[ 
https://issues.apache.org/jira/browse/FLINK-3444?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15227983#comment-15227983
 ] 

ASF GitHub Bot commented on FLINK-3444:
---------------------------------------

Github user zentol commented on a diff in the pull request:

    https://github.com/apache/flink/pull/1857#discussion_r58673287
  
    --- Diff: 
flink-java/src/main/java/org/apache/flink/api/java/ExecutionEnvironment.java ---
    @@ -780,6 +780,33 @@ public CsvReader readCsvFile(String filePath) {
                return fromCollection(Arrays.asList(data), 
TypeExtractor.getForObject(data[0]), Utils.getCallLocationName());
        }
        
    +   /**
    +    * Creates a new data set that contains the given elements. The 
elements must all be of the same type,
    +    * for example, all of the {@link String} or {@link Integer}. The 
sequence of elements must not be empty.
    +    * <p>
    +    * The framework will try and determine the exact type from the 
collection elements.
    +    * In case of generic elements, it may be necessary to manually supply 
the type information
    +    * via {@link #fromCollection(Collection, TypeInformation)}.
    +    * <p>
    +    * Note that this operation will result in a non-parallel data source, 
i.e. a data source with
    +    * a parallelism of one.
    +    *
    +    * @param type The base class type for every element in the collection.
    +    * @param data The elements to make up the data set.
    +    * @return A DataSet representing the given list of elements.
    +    */
    +   @SafeVarargs
    +   public final <X> DataSource<X> fromElements(Class<X> type, X... data) {
    +           if (data == null) {
    +                   throw new IllegalArgumentException("The data must not 
be null.");
    +           }
    +           if (data.length == 0) {
    +                   throw new IllegalArgumentException("The number of 
elements must not be zero.");
    +           }
    +           
    +           return fromCollection(Arrays.asList(data), 
TypeExtractor.getForClass(type), Utils.getCallLocationName());
    --- End diff --
    
    why does the scala variant throw a specific exception on getForClass() 
failure but this one doesn't?


> env.fromElements relies on the first input element for determining the 
> DataSet/DataStream type
> ----------------------------------------------------------------------------------------------
>
>                 Key: FLINK-3444
>                 URL: https://issues.apache.org/jira/browse/FLINK-3444
>             Project: Flink
>          Issue Type: Bug
>          Components: DataSet API, DataStream API
>    Affects Versions: 0.10.0, 1.0.0
>            Reporter: Vasia Kalavri
>
> The {{fromElements}} method of the {{ExecutionEnvironment}} and 
> {{StreamExecutionEnvironment}} determines the DataSet/DataStream type by 
> extracting the type of the first input element.
> This is problematic if the first element is a subtype of another element in 
> the collection.
> For example, the following
> {code}
> DataStream<Event> input = env.fromElements(new Event(1, "a"), new SubEvent(2, 
> "b"));
> {code}
> succeeds, while the following
> {code}
> DataStream<Event> input = env.fromElements(new SubEvent(1, "a"), new Event(2, 
> "b"));
> {code}
> fails with "java.lang.IllegalArgumentException: The elements in the 
> collection are not all subclasses of SubEvent".



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

Reply via email to