On Thu, Jul 5, 2012 at 12:42 AM, prasanna <srivatsav.prasa...@gmail.com> wrote: > On 4 July 2012 23:49, David Nalley <da...@gnsa.us> wrote: >> On Wed, Jul 4, 2012 at 1:55 PM, Pradeep Soundararajan < >> pradeep.soundarara...@citrix.com> wrote: >> >>> Yes, I have executed the below successfully:**** >>> >>> ** ** >>> >>> ant build-all **** >>> >>> package marvin**** >>> >>> ** ** >>> >>> >>> >> Yes, I don't doubt that it builds, but does it run (and are you getting >> paramiko installed in some other way?) > > So I'm kind of confused about the licensing and legal bits here. So > let me state the problem (for myself) clearly and perhaps we can find > a solution. > > Marvin's packaging happens using ant. This packaging will only create > a distributable source tarball under tools/marvin/dist/. That tarball > does _not_ contain paramiko. The actual installation of the source > tarball happens using distutils on the client machine. When you > install this tarball as: > > $~: "pip install marvin-0.1.0-tar.gz" > > pip takes care of downloading paramiko for you. > > If this is not legally okay - then we'll have to altogether get rid of > the tests and rewrite them without paramiko. As of now I don't know of > a good enough replacement library. > > There is both a pro- and con- to this : > > 1) without paramiko we would lose the backend verification in the tests > 2) without the backend verification the tests can be run with the > simulator (or any other hypervisor) making it easier to have a smoke > suite which everyone can run. > > HTH clear some things. > > -- > Prasanna.,
So reading this again and parsing it for a bit this morning I arrive at this: Marvin is not in the default build Marvin has a system (runtime) requirement of python-paramiko I think this is ok (extra eyes welcome) Pradeep: What about the other issues I pointed out?
