You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@camel.apache.org by "Luigi De Masi (Jira)" <ji...@apache.org> on 2026/10/14 17:08:00 UTC
[jira] [Work started] (CAMEL-25668) camel-semantic: Extract TypeSafe adapter and rename Wolf Defender module
[ https://issues.apache.org/jira/browse/CAMEL-25668?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Work on CAMEL-25668 started by Luigi De Masi.
---------------------------------------------
> camel-semantic: Extract TypeSafe adapter and rename Wolf Defender module
> ------------------------------------------------------------------------
>
> Key: CAMEL-25668
> URL: https://issues.apache.org/jira/browse/CAMEL-25668
> Project: Camel
> Issue Type: Task
> Components: camel-ai
> Reporter: Luigi De Masi
> Assignee: Luigi De Masi
> Priority: Major
>
> h2. Purpose
> Give the existing Semantic experts consistent, explicit artifact names while keeping the TypeSafe AI component independently usable without a dependency on camel-semantic.
> This task covers two different packaging changes:
> * Extract the TypeSafe AI semantic adapter from camel-typesafe-ai into a new camel-semantic-typesafe-ai module.
> * Rename the existing camel-wolf-defender module/artifact to camel-semantic-wolf-defender. Wolf Defender is already a Semantic expert module, with its local inference implementation and supporting exception; it has no separate Camel routing endpoint. This is a module rename, not an adapter/backend split.
> h2. TypeSafe AI: extract the adapter
> * Add components/camel-ai/camel-semantic-typesafe-ai, depending on camel-semantic and camel-typesafe-ai.
> * Move TypeSafeAiSemanticAdapter and the applicable Semantic-specific tests to the new module.
> * Keep the component, endpoints, producer, client configuration and managed transport in camel-typesafe-ai. Reuse that transport from the extracted adapter.
> * Remove camel-typesafe-ai's dependency on camel-semantic. Do not introduce a reverse dependency from the component to the new adapter module.
> * Preserve existing operations, accepted inputs, parameters, batching, typed results, provider uncertainty, exceptions and lifecycle behavior.
> h2. Wolf Defender: rename the module
> * Rename components/camel-ai/camel-wolf-defender and its Maven artifact to camel-semantic-wolf-defender.
> * Keep the adapter, inference implementation, supporting exception and tests together. Preserve their Java packages and class names; no inference API extraction or new routing component is required.
> * Keep the existing model provisioning, native dependencies, resource limits, concurrency, lifecycle, failure behavior and expert semantics unchanged.
> h2. Explicit input validation
> * Validate the input shapes and modalities accepted by each expert before provider requests or local inference, including direct adapter calls and batches.
> * TypeSafe AI accepts text and its supported structured application state (String, Map and List). Wolf Defender accepts text only, subject to its existing nonblank, character and token limits.
> * Reject unsupported images, audio, video, binary objects and explicit media parts with clear validation exceptions. Do not stringify unsupported inputs, silently remove parts or invoke preprocessing/fallback services. A structured wrapper must not bypass the checks for explicit media content.
> * Apply the applicable validation at the TypeSafe provider/component boundary as well as its Semantic adapter boundary, without introducing a dependency from camel-typesafe-ai to camel-semantic. Do not require shared Semantic media classes in the provider component.
> * Preserve legitimate structured state and ordinary textual content; do not infer a media modality merely from arbitrary user text or field names.
> * Add positive tests for each supported input shape and negative tests showing rejection before HTTP/model invocation, including a batch with an invalid input. Document any newly enforced rejection of previously accepted unsupported values in the upgrade guide.
> h2. Build, metadata and migration
> * Update reactor modules, dependency management/BOM entries, relevant dependency declarations, generated catalog metadata and documentation mirrors.
> * Update expert artifact identity to the module that contains each adapter, preserving provider identity and operation meaning.
> * Update documentation, examples and applicable Camel JBang dependency resolution/usage references to use the new artifacts. Preserve explicit named expert selection and the existing runtime metadata/tooling contract.
> * Document the migration in the appropriate upgrade guide: TypeSafe Semantic users add camel-semantic-typesafe-ai; direct TypeSafe component users continue using camel-typesafe-ai; Wolf Defender users replace camel-wolf-defender with camel-semantic-wolf-defender.
> * Preserve existing Java adapter class names where practical. Document any necessary TypeSafe bean-class or import changes explicitly rather than silently changing route behavior.
> h2. Acceptance criteria
> # camel-typesafe-ai can build and execute its component functionality without camel-semantic or camel-semantic-typesafe-ai on the application's classpath.
> # camel-semantic-typesafe-ai supplies the existing TypeSafe expert using the component's configured transport, including custom endpoint/client settings and batching.
> # camel-semantic-wolf-defender supplies the same Wolf Defender expert and local inference implementation under the renamed artifact, without introducing a separate backend module or changing the Java API.
> # Dependency graphs are acyclic, and the provider component does not depend on its Semantic adapter module.
> # Existing Semantic behavior remains covered by the moved/adapted tests: explicit registry references, capability inspection, supported inputs and operations, result types and probabilities, validation, exceptions and service lifecycle. Metadata inspection must not load the Wolf model or perform inference.
> # Relevant Java/XML/YAML examples and Camel JBang usage resolve the new artifacts. Existing audit/result rendering continues to use the common Semantic contracts.
> # Build metadata, generated artifacts, documentation and upgrade guidance consistently reflect the new artifact names.
> # Invalid or unsupported inputs fail before provider/model invocation for both experts, with regression tests for supported text/structured inputs and rejected media.
> h2. Scope boundary
> The OpenAI Decisions integration and camel-semantic-openai belong in a separate issue. This refactoring does not add OpenAI support, multimodal inputs, image audit rendering, a Semantic API/runtime split, or new inference behavior.
> h2. References
> * [TypeSafe AI module at the merged Semantic baseline|https://github.com/apache/camel/tree/12e8bdcdf1affa19b3d62786d4a4001db936b164/components/camel-ai/camel-typesafe-ai]
> * [Wolf Defender module at the same baseline|https://github.com/apache/camel/tree/12e8bdcdf1affa19b3d62786d4a4001db936b164/components/camel-ai/camel-wolf-defender]
> * [Merged explicit-expert and audit changes, PR 27791|https://github.com/apache/camel/pull/27791]
> _AI-generated by Codex on behalf of luigidemasi (Luigi De Masi)._
--
This message was sent by Atlassian Jira
(v8.20.10#820010)