rzo1 opened a new issue, #1522: URL: https://github.com/apache/opennlp/issues/1522
Since about Oct 28, every CI build on main and opennlp-2.x fails in tests that download the 1.5 models from `https://opennlp.sourceforge.net/models-1.5/`. The last green builds are from Oct 27. What happened: - `https://opennlp.sourceforge.net/` (project web space) still answers 200. It's the old redirect page to incubator.apache.org. - `https://opennlp.sourceforge.net/models-1.5/` and every file below it return 404. - The same files are still in the SourceForge file release area. `https://downloads.sourceforge.net/project/opennlp/models-1.5/<name>` works, e.g. `en-pos-maxent.bin` returns 200 with the same size as before, after one https redirect to a mirror. Why the tests fail instead of skip: `@EnabledWhenCDNAvailable(hostname = "opennlp.sourceforge.net")` only checks that `https://<host>/` answers 2xx. The host root is still up, so the tests run and then fail on the 404. Affected: - `opennlp.model.v15.base.url` in the root `pom.xml`, and the defaults in the `AbstractModelLoaderTest` classes (runtime, cli), `AbstractNameFinderTest` (tools) and `AbstractIT` (uima) - `POSTaggerMETest` (4 legacy tests), `NameFinderMEWithDatesTest`, `TokenNameFinderModelTest.testNERWithPOSModelV15`, `ChunkerModelLoaderTest`, `TokenNameFinderModelLoaderTest`, `ChunkerMEIT`, uima `SingleAnnotatorIT`, `FullAnnotatorsFlowIT` - `opennlp-extensions/opennlp-uima/createPear.xml` (plain http, same path) Proposed fix: 1. Switch the base URL to `https://downloads.sourceforge.net/project/opennlp/models-1.5/` in the pom, the four test defaults and `createPear.xml`. `base + name` still works, and `HttpURLConnection` follows the redirect. 2. Make `@EnabledWhenCDNAvailable` probe a URL instead of a host root. `https://downloads.sourceforge.net/` itself returns 404, so a hostname-only check would silently skip all of these tests. Probing the actual model path also means a dead path skips instead of failing CI. -- 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]
