On Thu, 10 Mar 2022 10:28:30 GMT, Julian Waters <jwat...@openjdk.org> wrote:
>> That might be. If that is so, then you've most likely spotted a bug. In >> practice, you're either building an official build and have a build number, >> or an adhoc build with an opt string, so I'm not suprised noone has run into >> this. > > Both version build and version opt have valid --without options though, if > both are set (As in the case of an adhoc build run with > --without-version-opt) that could be problematic, since the issue wouldn't be > stemming from a bug in that case. I'll double check just to be sure Ok, now I've tried to dig through Apple's somewhat confusing documentation on CFBundleVersion and CFBundleShortVersionString. My takeaway is that we should either publish the build number there, or 0 if we have no build number. This is basically what is done by the code, and that was *exactly* what was done by the code, prior to the fix were we allowed an empty build number (JDK-8274980). So this code should also have been updated then. I've opened JDK-8282948 to fix this. But still the effect for this PR is that a valueless --with-macosx-bundle-build-version is not accepted. ------------- PR: https://git.openjdk.java.net/jdk/pull/7713
