Base64EncoderStream.encode() produces characters outside the Base64 alphabet

[Noblebrown]

When invoking Base64EncoderStream.encode(), it will can produce a String containing characters outside of the Base64 alphabet. This should be fixed, calls replaced with Base64Encoder.encode(), or replaced entirely with a third-party library.

[praseodym]

I would suggest using a third-party library instead. Maybe Guava?

Another option would be to use Java 8, then there is java.util.Base64.

[Noblebrown]

Ews Java has two Base64 classes (Base64EncoderStream and Base64) for reasons unknown. Base64 seems to work just fine, but a third party library may do the trick as well. I haven't used Guava, but I've heard good things. Apache Commons would be a candidate as well. I don't think Java 8 is viable. It's too new and isn't widely used.

[praseodym]

Apache commons-codec is already used in Apache HttpClient and has base64 methods, so we could use them if we keep it as an explicit dependency (#35).

[serious6]

I looked at the places where the code needs to be changed. Here might be a problem because the api is just casting object to byte[]. If the object is no byte[] this will fail. For dealing with this I tried smth like this:

    @Override
    protected String toString(Object value) {
        if (value instanceof byte[]) {
            return new String(Base64.encodeBase64((byte[]) value));
        }else{
            ByteArrayOutputStream out = new ByteArrayOutputStream();
            ObjectOutputStream os = new ObjectOutputStream(out);
            os.writeObject(value);

            return new String(Base64.encodeBase64(out.toByteArray()));
        }
    }

but I dont know in which way we should deal with the IOException. Another bug might be converting the byte[] to a String with no Charset beeing set. Or am I wrong? Even though this wasn´t integrated in the encoding-process before.
example makes use of apache.commons.codec.Base64