Justin Mason

Iona Technologies was one year old in 1992. There were several researchers working for the company on consulting projects and on EU programmes. It was already working on the middleware product that it launched as Orbix in June 1993. But Iona did not have any full time staff until it hired me and I became employee number one.

I had heard about this job through the Student Computing Research Group in Trinity College Dublin (TCD). I had started a degree course in mathematics at Trinity in 1989, but dropped out after the first year. I wasn’t particularly academically minded at the time, but I was very keen on computing. I continued to hang around the Maths Society and to mix with students who were interested in writing software and in developing communication tools. Two years later I was the lucky candidate who was free to join Iona.

The company was still based in single room on the ground floor of a TCD building in Westland Row. Another campus company, internet service provider IEunet, had the office upstairs. A coax cable between the two floors gave Iona an Ethernet connection to the internet.

My first project was to write a printer driver for SunOS, contracting on behalf of Iona. Then I became a kind of sysadmin. Iona ran high end Sun workstations with graphical user interfaces. That was a big change for me from the text-based interfaces in the maths department.

I learned most of my tricks, and picked up new software, via the USENET news feed that came into the maths department news server. USENET was the primary means of international forum-based communication with like-minded techies and academics in the early 1990s. Most of the original team at Iona read technical newsgroups like comp.object or comp.lang.c++ on USENET. I experimented with running our own news server for a while, but mostly we used the servers in the TCD School of Mathematics or in the School of Computer Science.

The internet connection allowed us to download the software that we discovered through USENET more easily. There had previously been FTP-by-mail services, where people could receive software in multiple email messages and then laboriously reintegrate the pieces. Now, however, we could request software directly from FTP servers.

There were a lot of hypertext software products around in 1992, but they were very proprietary and users had to pay for them. The World Wide Web software, on the other hand, came as free-to-use hypertext systems. Open source software with open protocols looked like a revolution.

I tried out a number of web browsers, including Mosaic which was announced on USENET in early 1993. The first version I got hold of was Mosaic 0.9. There wasn’t a lot of content to look at on the WWW at that time, but the format worked. You could visit a site in America and then move seamlessly to a site in Switzerland with a single click. That was something really new.

Most the sites that existed then were in universities and research centres, but there were also a few in commercial organisations. These were mainly in the US and most of them were companies that had academic links. Iona Technologies was that type of company. I set up an Iona web server in July 1993. It was approximately the 70th HTTP server in the world and the first non-academic web server in Ireland.

To begin with, I just built HTML pages for my own use. I showed them to Annrai O’Toole, who was Iona’s vice president of development and one of its three founders, and to the company’s business manager Colin Newman. Then Colin and I put together a page for the company and its Orbix object request broker, which Iona had launched in June at the Object World trade show in San Francisco. The HTML code was hand-edited and really basic. The content was mainly text and the first image on the page was the company logo.

I have no idea who accessed that page. We didn’t have any analytics back then. But it was probably seen by other members of the OMG and by readers of the comp.object newsgroup on object-oriented software, along with whoever was lucky enough to have access to just click through the limited number of sites back then.

This artwork assisted users to navigate around Iona’s web pages. See https://www.iona.ie/ for reconstructed versions of the site as it looked in 1993 and 1997.

The initial site was the Plexus Perl HTTP daemon, running on a Sun SparcStation IPX, which I used as both a workstation and as the company’s main server. Perl had features for adding interactive elements to a page. We didn’t see any particularly heavy load, thankfully. It was not long before we replaced this with a more powerful SparcStation that also had other fileserver duties. I don’t think we used a dedicated web server machine until we’d moved away from Trinity in 1994. Sun’s SunSoft subsidiary purchased a minority stake in Iona at the start of that year and provided us with space at its premises in Percy Place.

The early versions of the site did not just display company information. There were also links to the Trinity web servers, to other Irish universities and to sites around the world with content that we liked.

Colin took over editing the site later in 1993. Simon Boyle joined the company in summer 1994 and took over from Colin. But I was still involved with the web activity as sysadmin and kept on downloading new web tools.

