[Python-Dev] [Python-checkins] cpython: Issue 14814: The new systematic tests aren't just about error reporting any

[Nick Coghlan]

On Mon, Jul 9, 2012 at 3:22 AM, Terry Reedy <tjreedy at udel.edu> wrote:
> On 7/8/2012 9:14 AM, nick.coghlan wrote:
> If you want to report both errors, when present:

I don't really - when there are multiple things wrong with the
address, ipaddress just reports the first one it notices (e.g.
IPv4Address("google.com") will complain about the lack of "."
separators rather than the fact that "google" and "com" are not valid
octets).

This was just a case where I started to reorder the checks to do the
faster one first, and then realised it was more user friendly to check
the other way around (since the success case will always include both
checks, it only makes a difference to the speed of rejection of some
invalid cases). Most of the other validation checks have a more
obvious suitable order (for example, there's no reason to start
checking individual parts if we can't even split the parts up
appropriately).

>> +class BaseTestCase(unittest.TestCase):
>>       # One big change in ipaddress over the original ipaddr module is
>>       # error reporting that tries to assume users *don't know the rules*
>>       # for what constitutes an RFC compliant IP address
>
>
> Good assumption. There *will* be people who learn by trial and error.

*cough* It's conceivable I started down this path due to using the
REPL for exploration while making an initial attempt at writing method
and data attribute docs over the weekend. I'm firmly in the category
of IPv4 network programming engineers with next to no IPv6 experience
:)

Cheers,
Nick.

-- 
Nick Coghlan   |   ncoghlan at gmail.com   |   Brisbane, Australia