Support for FreeBSD > 6 is missing from iputil.py

[francois]

On tahoe-dev, Randy Bush reported the following error when running a Tahoe node under FreeBSD 8.0.

2010-06-25 06:43:15+0000 [-] Unhandled Error
        Traceback (most recent call last):
          File "/usr/local/src/net/allmydata-tahoe-1.7.0/support/lib/python2.5/site-packages/foolscap-0.5.1-py2.5.egg/foolscap/eventual.py", line 26, in _turn
            cb(*args, **kwargs)
          File "/usr/local/src/net/allmydata-tahoe-1.7.0/src/allmydata/node.py", line 259, in _startService
            d.addCallback(lambda res: iputil.get_local_addresses_async())
          File "/usr/local/src/net/allmydata-tahoe-1.7.0/Twisted-10.0.0-py2.5-freebsd-8.0-STABLE-amd64.egg/twisted/internet/defer.py", line 224, in addCallback
            callbackKeywords=kw)
          File "/usr/local/src/net/allmydata-tahoe-1.7.0/Twisted-10.0.0-py2.5-freebsd-8.0-STABLE-amd64.egg/twisted/internet/defer.py", line 213, in addCallbacks
            self._runCallbacks()
        --- <exception caught here> ---
          File "/usr/local/src/net/allmydata-tahoe-1.7.0/Twisted-10.0.0-py2.5-freebsd-8.0-STABLE-amd64.egg/twisted/internet/defer.py", line 371, in _runCallbacks
            self.result = callback(self.result, *args, **kw)
          File "/usr/local/src/net/allmydata-tahoe-1.7.0/src/allmydata/node.py", line 259, in <lambda>
            d.addCallback(lambda res: iputil.get_local_addresses_async())
          File "/usr/local/src/net/allmydata-tahoe-1.7.0/src/allmydata/util/iputil.py", line 94, in get_local_addresses_async
            d = _find_addresses_via_config()
          File "/usr/local/src/net/allmydata-tahoe-1.7.0/src/allmydata/util/iputil.py", line 242, in _find_addresses_via_config
            raise UnsupportedPlatformError(sys.platform)
        allmydata.util.iputil.UnsupportedPlatformError: freebsd8     

[francois]

I don't see any easy way to implement meaningful tests for this change. It's probably better to have a new buildbot running a recent FreeBSD version.

What do you think?

[zooko]

I would be willing to apply this based on Randy Bush's FreeBSD 8 buildbot.

It would be even cooler to do a "mock out the underlying platform" hack like the way you, François, wrote portable unit tests of your unicode work. The unit tests could make sys.platform say "FreeBSD10" or something and then when iputil.py calls getProcessOutput() the test could patch out getProcessOutput() to return a sensible thing that looks like the output of "ifconfig -a" on FreeBSD 10.

[francois]

Replying to zooko:

It would be even cooler to do a "mock out the underlying platform" hack like the way you, François, wrote portable unit tests of your unicode work. The unit tests could make sys.platform say "FreeBSD10" or something and then when iputil.py calls getProcessOutput() the test could patch out getProcessOutput() to return a sensible thing that looks like the output of "ifconfig -a" on FreeBSD 10.

Zooko, I knew you'd be saying that ;)

I'll have a look into that.

[davidsarah]

This seems like a very safe patch that should be applied for 1.7.1, regardless of the absence of tests.

[zooko]

Oh all right. For the record, I still want a deterministic mockery-based unit test and I still want Randy's FreeBSD buildslave to go green:

​http://tahoe-lafs.org/buildbot/waterfall?show=Randy%20FreeBSD-amd64

But I guess this patch is extremely unlikely to cause a regression. :-)

applied in 968edfda119cb29c. Thanks!