[ 
https://issues.apache.org/jira/browse/LANG-1050?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14174832#comment-14174832
 ] 

Joerg Schaible edited comment on LANG-1050 at 10/17/14 7:56 AM:
----------------------------------------------------------------

Well, our unit test works for this:
{code}
    public void testEmptyArrayCreation()
    {
        final String[] array = ArrayUtils.<String>toArray();
        assertEquals(0, array.length);
    }
    public void testIndirectEmptyArrayCreation()
    {
        final String[] array = ArrayUtilsTest.<String>toArrayPropagatingType();
        assertEquals(0, array.length);
    }
    private static <T> T[] toArrayPropagatingType(final T... items)
    {
        return ArrayUtils.toArray(items);
    }
{code}
Therefore I'd extected that it works when replaceing the Object[] version with 
(ignoring BC for now):
{code}
    public static <T> T[] nullToEmpty(final T... array)
    {
        if (isEmpty(array)) {
            return ArrayUtils.<T>toArray();
        }
        return array;
    }
{code}
However, you're right, I assumed wrongly, that the type is propagated. The unit 
test above simply works, because null is never expected as the array itself and 
we get an array with a null element for this:
{code}
return ArrayUtils.toArray(array);
{code}
Since there seems to be no proper solution, I withdraw my -1.


was (Author: joehni):
Well, our unit test works for this:
{code}
    public void testEmptyArrayCreation()
    {
        final String[] array = ArrayUtils.<String>toArray();
        assertEquals(0, array.length);
    }
    public void testIndirectEmptyArrayCreation()
    {
        final String[] array = ArrayUtilsTest.<String>toArrayPropagatingType();
        assertEquals(0, array.length);
    }
    private static <T> T[] toArrayPropagatingType(final T... items)
    {
        return ArrayUtils.toArray(items);
    }
{code}
Therefore I'd extected that it works when replaceing the Object[] version with 
(ignoring BC for now):
{code}
    public static <T> T[] nullToEmpty(final T... array)
    {
        if (isEmpty(array)) {
            return ArrayUtils.<>toArray();
        }
        return array;
    }
{code}
However, you're right, I assumed wrongly, that the type is propagated. The unit 
test above simply works, because null is never expected as the array itself and 
we get an array with a null element for this:
{code}
return ArrayUtils.toArray(array);
{code}
Since there seems to be no proper solution, I withdraw my -1.

> Change nullToEmpty methods to generics
> --------------------------------------
>
>                 Key: LANG-1050
>                 URL: https://issues.apache.org/jira/browse/LANG-1050
>             Project: Commons Lang
>          Issue Type: Improvement
>          Components: lang.*
>            Reporter: James Sawle
>             Fix For: Review Patch
>
>
> Currently there are multiple Object based methods which could be replaced by 
> a single generic method.
> - public static Long[] nullToEmpty(final Long[] array)
> - public static Integer[] nullToEmpty(final Integer[] array)
> - public static Short[] nullToEmpty(final Short[] array)
> - public static Character[] nullToEmpty(final Character[] array)
> - public static Byte[] nullToEmpty(final Byte[] array)
> - public static Double[] nullToEmpty(final Double[] array)
> - public static Float[] nullToEmpty(final Float[] array)
> - public static Boolean[] nullToEmpty(final Boolean[] array)
> Recommendation, replace all of these with a single method that would also 
> allow a defensive programming style when not using wrapped primitives.
> - public static <T> T[] nullToEmpty(final T[] array)



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to