luigidemasi opened a new pull request, #27578: URL: https://github.com/apache/camel/pull/27578
Adds `camel-test-infra-wolf-defender` so the model integration tests can provision their own pinned files. With `-DwolfDefender.download=true`, the service downloads and verifies the model, tokenizer and configuration, then reuses them under `~/.camel-test/wolf-defender/<revision>`. An existing directory supplied through `wolfDefender.modelDirectory` takes precedence and is never modified. The service uses the test-infra JUnit lifecycle, a cache lock shared across processes, and checksum verification before publishing downloads. Inference still runs locally in the Camel component. Ordinary tests do not download model files, and model tests skip unsupported native platforms before provisioning. The component documentation and generated catalog mirror describe both modes. This follows #27564 and builds on the platform guard in #27577. The IT now uses a local condition method with a direct Java call to the shared platform check, addressing the string-reference concern raised there. Validation: - 10 test-infra tests and 34 component unit tests passed. - Both real-model integration tests passed with a fresh download and with existing files. - Cached and existing files were verified with network access blocked; architecture-property probes confirmed `ppc64le` and `s390x` skip before cache creation. These probes ran on x86 hardware. - Source checks and the full 703-module Camel build passed; full-reactor tests were skipped. _AI-generated by Codex on behalf of @luigidemasi._ -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
