recent \master\ build 231649114. Relevant published test runs that completed were green; no product-code test failure was identified.

[Rana Singh (ranadeepsingh)]

/copilot review

[Rana Singh (ranadeepsingh)]

Header-capitalization audit completed across SynapseML, SynapseML-Internal, and workload-ml. The canonical wire names are \X-Taxonomy-TrafficType, \X-Taxonomy-ExtendedProperties, and \x-llm-service-tier. Commit \�365962e1b\ removes the misleading capitalized Scala header-name fields and inlines those exact documented literals in the cached map. SynapseML-Internal and workload-ml already emit the same exact names, so they require no follow-up change.

[Copilot]

Review details

• Files reviewed: 4/4 changed files
• Comments generated: 0 new
• Review effort level: Lite

[Rana Singh (ranadeepsingh)]

/copilot review

[Rana Singh (ranadeepsingh)]

Corrected the emitted service-tier wire name to the documented \X-Llm-Service-Tier\ in commit \c7067331ca. The exact-casing regression test now asserts this spelling; the targeted suite passes 7/7 and Scala/test scalastyle are clean.

[Copilot]

Review details

• Files reviewed: 4/4 changed files
• Comments generated: 0 new
• Review effort level: Lite

[Rana Singh (ranadeepsingh)]

/azp run

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).