Duplicated `META-INF/MANIFEST.MF` entries when using `packageWithJVMJar := true` on Windows about sbt-osgi HOT 10 CLOSED

@He-Pin So good news, I managed to replicate this with a trivial test in the Windows CI

from sbt-osgi.

I think this is windows related? Unfortunately I don't have a Windows machine on me but willing to accept any PR that solves the issue.

Also we do run against Windows in our CI so its interesting we are not getting problems there.

from sbt-osgi.

Also we do run against Windows in our CI so its interesting we are not getting problems there.

Maybe the CI doesn't run any benchmarks?

from sbt-osgi.

Maybe the CI doesn't run any benchmarks?

Well benchmarks aren't really relevant here. There may be an interesting interaction between sbt-jmh and sbt-osgi in which case the best way to replicated this would be write a scripted test that adds sbt-jmh as a plugin to try and reproduce it.

The best way to diagnose this would be to minimize the issue as a sbt-scripted test within this project.

An interesting thing to note specifically wrt Pekko, sbt-osgi has been updated some weeks ago where sbt-jmh was updated recently so maybe thats the root cause? Would be good to bisect which at which exact Pekko commit the issues are being caused.

from sbt-osgi.

seems it's ok on mac, but fails on Windows 11.

from sbt-osgi.

seems it's ok on mac, but fails on Windows 11.

Also fine on Linux otherwise CI would be breaking. Try the find the commit which causes the breakage.

from sbt-osgi.

See apache/pekko#986 (comment) for likely cause

from sbt-osgi.

@He-Pin Also what JDK are you using, is it 8, 11, 17 or 21?

from sbt-osgi.

@mdedetrich tested both java 8 and java 11, both fails.

from sbt-osgi.

@mdedetrich tested both java 8 and java 11, both fails.

Thanks, created a PR at #115 to create a basic test with packageWithJVMJar := true to see if I can replicate it on Windows CI.

from sbt-osgi.