Ed Henning

Last month I originally intended to write about the type of operating system I wanted to see developed, but in the event I focused more on the perceptions of the majority of users.

It is, after all, that perception which drives sales and development, and not necessarily the technical quality or long term benefits.

When I reread the article after it had gone to press, I noticed a perception trap I had also fallen into when I referred to 'generic windowing operating systems'.

There should, of course, be no such thing, as the operating system and the windowing user interface are totally different in function and are arguably better kept as separate components.

The only reasons for them being integrated mainly concern marketing; from a technical point of view there is every reason for them to be separate.

This whole concept of integrating many components into one system may be good marketing for Microsoft, but it is not good for users, and it is also not good practice.

I never really believed Microsoft when it tried to claim that Internet Explorer was of necessity intimately integrated into the operating system - that they couldn't be pulled apart without something going wrong.

But I was very frustrated that none of the lawyers challenged Microsoft that if it were true, then it was simply appalling development practice, regardless of the marketing issues.

An analogy here is with object oriented programming. One of the key concepts here is that individual components are deliberately kept discrete, so they can be developed separately, tested individually and so on.

Making each component rock solid becomes much easier, and the overall system benefits as a result.

Microsoft has gone in completely the opposite direction with Windows and, if the system is as integrated as the company makes out, then there is trouble waiting down the line; what I call 'Ashton-Tate syndrome', for those of you old enough to remember what I mean by that.

The more complex a large system becomes, the more difficult it is to make changes and correct problems, with the eventual result that the system becomes impossible to maintain and develop further. The only way to avoid such a fate is to split the system into individual, manageable components, each of which can be replaced and developed without harm.

Once this has been done, it then becomes possible to sell these components discretely, giving the user better quality and more choice.

It should be in the interests of any company to give its users better quality products and greater choice, but the fact that Microsoft continually tries to force users down one road, with increasingly unwieldy software, is a symptom of dangerous short-term thinking.

In a properly component-structured operating system (many of these arguments also apply to applications), these individual components could be installed separately and interchanged.

A few years ago, a user could have different versions of Dos on a PC, and could run more than one windowing user interface as well. I had hoped that such component oriented development would continue, but Microsoft steered in a different direction.

An argument in Microsoft's favour is that more choice makes life perplexing for the majority of users, but that needs to be tackled at the supplier level, perhaps with PC vendors putting together a selection of components that only need to be changed by competent users.

There are other significant advantages to a component approach. For example, the file system itself should be interchangeable. Different situations need different characteristics in a file system.

A streaming video application in a digital video recorder or in a video server operates better with a different file system than that in a database server or desktop system.

The other components that should be separate include the internet browser, file management software, various utilities such as virus protection, encryption, file compression, backup - the list is long.

Microsoft's fear with this approach is that if it gives users more choice they are likely to buy components from other vendors. Of course that would happen, but you can only force customers for a certain length of time, and an eventual Microsoft backlash is inevitable.

I would much rather buy software from a company that gave me more choice and higher quality products, and I would expect such an approach to build proper brand loyalty over the long term. But Microsoft does not think long term.

Yet again, the eventual solution here does seem to be Linux, where the disk operating system is separate from the graphical interface, and there is in general a more healthy, component oriented approach.

The best thing that can happen to PC software development is the adoption of Linux in the workplace, on the business desk as well as in the back office, where its position is already well established.

A component-level approach makes it much easier for businesses to customise systems to their requirements, and the general robustness of Linux gives an added bonus.

Like this story? Spread the news by clicking below:

 del.icio.us     Digg this     reddit!