> <target name="init"/> > <target name="t1" depends="init"/> > <target name="t2" depends="t1"/>
versus > <target name="init"/> > <target name="t1" depends="init"/> > <target name="t2" depends="init,t1"/> Is there actually any difference in the way ant will process these dependencies? They appear to be identical, but the latter case lists explicitly that it depends on init (rather than relying on implicit dependencies), which implies "init" is on the dependency list twice. I know that Ant will only execute init once, though; I'm just curious if there any implicitations (performance, anything else???) that affect one approach more than the other. I've had several cases where I've wavered back and forth between these two approaches. In general, I try to end up with whatever "feels" like the real dependencies, even if it means repeating dependencies as in the latter case. For example, if the t2/t1 dependency is something I could envision perhaps changing someday (perhaps "t2" is "tests" and "t1" is "compile" - someday I might want to be able re-run a test without recompilation), then I'll treat it like the 2nd case. But if t2 requires t1 which required init, then I'll lean towards the former case. One minor annoyance with the latter, though: I use an ant2dot.xsl that was posted on this list back in June(many thanks Stefan!! (and Erik for the build targets to automate it!!)). The latter approach really clutters up the graph. I end up with two arrows out of t2 and two into init. But that's an issue with the stylesheet (and I'm still learning XSL syntax to see if I can clean that up....) THanks for any thoughts, Paul Christmann -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
