I see Gyula’s point. In the case of the TupleTypeInfo subclass it only works because the equals method of TypleTypeInfo is used, IMHO.
Stupid implementation mistakes should be caught by the Java type checker. I don’t think it would allows passing a Map<String, String> to the map method if the type of the DataSet is Integer. The problem is that it is possible to have several TypeInformation implementation that would be a valid TypeInformation for a given type. (And the downstream operator has to use the TypeInformation from upstream, that was set using returns. Otherwise the binary format would not be compatible.) > On 08 Nov 2015, at 23:45, Timo Walther <twal...@apache.org> wrote: > > MapFunction