- From: Danny Ayers <danny666@virgilio.it>
- Date: Fri, 16 Apr 2004 12:17:01 +0200
- To: www-rdf-interest@w3.org
I recently came across the new OpenStack project, which aims to develop
an open alternative to Microsoft's tightly-coupled stack of desktop
applications, leveraging the XML-friendliness of OpenOffice, Mozilla and
Chandler. It struck me that the integration/interop of these apps was
probably a very good fit for Semantic Web technologies. I suggested this
to Gary Edwards, the project lead, and below is his response. Sounds
rather promising...
-------- Original Message --------
Subject: Re: OpenStack and RDF
Date: Fri, 16 Apr 2004 01:15:05 -0700
From: Gary Edwards <GaryEdwards@Yahoo.com>
To: danny666@virgilio.it
References: <407D2C4D.7050906@virgilio.it>
Hi Danny,
Thanks for the interesting proposal. I feel most privileged that you
have taken the time to contact me about the OpenStack Project. And yes,
it does make sense to merge the OpenStack efforts with the Semantic Web.
I'm not sure what that means in practical terms though.
Your eMail has many thinking in so many directions i'd like to respond
by breaking things up into chunks of opinion that i might be able to
manage a bit better:
*
Merger <#Merger%7Coutline>
*
Metadata, Adobe & The OASIS OO TC
<#Metadata,%20Adobe%20&%20The%20OASIS%20OO%20TC%7Coutline>
*
Thoughts on The Semantic Web <#The%20Semantic%20Web:%20%20%7Coutline>
*
Brief History of OpenStack
<#Background%20on%20the%20OpenStack%20Project:%7Coutline>
*
OpenStack Presentation
<http://theequityexchange.com/openstack/slides/openstack.html>
Merger:
It seems to me, that the Semantic Web is a disciplined method for
putting the pieces of an information ecosystem in place so that live
documents (content, data, streaming media and discreet chunks of
computation) can be pushed, pulled, accessed, exchanged or
collaboratively interacted with across the infogrid by human and machine.
By way of contrast, the OpenStack is simply an alignment of core
applications trying to drive the human interface to your Semantic Web.
As long as OpenStack components practice the discipline of your infogrid
mathmatics, would that qualify as a merger?
Metadata, Adobe & The OASIS OO TC:
I also agree with your assesment of RDF and OpenOffice.org metadata. I
represent the OpenOffice.org community on the OASIS Open Office XML File
Format TC, and extending the metadata functionality of the universal
file format is one of the primary issues that will be taken up during
the phase II effort. Phase I was completed in early March, with the
final specification now ready to submit to OASIS membership for comment
and approval. Part of my personal responsibility for phase II issues has
been that of contacting Adobe concerning metadata. I'm also responsible
for the Contact-PiM-Project Management-Calendar-Scheduling schema issue
that will be part of the phase II work. For help on this i've turned to
the Chandler, Glow and Evolution Projects for assistance.
The Semantic Web:
Honestly? This stuff is way above my pay grade. Lucky for me RDF
provides a comparatively easy to follow implementation map. Emphasis on
“comparatively”. The Semantic Web vision represents a galactic blueprint
for a future way beyond the humble hopes of OpenStack. And the RDF model
describes exactly the method of presenting and moving information
through this newly christened universe, where mankind and the machines
of mind are meshed into flows of information rendering volumes of
knowledge wherever the spark of wonder strikes. You've just got to love
it. But how to get there from where we are today?
The vision is so big, so grand and all encompassing that one can easily
imagine open stack models living harmoniously with proprietary blends of
all sorts, across all kinds of platforms. The opportunity here is that
there is no way the Windows XP Stack will enable any kind of transparent
RDF implementation. Chairman Bill will no doubt embrace and extend
Semantic Web principles, but it's going to be a one way street. RDF in,
proprietary Longhorn only extensions out. That's good news if we can
establish an OpenStack environment across the many Windows distros. One
that properly implements RDF.
Background on the OpenStack Project:
The OpenStack Project began in early 2003 as the tech media began to
analyze and take the measure of Microsoft's latest MS Office beta
version. As the calls for response came into OpenOffice.org, we began to
realize that the next version of MS Office would be something far beyond
what OpenOffice, or any other traditional office productivity suite,
could be measured against. A feature by feature comparison just wasn't
possible. All of the advanced “collaboration” features in MS Office 2003
were being tied into MS Server suites of one sort or another.
There was every indication that Microsoft was hell bent on releasing the
next version of MS Office as a highly integrated application
environment. A new layer of entangling interdependencies bolted into the
increasingly integrated Windows XP Stack.
OOo could cut it as an “application environment”, but this integrated
stack business was something else.
While OpenOffice.org in and of itself didn't comprise such a sweeping
integration of dependencies and interfaces spanning versions of OS's,
applications, communications and messaging protocols, developers
framework (.NET), and server suites, we did realize that many of the
components an alternative “OpenStack model” would need are in fact
available. One just needs to know where to look.
