#SNMP Design: Breaking Changes Coming, Part III

(This post is updated on September 27th heavily.)

If you notice, both SNMP data type OCTET STRING and Opaque can be viewed as byte array type in .NET. Maybe you wonder why OCTET STRING is not viewed as System.String, the reason is simple. System.String does have one more property than a byte array, Encoding. It seems when SNMP was designed, it does not provide a default encoding for OCTET STRING explicitly (and sometimes it is even used just like a byte array when you store MAC address bytes in it).

Therefore, a problem has been there for a long time. Which encoding should be used there if an OctetString instance comes. In the old Change Sets, the answer is ASCII and there is no way in them to change it to Unicode or others.

It is in the latest few builds, I start to provide a way so that you can explicitly state which encoding you prefer.

As a reference at first, there are new methods and properties in OctetString class you should pay attention to,

1. public OctetString(string str, Encoding encoding). You may provide a specific Encoding so that the internal bytes are what you expect.
2. public string ToString(Encoding encoding). According to Encoding provided, the internal bytes are decoded to a correct string.
3. public Encoding Encoding. This is a read only property that illustrates what encoding is assigned to this OctetString instance.
4. public static Encoding DefaultEncoding. This is a property to define default encoding used for all OctetString instance if no encoding is assigned via the constructor described in item 1. The default value of this property is Encoding.ASCII.
5. public OctetString(string str). OctetString.DefaultEncoding is used to generate internal bytes from str.
6. public string ToString(). OctetString.Encoding is used to generate a string from internal bytes.

Now let’s review a few common scenarios and discuss what you may do to customize OctetString’s behaviors.

Default ASCII Way

If you don’t touch the properties and methods related to Encoding, then you are in a pure ASCII way. It should behave the same as old releases of #SNMP (0.5, 1.0 and 1.1).

Default Encoding.* Way

If you change DefaultEncoding to any other Encoding types, and don’t touch the properties and methods related to Encoding later, you are in a pure Encoding.* way.

Basic Hybrid Way

If some of your devices require ASCII while others require Unicode (aka UTF-16), then you may find it a little bit difficult. My suggestions are,

Whenever you need to construct an OctetString object, specify its encoding explicitly. Whenever you need to use an OctetString object, check its Encoding before calling ToString(). If the Encoding is not correct, call ToString(Encoding) explicitly.

Last Word

I understand the current implementation is not perfect yet. When #SNMP parses incoming packets, it always use DefaultEncoding to construct OctetString objects from raw bytes. This is not optimal for hybrid cases. Hope a better way can be found soon.

No matter what, please report any critical issues or your suggestions in this field.