Use testr instead of nose.

nose is invasive and can sometimes alter the outcome of a test run. testr,
on the other hand, keeps a distinction between running tests and displaying
results of the test runs. Additionally, it supports the stock python unittest
protocol.

Even better, testr supports parallel test running, which makes things faster,
and a command "testr run --failing" which will just re-run the latest failing
tests (often something one wants to do in iterative dev)

Part of blueprint grizzly-testtools

Change-Id: I0b3f1bcb5d4ff59c65eb3219b30a9e64f54d70bd

1 files changed, 4 insertions, 3 deletions

diff --git a/.unittests b/.unittests
index b977da4..4758296 100755
--- a/.unittests
+++ b/.unittests
@@ -1,7 +1,8 @@
 #!/bin/bash
 set -e
 
-nosetests --exe --with-coverage --cover-package swiftclient --cover-erase $@
-rvalue=$?
+python setup.py testr --coverage
+RET=$?
+coverage report
 rm -f .coverage
-exit $rvalue
+exit $RET