Projects in the 90's

aQute

In 1990, I founded aQute because the work at Mediasystemen lost a lot of its attraction from the wilder days. The company was established and moderately successful in the market. It had just sold a large system to a german newspaper.

Mediasystemen

1990 - Provided Mediasystemen with 1/2 day a week suppor for the transer. Assisted them in the design of the new editorial server and did some small turn-key projects.

Weekbladpers Quark Xpress

1991 - Weekbladpers, a major Dutch publisher, needed an interface between the Macintosh desktop publishing program Quark Xpress and their editorial system. This was developed with Smalltalk and some Macintosh programs.

Courses C++, Smalltalk

1992 - During the early 90's I gave many different courses. Some courses were given on behalf of Intel, others were for aQute. Among the customers were the Dutch PTT, Philips and others.

MacVonk Windmill

1992 - MacVonk was a Macintosh shop. Due to my experience with font technology, they asked me to develop a CD with a font selection program. The fonts on the CD would be crippled so that the customers could try out a font and then order it via telephone.

Metadesign Fontshop catalog

1993 - Metadesign is the design bureau of Erik Spiekerman, a well known designer in Germany. One of his endaveours is a shop that sells fonts called FontShop. At the time, FontShop had the largest catalog of fonts (over 4000). The production of the catalog had become so tedious that automation was required. Petr van Blokland and I were asked to develop a program that could print the catalog from a Filemaker database. This was a major problem because the Macintosh could not handle more than a few hundred fonts. I developed a Smalltalk program and a Macintosh extension that could link and unlink fonts as needed. This program has been used for many years.

Adobe Font Catalog

1993 - Adobe got interested once they learned from the FontShop catalog. At the time, Adobe was selling around 2000 different fonts and they needed a browser for their catalog. Together with Bur Davis (an Adobe employee) we developed a database of font information. I wrote a Smalltalk program that read the database and generated the files necessary for the Macintosh font browser. The font browser contained bitmaps for all the samples. The fonts could be found by family, by style, by name, and by designer. The browser was included on all their font CDs. In the late 90's Adobe used the same database to create their web catalog of fonts.

IBM Smalltalk Mentoring

1993 - In the early 90's IBM was heavily into developing OS/2 and they were looking for a faster way to develop applications. They saw an opportunity in Smalltalk, so one day IBM Ireland asked me to give a Smalltalk course in Dublin. After a few days it became clear that a few days would not be sufficient to make the group proficient due to lack of experience. After a discussion with the manager I got contract for 6 months to mentor the group developing personal information applicaitons like calendaring. This was a very interesting and pleasurable project, despite having to be away from home every other week. Unfortunately, one day the project was canceled because IBM had lost most of its interest in Smalltalk.

Ericsson Radio,Mentoring

1993 - In the summer of 1993 Ericsson Eurolab Aachen asked me to mentor a group of 15 developers in Aachen in C++ for 3 months. I spent about 2 days a week at the group and took great effort to teach them object oriented techniques. This was very much appreciated because the contract was extended to a year and after 6 months a swedish project manager asked me to come to Sweden and give some courses there as well.

The first project was MSA, the Mobile Subscriber Administration application. This application is responsible for managing the Home Location Registers, these are the computers in a GSM system that hold your subscriber information. The original program was highly convoluted and complex. By refactoring I had been able to remove most of the complexity. An add on to the MSA program to interface to a remote system turned out to be much simpler than anybody had thought due to the simplfications. Key issue what we did was describe the type information of the data we were handling instead of hard coding. Applications like MSA are very much screen, database, network element kind of applications. It is rare when the software has semantic knowledge of the variables. This model is therefore easy to parametrize. This way of working was very new inside Ericsson and I met some resistance in the beginning. However, after 6 months I was asked to come to Sweden and train some people.

Ericsson Radio, Graffiti

1993 - After MSA was on the rails, I was asked to get involved in BCP (I forgot what it stood for). The purpose of BCP was to become a library that abstracted map handling from an application. There were few people with graphic software experience so I was asked to supervise the design. We developed a very interesting library that handled the complex features to efficiently handle maps and combine it with database information. Maps are surprisingly hard because their scaling is several magnitudes. This library became used for several cellular applications.

Ericsson Radio, Blueprint

1994 - During one of my visits to Linkoping in Sweden the development manager of the Operating and Support Systems(OSS) for GSM, Leif Widlund, discussed with me the problems with the software development. I gave him a short presentation about about the problems I discovered and finished with a possible solution. Most OSS applications are very similar, they interface a user or a customer system to a network element. They rarely react on the content of the data. Such commonality could be captured in a framework. Additionally, there were many requirements to comply with OSI standards, something the OSS seriously lacked. A framework could go a long way to provide this standards compliance. He looked at me long and hard and then told me to make a presentation for Ericsson Radio top management in Stockholm. A few weeks later I did and was hired for 2 year. I moved to Sweden for this contract and worked for almost 3 years on this project. In the end we succeeded in delivering the framework though it has not been used in many applications due to severe non-technical cross-wind.

Ericsson Radio, Base Station Manager (BSM)

1995 - The Base Station Manager was application that managed GSM base stations in the OSS. It was the first application to be build on top of Blueprint. I was extensively involved in its design and execution.

