Am 20.11.12 11:08 schrieb "Babak Vahdat" unter
<babak.vah...@swissonline.ch>:
>Hi
>
>Looking at unit-tests all over the places we've got *tons* of this API
>call, like:
>
> assertTrue("File should not have been deleted", new
>File("target/files/report.txt").getAbsoluteFile().exists());
>
>Which could simply be modified to
>
> assertTrue("File should not have been deleted", new
>File("target/files/report.txt").exists());
>
>
>The only benefit I see is that using this API you would see the absolute
>file/directory path at the stacktraces when the asserts would fail, like:
>
>
> File file = new File("target/issue/test.txt").getAbsoluteFile()
>
> assertTrue("File " + file + " should exist", file.exists());
And here one concrete example of this:
https://svn.apache.org/repos/asf/camel/trunk/camel-core/src/test/java/org/a
pache/camel/builder/xml/XsltBuilderTest.java
There're totally 8 calls to this method! Of course removing all of those
calls has no effect on the test results, as they all would still pass.
Babak
>
>Note that by the example above we instantiate 2 file handles, one of which
>we don't reference at all, which's the "new File("target/issue/test.txt")"
>object.
>
>If there's no other advantages I'm missing here I would suggest to remove
>all such these calls, as it consumes both the CPU-time as well makes I/O,
>not sure though how expensive really these (native OS) calls would be, but
>for sure they're not for free.
>
>Thoughts?
>
>Babak
>
>
>
>
