Software architecture notes

This article or section is a stub. Stubs are articles that have not yet received substantial attention from the authors. They are short or insufficient pieces of information and require additions to further increase the article's usefulness. The project values stubs as useful first steps toward complete articles. See Wiki workflow and Wikipedia:Stub for more details on this process. See all the stubs in Category:Stubs

This specification describes what is required for setting up a platform, an organisation designed to form a network of completely independent nodes with other platforms, which we call the platform network. The high-level conceptual development path for platform technology is described here. We will describe our requirements and use cases in order to determine the correct components to use for various aspects. These requirements are based on our research over the past ten years or so regarding what is required for a decentralised, robust and efficient solution to maximise personal empowerment, freedom and privacy and allow for the emergence of bottom-up and grass-roots solutions, and organisations, based on the needs of the people. In our manifesto we describe the principles and values that guide our work and are foundational to these specifications, in particular the four criteria:

• Openness
• Completeness
• Think Global, Act Local
• All Aspects Changeable

Personal Platform

Beginning with the relevance to the individual, the platform is a software package that enables the user to organise his or her own life, to maintain an overview of goals, values, tasks and commitments, much as described by the GTD movement, for personal organisation. In addition, we require communications tools and common office productivity software such as email and schedule, web browser and an office suite. These needs are rounded off by online private and secure file storage services to ensure backup of vital data and synchronisation across multiple devices.

This can all be done now...

Now these things taken individually are nothing special, anyone can choose to set up a computer with a free or paid-for operating system, such as Linux, Apple or even Windows, then install free or paid-for applications to achieve office productivity and personal organisation requirements, followed by setting up online user accounts to access services and even applications in the "cloud". However, there are a number of limitations that become obvious when currently trying to fulfil even these basic personal requirements using off-the-shelf software available on the market today. Without going into full detail, the key issues are:

Fragmentation of data and user accounts

This toolbar along the bottom of a blog article allows people to share the information using their favourite service, possibly several services. Each of these services would have a separate log in and separate data such as user profiles that will need to be kept up to date.

Restriction of freedom and loss of data privacy

Who has time to read through all that boring legalese, just to set up an account or install some software? Unfortunately it turns out that some terms and conditions are anything but reasonable, claiming ownership and perpetual use of all of the data you upload or share on that service. Wikipedia:Information privacy Interviewing Richard Stallman: Freedom in the time of SaaS

Having to pay for proprietary software

Use a lot of online services? Those costs can add up, not to mention being prohibitively expensive in so-called "emerging markets".

Lack of offline synchronisation with most cloud services

Got no Internet? Tough, if your e-life is in the cloud. Most services don't allow for easy offline synchronisation.

Freedom in The Cloud

For more information, mainly on the aspect of privacy and the impact of the centralisation of data on our freedom, view Eben Moglen's talk that inspired the creation of a unified, private open source alternative to Facebook and similar services: Diaspora.

What we want instead

In contrast, what platform offers is an integrated, consistent user interface, with the context being navigated comprising the users areas of interest and personal projects, rather than fragmented applications. We envisage a fully packaged installation file (ISO) that allows an operating system to be set up from scratch on any device, offering a web browser interface to allow the user to define their areas of interest and projects, import legacy data and access any applications as required to do work or just socialise online.

The Platform is a self-contained solution that could for instance run on a laptop, whether or not there is Internet available, which could fetch updates, messages or files as and when connectivity is available. Imagine having one place from which everything is managed, messages and updates are sent from, a private, secure online home with a unified inbox, set of files, bookmarks, documents, contacts, to be easily shared, in a system that just works and is completely self-contained.

The good news is that this is all possible now, by way of integrating existing open source technologies using the right kind of "glue". Further on we will identify our current candidates for fulfilling various aspects described here.

One tree and tag cloud that holds it all together

The framework platform is based upon from can be thought of as a "glue" that integrates a number of existing technologies and allows new technologies to be integrated as they become available. At the interface level, a tree represents the ontology of the user or organisation being viewed. This tree is a meaningful and coherent structure that represents a small number of top-level concepts. The tree will be different when looking at one's personal "electronic life representation", in which case it consists of "areas of focus" in life, as opposed to viewing a shared organisational tree. David Allen coined that term and recommends having a perspective on life (20.000ft) at which only these areas of focus are visible in order to assess how much time one is spending in different areas and whether they are in balance. This area is in between the most high-level aspects, such as goals, vision and values and the more concrete day-to-day aspects such as projects, checklists and tasks, which would typically be associated with one or more areas of focus. On the level of organisations, this concept maps quite directly to what is normally called a "department" or "division".

