I guess i say pip is required because you use pip to install virtualenv.....

On Tue, Aug 30, 2016 at 6:12 PM, David Brown <[email protected]> wrote:

> Aha, that is in the virtualenv too...
>
> On Tue, Aug 30, 2016 at 6:06 PM, David Brown <[email protected]> wrote:
> > Ah wait, I guess you use it to install twine.
> >
> > On Tue, Aug 30, 2016 at 5:58 PM, David Brown <[email protected]>
> wrote:
> >> Nice work with this Stephen! One thing: if I am thinking about this
> >> correctly, pip isn't a requirement for building. I think we are only
> >> using pip inside the virtualenv, and since the virtualenv binary
> >> installs pip and setuptools by default when in creates the
> >> environment, I don't think we need it to be installed on the system.
> >> So basically, if you have Python(2) on on your machine, you only need
> >> to make sure that you install virtualenv before building...make sense?
> >> or am I missing something?
> >>
> >> On Tue, Aug 30, 2016 at 5:32 PM, Stephen Mallette <[email protected]>
> wrote:
> >>> With some help from Dave I've tweaked up the build for python some
> more -
> >>> the command to build is still:
> >>>
> >>> mvn clean install -pl gremlin-python -DglvPython
> >>>
> >>> but now:
> >>>
> >>> + The prereqs for building are just python 2.7, pip and virtualenv
> >>> + The build uses virtualenv to isolate the python environment to the
> >>> /target directories
> >>> + -DskipTests is now better respected by Gremlin Server - won't bother
> to
> >>> start it, if tests are skipped
> >>> + Minor adjustments to the deploy phase for python. We can't really
> deploy
> >>> to pypi on deploy in our release process because pypi has no staging
> >>> environment (like sonatype does) so the release would be immediately
> public
> >>> prior to vote on the release. So i made a special "-Dpypi" that
> activates a
> >>> profile that will hook into "mvn deploy" and push the artifacts to pypi
> >>> with twine.
> >>>
> >>> Note that these changes are all in the "virtualenv" branch at the
> moment. I
> >>> suspect I will merge it to master tomorrow after some more tests.
> >>>
> >>>
> >>>
> >>> On Mon, Aug 29, 2016 at 5:45 PM, Dylan Millikin <
> [email protected]>
> >>> wrote:
> >>>
> >>>> Yeah this is great. Will be useful for all GLVs.
> >>>>
> >>>> On Mon, Aug 29, 2016 at 6:13 PM, David Brown <[email protected]>
> wrote:
> >>>>
> >>>> > Wow Stephen thanks for all your hard work! This will really make
> >>>> > driver development a lot easier.
> >>>> >
> >>>> > On Sun, Aug 28, 2016 at 6:48 PM, Stephen Mallette <
> [email protected]>
> >>>> > wrote:
> >>>> > > Took me half of my Sunday, but I just got Gremlin Server to start
> up
> >>>> and
> >>>> > > shutdown in the standard maven integration-test phase of
> >>>> gremlin-python.
> >>>> > > That means we can now write native python tests that round-trip to
> >>>> > Gremlin
> >>>> > > Server in addition to the regular unit tests with pytest we
> already
> >>>> > have! I
> >>>> > > actual  Note that I actually start two gremlin servers one with
> auth on
> >>>> > and
> >>>> > > one without (ports 8183/8182 respectively) so that we can test
> >>>> > > authentication features. Further note that native tests are bound
> to
> >>>> the
> >>>> > > integration-test phase of maven and run automatically on
> -glvPython.
> >>>> > that's
> >>>> > > a bit different from our other projects where integration test
> phases
> >>>> > only
> >>>> > > run if you -DskipIntegrationTests=false. gremlin-python won't
> obey that
> >>>> > > setting.
> >>>> > >
> >>>> > >
> >>>> > >
> >>>> > > On Sat, Aug 27, 2016 at 6:05 AM, Stephen Mallette <
> >>>> [email protected]>
> >>>> > > wrote:
> >>>> > >
> >>>> > >> Now that gremlin-python has been merged to master, you might
> wonder
> >>>> > about
> >>>> > >> what this has done to our build/release process. Well, not very
> much.
> >>>> If
> >>>> > >> you do your standard
> >>>> > >>
> >>>> > >> mvn clean install
> >>>> > >>
> >>>> > >> on master right now you should see that everything builds and is
> good
> >>>> to
> >>>> > >> go - even gremlin-python. So, everything is good, right? right?
> well,
> >>>> > yes
> >>>> > >> and no.
> >>>> > >>
> >>>> > >> The "yes" aspect here is that the entire project still builds
> with
> >>>> maven
> >>>> > >> which keeps our build toolchain simple. Users can just have java
> >>>> > installed
> >>>> > >> as they always did and still get a clean build of TinkerPop. The
> "no"
> >>>> > >> aspect is that native python tests (and if you were deploy,
> >>>> > >> packaging/deployment tasks) did not execute.
> >>>> > >>
> >>>> > >> What's good however is that even the native python build tasks
> are
> >>>> still
> >>>> > >> just part of the maven toolchain. You just need to have python
> 2.x
> >>>> > >> installed and, if you do, build with:
> >>>> > >>
> >>>> > >> mvn clean install -DglvPython
> >>>> > >>
> >>>> > >> You will now see in your output the results of native pytest
> >>>> execution.
> >>>> > I
> >>>> > >> think this approach almost sets the basic pattern for future
> GLVs. I'd
> >>>> > >> prefer to not have -glvPython to some degree and simply detect
> python
> >>>> on
> >>>> > >> the system and then execute natively if it can, but then i think
> about
> >>>> > what
> >>>> > >> happens as we add more GLVs and then i sorta like the idea of
> having
> >>>> the
> >>>> > >> specific option to turn things on and off.
> >>>> > >>
> >>>> > >>
> >>>> > >>
> >>>> >
> >>>> >
> >>>> >
> >>>> > --
> >>>> > David M. Brown
> >>>> > R.A. CulturePlex Lab, Western University
> >>>> >
> >>>>
> >>
> >>
> >>
> >> --
> >> David M. Brown
> >> R.A. CulturePlex Lab, Western University
> >
> >
> >
> > --
> > David M. Brown
> > R.A. CulturePlex Lab, Western University
>
>
>
> --
> David M. Brown
> R.A. CulturePlex Lab, Western University
>

Reply via email to