[ofiwg] Travis CI testing of libfabric and fabtests

[Hefty, Sean]

> > I've also enabled branch protection to the upstream trees.  Now that
> Travis CI is working, I'd like to propose that we require status checks to
> pass before pull requests can be merged.  The only drawback that I see to
> this is that there will be a small delay (~5-10 minutes) before simple
> pull requests can be merged.
> 
> Can you publish your Travis CI setup so that others can use it?

The travis scripts are in the repos.  See .travis.yml in the root of each tree.  The scripts automatically run from github on some virtualized linux system. 

> Optimally, it would be great if there could be CI testing for every
> provider.

I agree -- we should be able to get them to build.  I can build test all but the mxm provider at the moment.  Running them requires the hardware. 

> We have a similar setup with Jenkins inside Cisco, but have not found the
> cycles to finish the setup.  The complication is that our Jenkins is
> behind the Cisco lab environment firewall.  This created two problems:
> 
> - Github can't ping our Jenkins when a new PR comes in; our Jenkins has to
> poll Github for changes
> - The results of a Jenkins test are available via a local web server,
> which is obviously not available from the greater internet.  We half-
> finished a solution of publishing our results to gists (e.g., so that when
> a test fails, someone outside of Cisco can see the full stdout/stderr from
> the test to know *what* failed -- e.g., see https://gist.github.com/cisco-
> usnic-bot/2291186ea50354528bef)

I think Intel has a similar environment.  Maybe we can find a common solution here that works for testing over real hardware.

- Sean