So we set out to advise the tech media that the many open source
communities and corporate participants are working towards a Windows XP
Stack alternative. One that would be much more fluid, enabling citizens
to interchange components as needed, without breaking critical
dependencies. Right.
So the OpenStack Project began as a loose cross community cooperative,
strictly focused on cross platform (Windows, Linux and OSX) productivity
components that would correlate to core Win XP Stack layers. We felt
that the only way OpenOffice.org could be credible posed as an
alternative to what became known as MS “Office System” was to complete
an Open Stack alternative to the XP Stack, and argue the importance of
OpenOffice in that stack.
We also knew that telling consumers who were seeking an XP alternative
to put together their own “stack”, simply wasn't going to fly.
Especially so with the great herd of 400 million Windows users fighting
the cost of migrating to XP. “Role your own” just doesn't cut it, and
sooner or later Microsoft was sure to get all the layers of the XP Stack
together, wrapped in an easy to understand marketing message.
The most costly message in technology is the cost of confusion. And
that's what we realized we were selling. Confusion.
Cross Platform – Cross Distro:
We quickly came to the realization that what we were working on could
also be described as a cross platform, highly portable, developers
environment. An environment based on the productivity trifecta of
OpenOffice.org, Mozilla.org, and Chandler.
A Managed Environment Developers could reliably target:
Perhaps most importantly, this environment had to also be a “managed”
environment. One managed by the core applications, not the end users (as
is so often the case with Linux, especially in comparison to the tight
fisted control Microsoft exerts over Windows). One that developers could
reliably target without having to worry about their dependencies being
broken.
Distro Independent:
We realized that what we needed was a cross platform, managed
environment, that was also “distro” independent. And, we sought to treat
all Windows versions as just another “distro” the environment could ride
over.
The “Windows as just another group of distros” part is very important.
Some 70% of OpenOffice.org downloads are for some version of Windows.
USAToday estimates that there are 400 million non XP Windows
installations out there. Microsoft has admitted that the greatest
challenge to their financial future is that of migrating 350 million
users over to Windows XP. The XP Stack components only run on Windows XP
desktops and servers, (with the only exception being that of Windows
2000 with service pack 3 – which has the DRM facility).
One of the more important OpenStack arguments is that this is a rare
moment in time. Microsoft has left the entire monopoly base flapping in
the wind, offering the great herd of Win32 API users one option: migrate
at great expense and effort to XP or die.
Another point about the OpenStack concept is that we lay off the
“management” aspect of the portable environment onto the core
applications. Our thinking is that we don't want each and every distro
manhandling the dependencies. Nor do we want every developer tapping
into the environment to optimize the arrangement of dependencies to meet
their specific needs. Instead, we would rather developers wrote to the
environment. Which means they need both a blueprint, and, confidence
that the core communities will reliably maintain things. And it's also
important that the end users, the computational consumers, are not left
having to manage what could be a thriving ecosystem of innovative
productivity.
Lifting management of the environment out of the hands of the distros is
problematic. The core applications seem to be up to the task. But the
distros are pushing beneath them, and the corporate stacks are moving
aggressively to create their own environments. Sun's Java stack is
nicely done, but noticeably not cross platform. Novel is well positioned
to finally challenge Microsoft with an alternative to the XP Stack. The
GNOME/MONO/WiNE effort is extraordinary. And i expect that the
GNOME-CygWin project will go a long way towards making the port of
Evolution easier.
Having witnessed the conformity efforts of groups like UnitedLinux and
FreeDesktop.org, we thought that the core application communities might
have a better shot at enforcing the manner in which dependencies were
managed. And if the core applications can take control of the
environment, there is no reason to doubt that they will not also be able
to reliably implement RDF and other principles and disciplines of the
Semantic Web.
OpenStack Principles:
Based on your eMail, a quick outline of objectives and principles might
look like this:
*Objectives:*
*
Cross Platform
*
Cross Distro
*
A managed environment developers can reliably target
*Open Standards Principles:*
*
Open Interfaces
*
Open Communications & Messaging Protocols
*
Open XML Technologies (including file formats)
*
Adherence to RDF
The Dell Model of Software Configuration:
Sadly we are a long way from the rather humble ambitions we started our
with. Our hope is that multi distro brokers such as “Progeny” will
endorse the OpenStack concept and enable us to take that first giant
step towards cross platform conformity. I have had rather promising
conversations with Ian Murdock about these possibilities.
The idea is that each of the core applications will take responsibility
for their related frameworks, with Sun taking over the Java component.
*Sun:*
*
JVM
*
client side Class Libraries
*
J2SE environments
*
J2EE accelerators
*OpenOffice.org:*
*
VCL
*
Jython components
*Mozilla.org:*
*
XPCOM environment
*
XUL environments
*
JavaScript
*Chandler:*
*
Python
*
wxPython / wxWindows
*
XML Object File System (SleepyCat)
*Novel:*
*
Mono/GNOME
*
cross platform version of YAST
It doesn't take a rocket scientist to see that the OpenStack has
problems of redundancy and turf conflicts. I have spoken with
representatives from IBM and Oracle, which further complicates things.
