Displaying #maven-dev/2017-08-13.log:

Sun Aug 13 08:40:45 2017  khmarbaise:Joined the channel
Sun Aug 13 10:26:13 2017  khmarbaise:Joined the channel
Sun Aug 13 10:44:56 2017  rfscholte:Joined the channel
Sun Aug 13 12:12:35 2017  khmarbaise_:Joined the channel
Sun Aug 13 12:12:35 2017  khmarbaise_:Joined the channel
Sun Aug 13 12:25:41 2017  olamy:Joined the channel
Sun Aug 13 18:24:10 2017  khmarbaise:rfscholte: Ping..just a question…during the testing with the new version of maven-compiler-plugin which contains the fix for pelxus-language I have realized that during the build instances of JShell are running ?
Sun Aug 13 18:26:42 2017  rfscholte:those tests should only run with Java9, in which case JShell should be available
Sun Aug 13 18:27:27 2017  khmarbaise:Not the test..I’m trying to use maven-compiler-plugin within a real project which used JDK 9…
Sun Aug 13 18:27:53 2017  rfscholte:ok
Sun Aug 13 18:28:40 2017  rfscholte:if you use toolchains, then that JavaHome will be used
Sun Aug 13 18:28:56 2017  rfscholte:if you run unforked, reflection is used
Sun Aug 13 18:30:16 2017  khmarbaise:Ok..but why is jshell started…furthermore I see a real markable performance issue…from 7-10 seconds down to ca. 30 seconds for building this project…but let me check something...
Sun Aug 13 18:30:16 2017  rfscholte:only case I haven't catched is when setting the executable
Sun Aug 13 18:30:54 2017  rfscholte:JShell is indeed slow, but that's the most reliable way to get the modulename
Sun Aug 13 18:32:22 2017  rfscholte:Ok, I can think of an improvement there: right now it is a jshell call per jar
Sun Aug 13 18:33:10 2017  rfscholte:I guess we need to bundle it, so it will be just one call
Sun Aug 13 18:34:33 2017  khmarbaise:Ok tested before the change ca. 8 seconds…now 30 seconds…Ok only related to Toolchains the slowdown…plain JDK9 ca. 8 seconds as before…
Sun Aug 13 22:07:25 2017  khmarbaise:Joined the channel

Comments