2010/11/9 Nicolas Lalevée <nicolas.lale...@hibnet.org>: > That's what I thought, this proposed task being quite trivial and having no > side effect. > Obviously for larger patch or behavior change I would come first to the ML, > like I did for the project helpers for instance.
Fair enough. A follow up email to the ML is good though, to explain rational etc... before the "commit watch patrol" has to ask for it :) >>> the use case, or more precisely why the use case you describe can't be >>> achieved some other way. > > It can definitively be handled without that task. With Ant 1.8.1, to bind the > targets "jar" and "source" to an extension point "dist" is to create a third > target: > <target name="bind-to-dist" depends="jar,source" extensionOf="dist" /> > > I find it cleaner to avoid creating yet another target and implement this > simple bindtargets task. > If there are objection, I'll remove it. Use that work around for classical > build files. And put this task in EasyAnt from which I got the idea. Not being quite up-to-speed on extension-point, I wasn't sure, thanks. The reason I'm a little reluctant on <bindtargets> is that it's a task that affects the dependency graph of targets, but bypassing the normal means to do that, via <target>. Since it's a task, it can be run at any time, conditionally or not, inside a target or not, and especially after the dependency graph was computed, when it does/can change the dependency graph. Maybe that's OK, but it just make me a little uncomfortable and I'm not sure we see all the possible ramifications. It doesn't mean I'm necessarily against it, but if it's only notational convenience, and the alternative is hardly longer or "uglier", I'm not sure it's worth it. My $0.02. --DD --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@ant.apache.org For additional commands, e-mail: dev-h...@ant.apache.org
