tballison commented on PR #3260: URL: https://github.com/apache/tika/pull/3260#issuecomment-5878104962
From my :robot: ``` Thanks, the direction looks right, and the new BundleIT checks are great. A few things before merge: 1. **Make the imports of the un-embedded bundles mandatory.** They currently come in through `*;resolution:=optional`, so an OSGi deployment that misses one of the new bundles still resolves and then fails with `NoClassDefFoundError` mid-parse. With explicit, non-optional imports, Felix refuses to resolve and names the missing package. 2. **Floor the versions at what Tika builds with.** bnd's defaults allow older patch releases than we test (`org.apache.pdfbox` `[3.0,4)`, `org.apache.commons.compress` `[1.28,2)`), and `org.apache.commons.io` comes out as `[1.4,2)` (from mime4j's import). That range is only satisfied by commons-io 2.22 through its `1.4.9999` compat export, and a real 1.4 bundle would satisfy it too. The attached patch does both, using the parent-pom version properties so dependabot bumps carry through. BouncyCastle stays literal because its packages export major.minor only. `com.adobe.internal.xmp.impl` and `org.apache.pdfbox.debugger` stay optional since no deployed bundle exports them. With it, BundleIT passes, and removing `commons-compress.jar` from `target/test-bundles` makes the bundle fail resolution with `missing requirement ... osgi.wiring.package`. 3. **CHANGES:** main now has a 4.2.0 section; please add an entry listing the bundles OSGi users must now install. 4. **PDF test:** the `DefaultParser` recursion you found is fixed in TIKA-4942 (the detector recursed the same way). Once that's merged, could `testPdfParsing` go through the registered `Parser` service instead of `PDFParser` directly? Minor: the description says commons-io 2.x doesn't satisfy mime4j's `[1.4,2)`. It does, via the compat export mentioned above. Keeping mime4j embedded is fine either way. ``` -- 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]
