Open Stack

On 2013-06-20 15:56:22 +0400 (+0400), Dina Belova wrote:
> I'm writing to suggest that git-review requires to be covered by
> tests. There are no testing part in it at all, so maybe it is a
> good idea to cover it.
[...]

(Cc'ing you since I'm unsure whether you're subscribed, but setting
MFT to the infra list to avoid unnecessary cross-posting)

We've got an open bug on that topic I've been wanting to address for
a while but haven't been able to prioritize it sufficiently since
git-review is a fairly low-volume project as contributions go:

    https://launchpad.net/bugs/1048724

In short, right now I manually perform integration tests of every
proposed change using a very hackish shell script that interacts
with a Gerrit sandbox in the ways we're concerned it might break
under Python 2.6, 2.7, 3.2 and 3.3. This is of course very time
consuming, so the current thought is that we should have an
expect-like dummy shell underneath git-review which expects it to
attempt to run particular commands and then returns the correct
responses. This could be initially constructed by inserting a simple
shim to record all shell I/O during one of my integration tests, so
that it could be played back through the dummy shell later in
automated tests.
-- 
Jeremy Stanley