So I just tested "--dry-run" and I see it does load the class. My apologies. I was following the commit trail but somehow the loading of the class escaped me. I swore at one point it wasn't loading, but my error, nevertheless.
Was there any debate on this issue? The problem I currently see with loading the class is you're still allowing static initializers to run. I don't see a purpose in loading the class here because you're potentially allowing user code to execute and causing side-effects in the JVM. If the purpose of JDK-8159596 is to test the resolving of the configuration/options, I don't see how the current behavior is therefore desired. To take an analogy, the Maven Release Plugin has a "dry run" feature to test an install. It actually doesn't do an install. But someone can make the argument (and I am) that Java's --dry-run is not actually dry. It's really just a "--no-main" but user code can still run. Cheers, Paul On Mon, Jul 11, 2016 at 4:39 PM, Alan Bateman <alan.bate...@oracle.com> wrote: > On 11/07/2016 22:01, Paul Benedict wrote: > > The current help of --dry-run is this: >> "create VM but do not execute main method" >> >> But I think it's pretty important to note that the class is also not even >> loaded. >> > The main class should be loaded. The original intention was that it do > everything except invoke the main method. > > -Alan >