On Fri, 29 Aug 2025 04:42:24 GMT, Phil Race <[email protected]> wrote:
>> Yes, all 3 points are correct. Probably the native URL handler is installed >> during the startup of the bundled application in order to catch the first >> OpenURL event that might have launched the application in the first place >> (I'll conduct more tests to see how an early handler registration affects >> the startup behavior). Java code technically is not required to install any >> URL handlers even in case of a bundled application, however declaring the >> capability to handle OpenURL events in the application bundle requires some >> actions from the application developer. I guess it was thought of as a >> "common sense" not to declare OpenURL handling capability in the bundle >> without actually handling the URLs in the Java code. >> >> Searching the Internet so far did not show up any references to `[bundle >> _hasEAWTOverride:@"URLHandler"]` (or the respective keys in the property >> list file) except the JDK code itself. Under such circumstances I've decided >> to leave these checks intact. >> >> I'm investigating the possibility of composing a regression test for this >> change. It does not look straight impossible so far. I'll post an update on >> this matter shortly. > > @ProjectD8 any update on the regression test ? > Adding one (or not) is the only thing between this being a PR and a pushed > fix. @prrace I've come up with some ideas on encapsulating the required testing actions into a JTReg test. I've tested these concepts separately, and the results look promising. When putting everything together, however, I've encountered some difficulties with the programmatically generated macOS bundle, which may take some more time to rectify. Please expect an update in the next few days. ------------- PR Comment: https://git.openjdk.org/jdk/pull/25967#issuecomment-3283809663
