Of PR
Klint Finley
only person I'm naming in it, but I still wanted to check w/ the group
to see if you are ready to see press coverage or if I should hold the
story for a while. I write for a small Web publication called
SiliconAngle, which is read mostly by a small group of cloud
enthusiasts. But there's always the chance the coverage would get picked
up by Hacker News or something and be amplified to a larger audience so
I don't want to risk opening the flood gates if you're not ready for the
attention (on the other hand, it's on Github so that potential is
already there).
--
Klint Finley | SiliconAngle (DevOpsAngle Editor /ServicesAngle Senior
Writer) | http://siliconangle.com
Troy Howard
Ingy dot Net
Call it CloudFreeStyle or cloudfreestyle, please.
Ingy
Ingy dot Net
My opinion:Regarding names mentioned: I don't mind having my name included as long as my role is presented as an Apache Software Foundation representative/advocate/committer vs an AppFog employee.Regarding PR and timing: I think it would help if we got our "manifesto" online before getting too far. I think it's important to have a resource available that talks about how and why we came to the decision to create a community fork. Otherwise it's open to lots of speculation/rumination that could set things off on the wrong note. Perhaps we could wait until we've gotten to a point where we're more "ready" for the public eye?
+1
I think you are on the hook to get that manifesto started, Troy :)
I'd start it on the wiki, so we can all help you.
Klint Finley
So all one word, and Style capitalized?
>I think it would help if we got our "manifesto" online before getting
too far.
I can wait then. I tried to summarize the reasons for a fork, as
apolitically as possible, but if traffic is flowing to the project site
then it would probably help to have that info up there. It would also be
helpful for me to have that to refer to so that I can be as specific as
possible w/o overstepping.
Troy Howard
Adron Hall
Troy Howard
https://github.com/cloudfreestyle/cloudfreestyle.github.com/wiki/History
Anything else we want to do before announcing this to the world?
It's worth noting that Adron and I discussed the project at the most
recent PDX DevOps users group meeting on 4/16... so there was some
minor publicity that happened there.
I think it would be nice to show a roadmap, and perhaps get some
commits to our fork before going much further -- there should be
something a bit more concrete than the social aspects of this. There
should be real code changes that go into this fork as well. For
example -- even just changing the READMEs and code files to update
naming would start a legitimate divergence. Also, contacting folks
with outstanding pull requests to CF and asking if they are ok with us
merging in their changes to CFS. We need some legit momentum in the
codebase. :)
Thanks,
Troy
M. Edward (Ed) Borasky
> I've updated the wiki page regarding history of project formation:
>
> https://github.com/cloudfreestyle/cloudfreestyle.github.com/wiki/History
>
> Anything else we want to do before announcing this to the world?
>
> It's worth noting that Adron and I discussed the project at the most
> recent PDX DevOps users group meeting on 4/16... so there was some
> minor publicity that happened there.
>
> I think it would be nice to show a roadmap, and perhaps get some
> commits to our fork before going much further -- there should be
> something a bit more concrete than the social aspects of this. There
> should be real code changes that go into this fork as well. For
> example -- even just changing the READMEs and code files to update
> naming would start a legitimate divergence.
+1 - I've almost convinced myself to volunteer to update the "building
a local instance to hack on" instructions, since it results in
something I want to play with. But I'm not at all clear what the time
commitment is for it.
So ... is this right:
1. Fork the cfs repository to my Github account and hack the code so
it makes a working cfs command-line tool, then fix the docs in my
repo, then issue a pull request?
2. Fork the cloudfreestyle repository to my Github account and use my
newly-created cfs to build a virtual machine running a "micro
CloudFreestyle, then fix those docs in my repo, then issue a pull
request?
GIven my unfamiliarity with Ubuntu this looks to me like a couple of
weeks assuming six-hour days. In other words, I abandon
http://j.mp/compjournoserver's current path and throw myself under the
Lucid Lynx bus. ;-) This does have some advantages:
1. I get to learn Ubuntu command-line sysadmin skills
2. More of CRAN is packaged for Ubuntu than for Fedora (and more for
Fedora than openSUSE)
3. OpenStack is further along on Ubuntu than OpenSuse (I rate Fedora /
Ubuntu as a tie here)
4. *If* there is a better shot at world domination for
CompJournoServer on CloudFreeStyle than there is for it going alone or
on Cloud Foundry itself or on OpenShift, I should pivot now.
--
Twitter: http://twitter.com/znmeb Computational Journalism Server
http://j.mp/compjournoserver
Data is the new coal - abundant, dirty and difficult to mine.
M. Edward (Ed) Borasky
infrastructure myself. ;-)
M. Edward (Ed) Borasky
Fedora 17 (Beta), Ubuntu 12.04 Beta and openSUSE for an OpenStack
Essex Nova compute node. So far Fedora is winning because they have
the Nova documentation in the repositories and openSUSE doesn't. I
haven't had a chance to dig into Ubuntu's Essex yet - some time this
week I will, though.
M. Edward (Ed) Borasky
1. I'm pushing the 0.9.0 release out the door late Tuesday night,
April 25, 2012. It will more or less definitely have
a. a LiveDVD that can be installed to a real or virtual machine.
