Default encapsulation of strings fields into unions

[Youssef Hatem <yo...@rwth-aachen.de>]

Hi,

Assume I have the following schema:

{
  "name": "Foo",
  "type":"record",
  "fields":[ {"name": "Bar", "type":"string"} ]
}

Since a string can be null, will Avro (Java implementation) encapsulate Bar in a union? (i.e. type of Bar field will be regarded as ["null", "string"]).

I am almost certain that it should not because it will lead to inconsistencies with encoding/decoding but I want to make sure.

Another question, what happens when we want to deserialize a null string?

Best regards,
Youssef Hatem

Re: Default encapsulation of strings fields into unions

[Doug Cutting <cu...@apache.org>]

On Fri, Oct 11, 2013 at 9:24 AM, Youssef Hatem
<yo...@rwth-aachen.de> wrote:
> Since a string can be null, will Avro (Java implementation) encapsulate Bar in a union? (i.e. type of Bar field will be regarded as ["null", "string"]).

If you want to intermix nulls and strings then you need to explicitly
use a union.  If null is passed as a string then a
NullPointerException is thrown.

> Another question, what happens when we want to deserialize a null string?

A serialized string is never null.  It may be an empty string.

Doug