ant-ivy-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Burt Leung <>
Subject Re: Redundant JAR files in configuration groups
Date Wed, 06 Jan 2010 23:53:48 GMT
Hi Mitch,

Thank you very much for giving me a bit of your time and expertise!

I read the documentation for the ivy:cachepath operation and got it to work
for our project just now. As you mentioned, I no longer need to call
ivy:retrieve in our build too (which created duplicate JAR files in the /lib
folder of the module).


On Tue, Jan 5, 2010 at 2:44 PM, Mitch Gitman <> wrote:

> Burt, you wrote: "I personally think it would be nice if Ivy setup/provided
> ANT path elements behind the scenes based on the Ivy configuration groups."
> Ivy does offer this, just not behind the scenes. Look at the ivy:cachepath
> Ant task:
> This would obviate the need to use the ivy:retrieve task and create all
> those lib directories. To me, the use of ivy:retrieve and lib directories
> signals that you've kinda, sorta gotten into Ivy, but you haven't
> completely
> bought into or embraced Ivy yet. You've created a redundant set of
> directories to make your file setup look more like the pre-Ivy days.
> Also, about the whole matter of confs extending confs, I know this is very
> subjective and not right-or-wrong, but to me the whole idea of having your
> test Ivy conf extend your main Ivy conf is a case of meshing two things
> together too early. What if you want to see just the dependencies that
> apply
> to test and not main? You're stuck having to sift through maybe 30 JARs to
> see just three.
> Instead have your test Ivy conf pick up test-only dependencies. In other
> words, don't let your test conf extend your main conf. When you need to
> combine the JARs from both confs in a path, just do it as-needed using
> composition. So using the old-fashioned ivy:retrieve and lib dirs, you'd
> have:
>    <path id="classpath.tests.compile">
>        <fileset dir="${compileSrcLibDir}">
>           <include name="**/*.jar" />
>       </fileset>
>        <fileset dir="${compileTestsLibDir}">
>           <include name="**/*.jar" />
>       </fileset>
>   </path>
> Or to use the more direct ivy:cachepath where ivy.src.classpath and
> ivy.test.classpath are the paths you've created therein:
>    <path id="classpath.tests.compile">
>       <path refid="ivy.src.classpath" />
>      <path refid="ivy.test.classpath" />
>   </path>

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message