ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gintautas Grigelionis <>
Subject Re: ant git commit: Add magic names for tests, run more tests in Surefire
Date Sun, 28 Oct 2018 18:10:49 GMT
On Sun, 28 Oct 2018 at 18:48, Stefan Bodewig <> wrote:

> On 2018-10-28, Gintautas Grigelionis wrote:
> > On Sun, 28 Oct 2018 at 18:17, Stefan Bodewig <> wrote:
> >> On 2018-10-28, Gintautas Grigelionis wrote:
> >>> On Sun, 28 Oct 2018 at 17:59, Stefan Bodewig <>
> wrote:
> >>>> I wonder whether it wouldn't be a good idea to create a separate class
> >>>> for constants used during testing that lives in ant-testutil rather
> than
> >>>> poluting the "magic names" of Ant.
> >>> The problem is that should such class end up in ant-testutil, it makes
> >> ant
> >>> dependent on ant-util and that creates a dependency loop.
> >> I'm talking about the "magic names" only used in tests. The four
> >> properties with names that start with TEST_ should not be used inside
> >> the rest of Ant, are they?
> > I understand your point. TEST_ properties are only for tests. But, the
> > tests themselves are a part of Ant core.
> Are they? I really hope there are no test classes in ant.jar.
> The constants can be moved to the tests jar of Ant core (where Ant core
> likely means org.apache.ant:ant in Maven speak) and ant-testutil. All
> Maven artifacts that are not org.apache.ant:ant can have a test scope
> dependency on ant-testutil.

The scope does not matter, should the test constants be in ant-util, there
will be a dependency of ant on ant-testutil creating a loop.
Test classes in Ant core won't compile before ant-testutil is compiled and
packaged, which in turn would require ant to be compiled and packaged.
It's all nice and easy when there is a pile of code that can be compiled at
once and divided arbitrarily... not so simple when it has to be split
beforehand (see my remarks on AssertionsTest depending JUnit tasks or
launcher tests depending on Os).


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