Impove coverage job accuracy

by informing it that we use eventlet for concurrency.

We have many false negative coverage runs, even for changes that
do not touch the code under test, for example in doc only change [1].
According to the coverage documentation, unless we indicate that
we are using an alternate to the thread library for concurrency
we will get "very wrong results." [2]

Set this option correctly.  Also avoid false precision in the
coverage report.

[1] http://logs.openstack.org/01/514901/2/check/openstack-tox-cover/6df6369/

[2] https://coverage.readthedocs.io/en/coverage-4.4.1/config.html

TrivialFix

Change-Id: I7398a45cdab8aa4191a55cbb86b9e6499b09f021
This commit is contained in:
Tom Barron 2017-10-25 11:51:46 -04:00
parent c86e7658af
commit c22c073734

View File

@ -2,6 +2,7 @@
branch = True
source = manila
omit = manila/test*
concurrency = eventlet
[report]
precision = 2
precision = 0