• Some example tree structures for personal organisation and business
• 

  David Allen, creator of GTD, shows his company "areas of focus"/departments in this tree

• 

  David Allen's personal organisation areas of focus

• 

  A personal organisation tree done with mindmap software

As you can see from the examples, the structure can be quite simple and encompassing at this level.

Our goal is to enable such a tree to replace any hierarchical navigation structures being used by existing applications, such as the folder trees of email programs, file managers or bookmarking programs. Instead, the personal or group organisation tree will be created once, using a simple, forms based process, then linked to applications. This will then allow one structure to be navigated and added to within a consistent interface, with each area of focus or department representing a portal dedicated to that concept, and the applications serving various types of "items" or "content", without confronting the user with a different interface, behaviour, or data synchronisation headaches.

Now that we have a shared hierarchical context structure that is consistent across applications, we also need a global system of tags which applies across application boundaries and can be used in conjunction with the tree to filter down what is being viewed. That way we could have quite a powerful framework that not only allows hierarchies but also the more whimsical and flexible tags (known in wiki terminology as "categories"). The way the Firefox web browser organises bookmarks is an example of accommodating both trees and tags. The most popular way to display tags is as a cloud within which the most popular tags are displayed in larger font and clicking on them will lead you to a collection of every item that has the particular tag associated with it. The problem with tags currently is that you will need to create a set of tags for each application that can handle the concept and there is no way to click on a tag that will show you all items across application boundaries that have that tag. We would like to see that solved, giving us a unified navigation and filtering paradigm and allowing applications to be integrated "behind the scenes" to allow the manipulation of chats, projects, tasks, events or whatever "item" they might be providing. The Cynapse application views are an elegant example of how this could be displayed:

Example

For example, imagine you are viewing the tree of "Open Corporation" and wish to follow up on some sales opportunities. You would have selected the "Sales" branch, which gives you the sales portal in the right hand application view. In addition you select "Opportunities" in the tag cloud of "Open Corporation". Then, in the right hand pane you select "Contacts" from the items button. You now have the total amount of contacts filtered down by contacts assigned to the sales department, tagged with "opportunity".

Further topics to cover

• User account synchronisation and storage
• Smart caching of content
• High-level portal structure
• User interface and preferences
• Definition of objects and queries, easy creation and adjustment of such
• Subscription to packages of content and functionality across underlying applications
• Robot framework for import, synchronisation, account logins

Record Administration

One aspect of our prototype wiki organisation system has which we haven't been able to find in existing open source solutions is the ability to replicate our portal structure. It's based on the idea of extending the CMS/Wiki to allow the users to adjust their portal structure to suit their own needs using wiki organisation packages as an initial starting point.

In terms of functionality requirements of the portal structure, the foundation concept is that of records and record-types. Records are sets of properties that can be associated with an article. A record-type consists of a form which is used to edit the records of that type, and a template which defines how that set of properties should look in the page.

In conjunction with this is the ability to create simple queries based on property values and resulting in a set of matching records. The results can be rendered as a table or a list, and can be further transformed by using the host CMS's templating system, for example to render the results as a select list in a form, or be adjusted to suit a javascript widget, report or chart.

Queries can be done in a user-friendly way since the templates that need to be added to an article to create them can themselves be record-types so that they're created and modified via their forms rather than difficult syntax.

This approach allows the organisation members to effectively define their own applicational structure which is a dynamic and integral part of their documentation and knowledge structure.

Current solutions and functionality to integrate

• Groupware applications --> emails, contacts, calendars, tasks, projects and Group decisions
• Global accounts and synching via robot framework - OpenID?
• Personal portal and broadcast centre
• CMS
  • Content (procedures, roles, etc.)
  • Ontology portals structure
  • Records, forms & queries
  • Organisational setup (forms) and org. templates
• E-commerce
  • Payment methods/merchant accounts
  • Accounts management
  • Invoicing & reporting
  • Book keeping
• Physical network

Our development plan

Beginning with being able to manipulate one tree across several applications, followed by the use of a unified set of tags across applications, using our Wiki Organisation approach, we will begin to implement the ideas outlined here. Further detail is described in the mashup article.