David Aldrich wrote:
> Consider a 'send' method that sends a message to another system via a
> socket. This method will wait for a response before returning. There are
> two possible error conditions:
>
>
> 1) Timeout - i.e. no response received
>
> 2) Illegal response received
>
> I need to communicate these errors to the caller of send(). So far I have
> just raised a RuntimeError exception for both errors, and stated what
> happened like this:
>
> raise RuntimeError("Message timeout")
>
> That's fine if the caller just wants to print the error but not so good if
> the code needs to act differently according to which error condition
> occurred.
>
> So, my question is, what's the pythonic way of doing this? Should I
> subclass RuntimeError for each possible error condition? E.g.:
>
> class MessageTimeoutError(RuntimeError): pass
> class IllegalResponseError(RuntimeError): pass
If you don't want to let the original timeout error bubble up you can create
your own little hierarchy of exceptions:
class ResponseError(Exception):
pass
class TimeoutError(ResponseError):
pass
class BadDataError(ResponseError):
pass
Then the baseclass of ResponseError doesn't matter much as client code that
wants to catch every expected error can catch ResponseError. You can later
add subclasses as needed without breaking this catch-all client.
_______________________________________________
Tutor maillist - Tutor@python.org
To unsubscribe or change subscription options:
https://mail.python.org/mailman/listinfo/tutor
