I ran my tests on Windows NT 4.0. I guess I should have stated that in the test case some where.
I just tested on Solaris 7 and observed the same behavior you did. The unix-jdks must be using SIGKILL for Process.destroy()? Well, on Windows NT, the subprocesses shutdown hooks are run. So, there is an inconsistency in the way the JDKs operate. Great.... I guess there isn't any harm in leaving it in, or should we remove it all together since it is inconsistent? Sorry for the trouble, Michael -----Original Message----- From: Stefan Bodewig [mailto:[EMAIL PROTECTED] Sent: Tuesday, December 11, 2001 9:03 AM To: [EMAIL PROTECTED] Subject: Re: [PATCH] Execute.java now registers forked processes so they are d estroyed when VM exits (i.e. ctrl-C) Hi Michael, I've committed your patch, please take a look, thanks. On Wed, 21 Nov 2001, <[EMAIL PROTECTED]> wrote: > ProcessDestroyer.java (new file, currently in the > org.apache.tools.ant.taskdefs package) I have changed the class so that it doesn't use JDK 1.2 collection classes, that way it compiles on JDK 1.1 as well (and does nothing, of course). > When you run with the patch, and hit ctrl-C, you get the following > output (shutdown hook runs): I can verify, that the shutdown hook in ProcessDestroyer gets called, but I never get > [java] shutting down TestProcess on my RedHat 7.2 box. I guess Process.destroy() is doing something nasty enough on Linux that the spawned JVM doesn't get a chance to run its exit handlers. The testcase can be run from Ant's build file via the interactive-tests target. Stefan -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]> -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
