Tuesday, May 25, 2004

Partial Trust - response

Jon Udell explain's in his article The challenge of partial trust that he experienced trouble with mixed permission's on XP. I have also experienced similar frustrations. The main culprit seems to be Office and on two machines with repeated installs I have failed to make this work, where I can control the machine and my family can also use it safely without knowing how it works.

It seems to me that the analogy of the car works here. The mechanic keeps it running and I do not have to know what they do, just pay the bills. The computer needs to get to this point where I can do all I need without having to resort to sledge hammers! It is interesting that the Audi A2 has a special hatch for all the functions that a user is expected to interact with, oil, water and fuel. The mechanic has access to the engine compartment, but the user does not need to ever enter the hallowed bay. Would it not be excellant if when I install new code, it could act as the mechanic and I could sit back and relax knowing it would still all work after the event.

Thursday, May 20, 2004

XML template files

Why do so many applications that say they are XML compliant actually use simple fill in the blank templates. Is this an issue? There seems to be a problem with empty versus optional in XML. A number of apps I have com across seem to take a best case template and then fill in the fields. If a filed such as a date is not supplied they leave the tags in with no content. This then trips up on schema checks and other format checks as an empty date may be interpreted as incorrect, rather than optional.

What seems to lie behind a great deal of these type of apps is the fact that XML is tacked on and not XML parsers or DOM tools are used to create the XML output.


Thursday, May 06, 2004

Is the Post Office the best custodian of Addresses

For a number of years now the Post Office in the UK has been the custodian of Address data. This is based upon the fact that they produced an encoding of an address known as the Postal Code. This code is designed to aid sorting of mail down to the level required by the individual postman's round. For postal items this is a very clever and useful solution to a problem. However, it suffer from a clear computing problem - As a primary key for an Address the Postal Code does not conform for Codd and Date's definition. This is because, a) it does not uniquely identify a property b) it changes over time c) it does not cover the entire set of properties and buildings to which other might need to refer.

Identification of a Property

All properties that will have mail delivered to them are associated with a postal code. This code may cover many properties within a street. or even a particular company working within a property

This means that it is not possible to uniquely identify a single property from a postal code. This means that the properties number or name has to be added to the postal code

Changing Postal Codes

For the Post Office the requirements for a code are based around their needs for the idenitifying appropriate delivery routes. If a route has a property that receives large amounts of mail this is identified by its own Postal Code. Therefore when a new compoany moves into a street or premises, the postal code for that property may change. These changes are sent out regularly with the Post Office Address Finder CD-ROM. Most applications that update addresses at key transaction point rarely update this information leading to future problems when transactions are initiated using the new Postal Code which is not recognised.

Not Every property has a Postal Code

The Post Office quite rightly does not identify properties to which they do not deliver mail. However, other companies do wish to identify these premises. This means that when trying to identify an address in the UK any technical solution needs to be able to handle both postal Code to Premises matching and also some other mechanism such as GPS ref. or grid ref.


As I contemplate this complexity I begin to wonder how much UK plc losses because business up and down the line do not have a satisfactory way to unique identify buildings and locations that is generic across industries and is as accessible as the Postal Code solution.

The Ordanance Survey have a solution based on uniquely identifying every building in the Uk using a 16 digit number. Associated with this number is extra information such as apartments and possible the postal code information. The main draw back is it is completely inaccessible to the person wishing to send a post card to Aunt Milly, but is is great for getting that Aerial site on that remote hill side if all parties use the OS software.

I have come to the conclusion that the government of the UK should oil the wheels of commerce by introducing a system that is more comprehensive that the Postal Code system, but is utilised by the Post Office and other delivery companies to manage their businesses.

Why should the governemt be interested in this? Virtually every transaction carried out by the utily companies in the UK requires an address match to check the premises are correct. This is probably the single most complex step in the transaction and it subject to the most failures and require most manual intervention. Reducing the cost of these transactions across the whole of teh country would benefit every one involved.

Therefore I believe the Post Office is NOT the best custodian of Addresses - here in the UK. Is this the same in other countries?

Workflow versus Process Automation

When does a workflow engine become a process automation tool and vice verse?

This seems to be a non-question in my mind but it seems to exercise people greatly, especially vendors and the various standards groups that are working in the web services area.

When I first used a workflow tools (Oracle Workflow 2) I was implementing a solution that was designed to connect our front portal applications through to our back ends. As happens from time to time in work, I found the exercise immense rewarding in that it forced me to rethink the way in which I imagined a solution. No more monolithic designs, not even the sweetness of a OO but a simple set of modules that could if possible stand alone and be played with by a process designer at will.

As the solution involved both routing work to humans as well as robots I did not understand that this was not normal for most 'Workflow' applications. In fact I designed the application as 90% manual with the aim of getting stats to say which parts should be automated. In this way the application itself along with suitable humans actually drove the requirements process. Oracle's Workflow tool was very suited to this type of work particularly as most of the robots were database centric.

So where does this question arise? It seems that the next generation of tools based on forthcoming BPEL based technology have simply overlooked the benefits of dove tailing Workflow solutions with Process Automation. When a Process fails where do you need to route the fault to? Normally a human - so why do most tools make this a cumbersome task? Why do these so called next generation tools find dealing with people such an alien idea? I believe the answer lies in the fact that most of these emerging tools have been built by people used to handling classes that rarely touch humans directly. They tend to be focused on the J2EE/.Net like frameworks which are low level in the inspirations and have failed to take into account the gains of the 4GL world of the early 1990's.

This inability to see the bigger picture has meant that large investments in Process Automation and EAI technology has failed to benefit from the humdrum, low tech solutions that are sidelined as being manual.

Part 3: Architecture - we need a better analogy

It has been a couple of months since I published parts 1 and 2 of my ramblings on Architecture in IT. The original article was an internal document that I have tried to make more generally applicable. Most of the comments in my expected Part 3 were linked with the various internal architectural initiatives which I can not comment outside of my company.

Here are some of the conclusions I can share

The Structural plan should be based on an expression as to how we want to do business and not constrained by our current creaking (yet working) system infrastructure. The changes in the atmosphere of London over the last few years are tangible and have come about through careful thought of the overall vision and the implementation of key yet sometimes small changes. The new bus map in London has moved me from a rat in the underground to a top floor tourist enjoying and appreciating the landscape of London.

On the detailed level I would very much like to see Architects involved in developments in the same way the architects are involved in major developments. The name Norman Foster is associated with many key amazing buildings and developments. It is his name that is linked with the result. This is because not only did his team design the buildings but also they maintained control of them through to the final opening. This means that the concepts put to planning authorities are seen through to fruition by the same people who expressed the plan. Without this continued involvement and enthusiasm you would find a set of mishmash results, which fail to express the real solution and exude an air of missed opportunity.

In this view of Architecture, I am an Architect. I am not a Town Planner. I do not have that grasp of the trends of change, or the business needs to construct the Plan, but I along with the majority of 'Solution' people are capable of understanding such a Plan. In fact when expressed in visionary language I feel that most people would find it more inspiring than the current 'Architectural' language. You will notice that there are key areas left to architects to decide - the technology of Systems outside of basic infrastructural pieces is left in their hands. This freedom to innovate within a framework is key to any respectable architect. Without this the enthusiasm does not get expressed in the designs and instead of beautiful landscapes of juxtaposition designs you get the homogenous repetition of the authority housing estate.

This page is powered by Blogger. Isn't yours?