Rewrite binary operations implementation

[kshchepanovskyi]

Previously binary data was stored as new String(byteArray) but it is not correct - as stated in String javadoc, it can replace unmappable characters.

As a result, when you save serialized java object it is not possible to deserialize it later.

[idyedov]

does it make sense to convert everything to byte arrays? ie use that logic for String methods as well

[kshchepanovskyi]

does it make sense to convert everything to byte arrays

Generally, it does not matter - because we must call String.getBytes() on read operations.

ie use that logic for String methods as well

It is used. Each pair of string/binary method uses private method that takes DataContainer type as arguments. For example:

@Override
public synchronized Response<String> set(final String key, final String value) {
    return set(DataContainer.from(key), DataContainer.from(value));
}

@Override
public synchronized Response<String> set(final byte[] key, final byte[] value) {
    return set(DataContainer.from(key), DataContainer.from(value));
}

private Response<String> set(final DataContainer key, final DataContainer value) {
    /// ...

[idyedov]

I guess I'm just concerned that we're storing it twice - one as a String and one as a byte array

[kshchepanovskyi]

Thanks for your comments - I will make improvements soon.

I guess I'm just concerned that we're storing it twice - one as a String and one as a byte array

I don't think so. It is mocking tool, memory usage is not a priority. Also, it is useful for debugging - it is almost impossible to identify your keys/values when they are stored as byte arrays only. Anyway, it is easy to optimize.