Ericsson Radio, Cellular Network Administration (CNA)

1996 - After Blueprint, I was contracted by Ericsson Telecom to perform work on the most important GSM OSS Application: CNA. This application had grown to tremendous proportions. Analyzing this application made it clear there were tremendous redundancies that could be removed. We used many of the lessons and code from the Blueprint project to simplify the CNA application. We reduced the codebase (excluding the GUI) from 300,000 lines to less than 15,000.

Ericsson Telecom, Business Communication Manager (BCM)

1996 - Ericsson Telecom in Gothenburg asked me at the end of 1996 to develop a quick prototype for Telia's (Swedish PTT) business unit that handled Centrex. Centrex is a virtual Private Business eXchange (PBX) that is hosted in a normal district telephone exchange. I used techniques that were very uncommon in Ericsson at the time: internet, urls, icons, drag & drop, browser, and Java. During a 2 week period we developed a working prototype, including a URL based interface to the Ericsson AXE. The customer was so impressed that their managers ordered the further development of such an application. We developed the application as an applet, solving the myriad of problems that Sun nd Netscape had placed on the applet model. In the end, we could deploy the full application one from one location and showed that the ideas behind applets were sound. This products was developed for many different markets under the guidance of Ericsson Ireland.

Cellkeeper

1996 - Cellkeeper was a private project that developed a design for a cellular network manager. It resulted in a report.

Ericsson Research, Perspective

1996 - Also in 1996 I was asked by Ericsson Research in Kista, Stockholm, to do a research project with 3D user interfaces. At that time, VRML was the hype. This was a new standard (just before the XML hype and therefore is not XML based!), that was heavily promoted over the net. All browsers were supposed to support VRML any time soon. I developed an interface with Java that coupled the browser to a personal assistant system. This was a brilliant product developed by the department that had hired me. It was written in Erlang on specially developed hardware. It was connected to the phone system and had many features that were extremely useful. Mail server, answerinf machine, tracking of where the user was, calendar, etc. After the basic interface worked I developed a VRML application that showed all the posisble settings. Nice idea, was it not that VRML never made it to the mainstream. The best browser plugin was Cosmo, and that was a pretty bad piece of software. The VRML plugins in the end never made it to the market due to this bad quality. The project was finished but never made it to productization due this quality problem.

Ericsson Research, Marvin

1997 - In 1996 I had given a presentation about a personal assistant, a home server that had only one task: keeping its owner happy. After this presentation, the manager at Ericsson Application Research asked me if I wanted to work on such a project. The answer was YES! So in 1997 I started working on Marvin, the personal assistant. This involved making an infrastructure that could handle plugins. I developed an Infobase, which was a database that was really good in storing objects and relating it to other objects. The relations were meant to be implemented by Java classes. The idea was that you could extend the database with new relation types, so a plugin model was needed.

A lot of interesting work was done in Marvin. However, Ericsson Research was asked to assist Ericsson Radio with the development of the ebox.

Ericsson Research, ebox

1997 - The Wireless Internet unit in Linkoping had developed a small computer (20cm x 20cm x 7cm) that contained a 486 CPU, 32 Mb RAM, and 8 Mb flash. It had 2 ethernet interfaces, a serial port and a Lonworks interface that allowed it to control devices over the powerline. The business idea behind the box was to make it a consumer device, similar to the telephone. The ebox was programmable by different service providers that could provide unique functinality: phone services, home automation, storage, internet functions, etc. The services would be provided in an environment that was under control of an operator, the traditional customers of Ericsson. However, the potential customers had informed Ericsson that it was an interesting idea but that they would only consider this model if it was standardized. Ericsson therefore worked with Sun, Sybase, IBM, Nokia, Sony, Philips, and many other big and small companies to create a specification. The involvement of Sun implied Java. As a Java expert I was asked to prepare a demo and after the success of this demo I was asked to lead the standardization of the specification on behalf of Ericsson. This lead to the OSGi Alliance.

During this period I also worked with the Wireless Internet unit to further develop the ebox.

Ericsson Telecom, Service Application Manager (SAM)

1998 - In 1998 I presented an idea for general service management to Ericsson Ireleand. They funded to work a couple of months on working out this idea. Normally, all service providers use their own interface to their customers. Though this has certain advantages for the providers, it is awkward and difficult for the end users. Much redundant has to be entered over and over again. SAM was a model where service provider could host their end user applications in a trusted environment that was controlled by a trusted party. Service providers were able to communicate with their users but could use shared information. This project resulted in a working prototype that was presented to management. No follow up was done due to too much activities on my side.

Ericsson Research, the Point to Point Invention

1998 - During our many brainstorm sessions in Stockholm we came up with a very interesting idea. A device that could be plugged in anywhere in the net and regarldess of its position it would be able to find its twin. For example, a baby phone could be plugged into the Internet in Stockholm and its twin in Gothenburg. We also played with the idea of having Lego blocks that would be able to communicate worldwide with their twins. You can think up all kinds of interesting scenarios. I got the task to work out this scheme. I developed lego blocks that had a sensor and a LED. They had a little 8-bit, 4-pin CPU (PIC 12F508) that was connected with 2-wires to a PC that carried both the data and power. The PC was connected on the internet. Each PIC processor had its own unique ID, and the ID of its twin. The could therefore find eachother through a shared directory server. Ericsson obtained a patent on this idea but unfortunately has so far not done anything with it.

Ericsson Wireless Internet unit, OSGi specifications

1999-2001 - The OSGi specifications took a great deal of time to develop. I had to travel every 4 weeks to somewhere in Europe or the US. First specification was delivered in 2000.