Chris Friesen <chris.friesen at windriver.com> writes: > On 06/18/2014 08:35 AM, Duncan Thomas wrote: >> On 18 June 2014 15:28, Matthew Booth <mbooth at redhat.com> wrote: >>> The answer is not always more >>> review: there are other tools in the box. Imagine we spent 50% of the >>> time we spend on review writing tempest tests instead. >> >> Or we push the work off of core into the wider community and require >> 100% unit test coverage of every change *and* record the tempest >> coverage of any changed lines so that the reviewer can gauge better >> what the risks are? > > 100% coverage is not realistic. I was thinking the same, but there are actually some non-trivial projects that have 100% code coverage in their tests. These are two large projects that I know of: * http://www.pylonsproject.org/projects/pyramid/about: "Every release of Pyramid has 100% statement coverage via unit tests" * http://www.sqlite.org/testing.html: "100% branch test coverage in an as-deployed configuration" Whether 100% test coverage is worth is it another question. People sometimes confuse "100% test coverage" with "100% bug free", which is just wrong. -- Martin Geisler http://google.com/+MartinGeisler -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 818 bytes Desc: not available URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20140618/7e80d87c/attachment.pgp>