[Devtools] Atomic App / Nulecule spec reviews

[Charlie Drage]

See inline comments below

On 05/18, Pete Muir wrote:
> On 18 May 2016 at 14:50, Charlie Drage <cdrage at redhat.com> wrote:
> > Hi all!
> >
> > I've made some considerable progress in regards to the Nulecule /
> > Atomic App direction and how we're becoming more of a deployment tool
> > for distributing and deploying multi-container applications.
> >
> > However! I'm starting to become blocked now due to awaiting spec /
> > code reviews as I begin to finish / catch up to my original workload.
> > Although I could merge it individually, it's best to get a community
> > review / support on it before continuing :)
> >
> > One of the biggest changes is the specification, as per pull:
> > https://github.com/projectatomic/nulecule/pull/205 which makes some
> > considerable changes to the way we write Nulecule files. If I could get
> > some input on it, that would be great as it would create some work to
> > get added to Atomic App / Nulecule as well as decrease the learning
> > curve associated with creating and generating a Nulecule file.
> 
> Could you summarise the changes you have made. I couldn't find an
> actual spec change in the changeset, just changes to example files.

Ah yes! It was just the addition of the example files as I found that
the easiest way to reflect the largeer changes without re-writting the
entire documentation.

If the original spec writers give their approval for the change I can
write a large PR which will change/re-write the original documentation
to conform to the new spec.
> 
> >
> > Another large change is the addition to *binary* generation for Atomic
> > App, creating support for not only Linux, but Mac OS X and Windows
> > platforms, allowing the ability to use Atomic App to deploy a
> > multi-container application on a cluster from whichever platform you
> > wish, whether or not it uses Python on the distribution. I've tested
> > the generated binary on multiple operating systems and it appears to
> > work awesomely!
> 
> I don't quite understand this one - do you have a video of it in action?
> 
> Would be good to add this to the sprint review.

Not yet! I've tested it on some other platforms such as CoreOS /
Project Atomic / Debian / CentOS which are of course all Linux
specific. *theoretically* there shouldn't be a
problem running it on Mac OS X / Windows. As the tool used for the
binary generation (pyinstaller) works on all platforms.

I'm awaiting for the PR's to be commented on if this is a step in the
right direction and then I can pull out my Windows machine and start
testing it more then. I'm sure I'll encounter a few bugs (I know of
one right now with .kube/config pulling...)

> 
> >
> > A third change is the addition of the Kubernetes binary usage to the
> > API, which (at the moment) contains an additional 1200 lines of code
> > to be added to Atomic App. Both functional and unit tests pass :)
> >
> > Many thanks and happy coding!
> > --
> >
> > Charlie Drage
> > Red Hat - OSAS Team / Project Atomic
> > 4096R / 0x9B3B446C
> > http://pgp.mit.edu/pks/lookup?op=get&search=0x622CDF119B3B446C
> >
> > _______________________________________________
> > Devtools mailing list
> > Devtools at redhat.com
> > https://www.redhat.com/mailman/listinfo/devtools

-- 

Charlie Drage
Red Hat - OSAS Team / Project Atomic
4096R / 0x9B3B446C
http://pgp.mit.edu/pks/lookup?op=get&search=0x622CDF119B3B446C