I felt that the Mosaic browser was super-successful. However, when its developers released Netscape Navigator as a commercial version in December 1994, it was way more polished and clearly going to become a mass market product.

I got into writing Java after (again) discovering it via newsgroup traffic at around the time of the first beta release. I was intrigued by the idea of an interpreted object-oriented language which could be used to add interactivity to web pages. I’d been looking for a way to display some network management stats interactively, so I wrote an applet to display them in a web page.

Open source was another interesting area in 1994. I took over the maintenance of a piece of open source printer software: the Public Line Printer (PLP) spooler system. At the time Iona had a diverse set of computers around the office for porting the Orbix software to the many different Unix platforms it was sold on. Getting all of those variants to successfully support interoperable printing was a big challenge, so I did what I normally did in this situation. I found an open source implementation of the Unix printing software to use instead of the default implementations. Unfortunately, PLP had a few rough edges and bugs. I fixed a few of them myself, found a few patches around the internet, and collated the results together. I also added support for platforms that the existing version did not cover and then republished the new and improved version. It gained popularity quite quickly.

I set up a mailing list for users of the product, and lots of users appeared, including McMurdo Base in the Antarctic! The original author of the PLP software also got in touch, and proposed collaboration on a paper for the 1995 USENIX large installation system administration conference. We put this together, renaming the software from PLP to LPRng in the process, and delivered the paper in Monterey in 1995.

This was a fantastic introduction for me to open source software development and online publishing. I was grateful that Iona allowed me to hack away on PLP while I should really have been concentrating on the day job. It was very exciting, as a solo software developer, to have my software used in high profile installations and to deliver papers at conferences in California. This project got me hooked on open source for many years.

In 1995 I joined Iona’s OrbixWeb team, which wanted to implement CORBA with Java. Paddy Benson led the project and there were about ten people working in the group. It was clear that a CORBA implementation in Java was going to be a useful feature – a way to expose information in a Java applet that CORBA could access from a remote server. Iona had big customers that wanted this.

My initial focus in the OrbixWeb team was to write a firewall proxy for Iona’s CORBA products. I’d learned about internet firewalls while attending the USENIX LISA’95 conference earlier in the year, and it made sense that customers who wished to safely expose parts of their internal systems data over CORBA to Java applets running in web browsers on the external internet, would need a safe way to do so, so a firewall proxy, to safely filter and sanitise this traffic, made sense. Annrai O’Toole was going through an Oasis phase at the time and decided to name the product “Wonderwall”, unfortunately.

After the release of the Wonderwall product, it became clear that a lot of internet firewall vendors were not all that keen on supporting a new internet protocol for CORBA-based applications; in effect, the HTTP protocol used by the web had already “won”. We came up with a way to deliver CORBA requests “tunnelled” over HTTP, adapted Wonderwall to perform this role and, in effect, HTTP became “the new TCP/IP” – applications had their own protocols, but they were written on top of this protocol layer.

Our main competitor for OrbixWeb at the time was Silicon Valley-based startup Visigenic. It developed a CORBA implementation in Java called Visibroker and we watched closely to see how it was coming along. Iona released OrbixWeb in May 1996. The product was widely used and I think quite successful as part of the whole Orbix suite of products.

During this time, Iona had an expanding team of engineers working on Orbix products of all varieties, covering systems from mainframes to Java, and was outgrowing its offices. We moved to premises in Lower Pembroke Street in 1996 and, again, into a new four-storey building on Shelbourne Road in 1998.

I left Iona Technologies in 1999. Along with Paddy Benson, l joined Netnote International in Shankill. The company was developing a low-cost smart card-based device – Webnote – that aimed to make internet access easier and a lot more affordable.

In theory the appliance would allow home users to browse the World Wide Web and to use a web-based email account for a much lower price than running a regular computer. In practice we had a hard time keeping the production costs in the desired range while ensuring the usability of the device. Instead, we diverged into building pay-as-you-go internet kiosk systems.

After NetNote, my next project was SpamAssassin. a Perl-based email filtering application that I released as open source in 2001. Three years later SpamAssassin became an Apache Software Foundation project and in the following decades it protected some 100 million email users around the world.

Last edit: August 2023

© Justin Mason 2023