----- Original Message ----- From: "Andrew Goodnough" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]>; <[EMAIL PROTECTED]> Sent: Thursday, March 13, 2003 08:54 Subject: RE: [PATCH] Ant Task - Proposed Enhancement
> This new ant task has no support for running several targets, forking, > child->parent props,refs (except the default handling of props/ref > provided by the original ant). This was outside of my scope. I don't > see why these features couldn't be added though. <java> and <javac> > have a fork attribute - why couldn't <ant>? Based on *your* reasoning > we should have created a <subjavac> if <javac> wasn't meeting our > needs. Forking is something I'd like to see in big ant projects, though eliminating memory leaks is more important. the hard part about forking ant is that we want a tight channel between the two so that log messages and build exceptions trickle up; the subprocess should be fairly tightly coupled. In pure unix I'd do that with a fork(), a pipe and some structs, but java is fiddler. RMI might be better -add a special param to ant to give it the hostname:port of an RMI receiver that status should go to. Maybe that could all be done with a special listener. anyway, that is all in the post 1.6 timeframe. I'd be nice to do remote builds as well as local in this framework.