And then there are the Server side participants like Zope and Plone who
have a significant presence in the OOo community. They are primarily
interested in writing server side accelerators, and need the reliability
of a managed environment. However, they were a bit taken aback when
Chandler switched from ZODB to the SleepyCat DB.
Many in both the OpenOffice.org community and the Chandler community
were hoping for a foundation level implementation of an XML Object DB
that all applications could share. When Chandler swapped horses in mid
stream, it sent ripples throughout open source communities. One soon
realizes that open access and the transparent sharing of core components
is vital if we are to make that shift from the limiting mindset of
holistic application to that of a very rich and robust portable
productivity environment. Oh well. If it was easy, we'd all be home by now.
An XML Engine:
One of the more important aspects of the OpenStack environment is that
it is an XML engine purring at the critical inflection point where end
users interface with information systems and networks of information
systems. And of course, we would like to mount that engine on a solid
RDF block. That way, knowledge workers will be able to dock at will with
informations systems not yet dreamed of, but certain to be part of our
future.
Moving the XML power train to the end point where knowledge workers work
has to be an important part of the Semantic Web plan. While working on a
project for Comcast, i saw first hand how important XML is. Especially
so at all the intersections of an enterprise web application where
information collides with other information and with foreign user
interfaces.
The Comcast dilemma is perhaps typical of many if not most enterprise
problems. Following a rapid acquisition and merger strategy that
dramatically expanded the company, they needed to connect many disparate
silos and proprietary information processing systems. Wherever we
looked, the core business process of sales staff placing advertising
orders for cable time slots was both manual and unmanaged. Sales agents
spent 80% of their time writing reports and fighting the ad placement
system. And 20% of their time actually doing presentations, servicing
existing accounts, and selling. What a mess.
I won't bore you with the details, except to say that the solution we
came up with was based on a very simple dictum that i think would apply
to most web applications. Especially those designed to integrate many
business processes by working across many disparate information
infrastructures and systems.
The simple dictum is to first, get everything into XML. IN the case of
Comcast, it was extremely important that the office productivity suite
produce XML, that could then be meshed at a central XML Hub, and routed
to back end silos and time processing systems, or lifted into the Tomcat
portal, Plone collaboration center, or beyond through XUL based web
applications.
Once the information was in XML, there was no limit to what we could do.
Capturing the desktops was critical though since that was the major
point of disconnect and business process failure. One of the more
interesting issues was that we could dynamically link OpenOffice.org's
Impress presentation interface to most of the third party research and
analysis services critical to the sales agents account management and
sales services. Very cool. Very XML! And not possible with Microsoft
PowerPoint.
Honestly Danny, my passion for the OpenStack really comes from
witnessing first hand the amazing “systems” potential made possible by
getting all the information into an openly structured format. People
simply don't yet realize the possibilities that will avail themselves
once we start to put the volumes of knowledge worker fodder into XML.
Where we can finally apply the awesome power of our computation
machines. And if we do this under the auspices of a methodology like
RDF, the future will be one where all opportunities are a go. Without
having to reinvent the wheel.
Thanks for taking the time to contact me Danny. There is an OpenStack
slide presentation
<http://theequityexchange.com/openstack/slides/openstack.html> at:
http://theequityexchange.com/openstack/slides/openstack.html
It may be a bit out of date. And for sure it suffers from the desire to
please everyone. OpenOffice has an official policy of making the best
effort to reach out and include anyone who wants to participate, whether
it works or not.
Hope this helps,
~ge~
Gary Edwards
(650) 365-0899
(650) 888-2268 c
Redwood City, CA 94063
garyedwards@openoffice.org
*Danny Ayers wrote:*
> Hi Gary,
>
> I've just been skimming the material on site, and note that two of the
> components cited - Mozilla and Chandler - have strong tie-ins to RDF,
> a language with a model beyond that of multi-flavour XML (which
> doesn't actually have a model), which allows interop at a more
> semantic level. Web-friendly semi-structured data with arbitrary
> relationships can be represented, and content can either be marked up
> with metadata internally or externally. In what it brings this
> approach is close to the much-vaunted ideas in Longhorn, except in a
> standards based, distributed, non-proprietary form. RDF is being
> adopted elsewhere in spaces not unlike that of OpenOffice - e.g. all
> data from Adobe products now carries RDF embedded in its content
> (XMP). The XML base of OpenOffice would make RDF integration very
> straightforward. Given that the object of the Open Stack exercise is
> very close to the W3C's Semantic Web initiative (high level interop),
> wouldn't it make sense to use those technologies (primarily RDF and
> OWL) for the common languages of desktop (and server) productivity
> systems. This wouldn't really be an different approach to plain-XML
> interop, because the RDF model is designed to work together with XML
> documents and data on the web. It would act more as a consolidation
> tool. So might it not make sense to merge the Open Stack stack with
> the SemWeb stack, as shown in the "layer cake" diagram? :
> http://www.w3.org/2001/09/06-ecdl/slide17-0.html
>
> Cheers,
> Danny.
>
--
----
Raw
http://dannyayers.com
Received on Friday, 16 April 2004 06:17:29 UTC