Exactly which virtual hosts will work is at least VirtualBox and
VMware Player / Workstation. KVM is desired as well. Xen is too
difficult for me to test on at the moment with openSUSE; I'll revisit
it when I have a coherent strategy for OpenStack.
b. A VMware Virtual Machine image. The Open Virtualization Format
(OVF) images don't work in KVM.
c. Redis, RabbitMQ and MongoDB. I have Perl packages for Redis but
not for Mongo. I have R packages for both.
d. Node.js
e. A full LAMP stack, PHPMyAdmin and PHPPgAdmin
f. amd64/x86_64 "vanilla" Automatically Tuned Linear Algebra
Subroutines (ATLAS).
g. Scripts to recompile R to the architecture and link to ATLAS on
a real or virtual machine
h. Scripts to recompile ATLAS to the architecture on a real
machine. This can't be done on a virtual machine, but libraries tuned
on the host can be used if the guest is running in the same machine. I
probably won't do that.
2. Decision dates:
a. Ubuntu 12.04 releases April 26. Fedora 17 releases May 22. I've
got the betas but haven't had a chance to dual-boot a test system yet.
The decision here is whether to spend energy porting the test host
setup to Ubuntu or Fedora to get distro-supported OpenStack Essex or
work with community-supported Essex on openSUSE. I'm leaning towards
Ubuntu / Fedora because the next openSUSE release isn't till July. I'm
pretty sure OpenStack will be in it, but I *know* it's in Ubuntu and
Fedora. See below for Ubuntu vs. Fedora.
b. OpenShift goes open source April 30. The decision here is
whether to keep building on openSUSE, switch to Cloud Foundry, switch
to OpenShift (RHEL 6.2 / Fedora) or switch to CloudFreeStyle. If I had
to guess now, I'd say OpenShift will win, if only because the Fedora
17 beta OpenStack Essex documentation exists and it doesn't for Ubuntu
or openSUSE.
I want to emphasize that the Computational Journalism Server, like
many other journalism projects, is deadline-driven. Fedora's taken two
slips on their way to a show-stopper-free 17, and that's a problem.
openSUSE is on an eight-month release cycle and they seem to be taking
slips as well. That's not something a typical journalist will
tolerate. So I'm going to be a hard-ass about it on this project.
M. Edward (Ed) Borasky
new item is some screenshots of the console during the install process
- there won't be any change to the appliance or the actual code.
Details are at
http://j.mp/compjournoserver
Ubuntu Precise Pangolin is due out tomorrow. If I can get a download
I'll probably dual-boot my workstation and see how well it hosts KVM.
Klint Finley
M. Edward (Ed) Borasky wrote:
> Version 0.9.0 was released today. 1.0.0 is due out Friday but the only
> new item is some screenshots of the console during the install process
> - there won't be any change to the appliance or the actual code.
> Details are at
>
> http://j.mp/compjournoserver
>
> Ubuntu Precise Pangolin is due out tomorrow. If I can get a download
> I'll probably dual-boot my workstation and see how well it hosts KVM.
>
--
M. Edward (Ed) Borasky
> Is it based on Cloud Freestyle?
OpenShift once I have the R parallel code nailed down. OpenShift
supposedly goes open source next week, so I'll know more once I can
look at some code repositories. I'm hoping to be able to do both so I
can have apps in the Interoperable Cloud App Gallery.
By "default", the openSUSE appliances only work on Hyper-V, Amazon EC2
and lower-level infrastructures (VMware / VirtualBox / KVM / Xen and
of course bare metal.) The R "cloud" parallel code use cases are also
mostly Amazon-based, including Elastic Map Reduce. I want to build on
top of OpenStack rather than Amazon if at all possible.
M. Edward (Ed) Borasky
<zn...@znmeb.net> wrote:
> On Thu, Apr 26, 2012 at 11:40 AM, Klint Finley <m...@klintfinley.com> wrote:
>> Is it based on Cloud Freestyle?
>
> Not yet - I'm planning a port from openSUSE to both CloudFreeStyle and
> OpenShift once I have the R parallel code nailed down. OpenShift
> supposedly goes open source next week, so I'll know more once I can
> look at some code repositories. I'm hoping to be able to do both so I
> can have apps in the Interoperable Cloud App Gallery.
through the documentation today for a blog post tomorrow. Fair warning
- I don't impress easily and what little I saw last night blew me
away. It doesn't help any that the central openSUSE repository server
lost all its hard drives yesterday and the community had to scramble
to hunt down mirrors. ;-)
Adron Hall
M. Edward (Ed) Borasky
> I gotta admit. They look like they landed almost on feature parity with
> Cloud Foundry.
>
> If that's a sign (especially since they started after the Cloud Foundry
> effort) of things to come, then Cloud Foundry - any and all efforts, is
> going to have to seriously get down to business.
>
> I've still got a ton of things going with Cloud Foundry/Iron Foundry, etc...
> but looks like I might have to double my efforts.
>
> -Adron
updated to 1.2 and for a *business* application with a pre-defined
business model I wouldn't consider OpenShift. But for my pure "hobby"
/ open source project, it's a toss-up between CloudFreeStyle and
OpenShift Origin. I think SUSE Studio comes in third because there you
have to build your own "glue logic" for the platform / infrastructure
interface pieces. That costs me cycles I could be spending on the R
parallel and computational journalism application coding.
Klint Finley
this? I'm done at SiliconAngle, but will be contributing regularly to
ReadWriteWeb and Wired Enterprise in the near future - both of which
seem like good places for a story on CloudFreeStyle.
M. Edward (Ed) Borasky wrote: