Creating your own exceptions is a good thing. It lets you catch something very specific without the need to inspect more generic exceptions for the right messages every time.

Catching All Exceptions

Catching all exceptions is bad. Why? Have a look at the following:


#!/usr/bin/env python3

    f = open('/tmp/potatoes.txt', 'r')
    f.write('I really love potatoes!')
except Exception as exc:


#!/usr/bin/env ruby

  f ='/tmp/potatoes.txt', 'r')
  f.write('I really love potatoes!')
rescue Exception => e
  puts e.message

Given the code above, it’s nearly impossible to know why your code fails. At best, there are two reasons for the failure: the file doesn’t exist; or you’re attempting to write to a file when you’ve opened it as read-only. In reality, there are probably many more exceptions that could be generated, and you probably want to handle them all differently.

Custom Exceptions in Python

Let’s look at Python’s ncclient library. What if we want to catch unauthorized RPCs? ncclient doesn’t raise an exception for this. To see how we can catch an unauthorized command, let’s create a small miniature library.

#!/usr/bin/env python2

from ncclient import manager
from ncclient.operations import RPCError

class UnauthorizedError(Exception):

class Device(object):
    def __init__(self, host, port, username, password):
        self.device_manager = manager.connect(host=host, port=port,

    def send_config(self, xml_config):
            return self.device_manager.edit_config(target='running',
        except RPCError as error:
            if error.tag == 'access-denied':
                raise UnauthorizedError('User is not authorized for this RPC.')

Now if a user uses our library, they’ll get the appropriate error for trying to configure something for which they are not authorized. After the if statement, we re-raise the exception with a bare raise so that we don’t skip over something other than an unauthorized command.

You could catch many other types of exceptions at the same time, but I’ve ommitted them for brevity.

Custom Exceptions in Ruby

Now let’s look at custom exceptions in Ruby. We’ll do this example as a simple script instead of assuming it’s a library. Let’s assume that a user bob with the password bob exists on the switch but is not allowed to change configuration on the device. This example assumes a Brocade VDX is the target device, but the principles remain the same regardless of vendor.

#!/usr/bin/env ruby

require 'net/netconf'

class UnauthorizedError < StandardError

login = {target: '', username: 'bob', password: 'bob'} do |dev|
    dev.rpc.edit_config do |x|
      x.interface(xmlns: '') {
        x.tengigabitethernet {
          x.description "to sw37 te 201/0/1"
  rescue Netconf::EditError => e
    if e.rsp.xpath('//error-tag').first.content == 'access-denied'
      raise UnauthorizedError, 'User is not authorized for this RPC.'

If we were to execute this script, this is what we’d get:

$ ruby custom_exception.rb
custom_exception.rb:21:in `rescue in block in <main>': User is not authorized for this RPC. (UnauthorizedError)
        from harbl.rb:10:in `block in <main>'
        from /home/tyler/.rvm/gems/[email protected]/gems/net-netconf-0.4.3/lib/net/netconf/transport.rb:27:in `initialize'
        from /home/tyler/.rvm/gems/[email protected]/gems/net-netconf-0.4.3/lib/net/netconf/ssh.rb:21:in `initialize'
        from custom_exception.rb:9:in `new'
        from custom_exception.rb:9:in `<main>'


I hope you can see how useful creating your own exceptions can be. It makes it much easier to diagnose and debug when a user (inevitably) reports a bug in your software.