Over the years, I have had several projects that remained unfinished, undeployed. The reasons varied: business decisions, technical merit, you name it. But darn it, the work I have done on at least some of these projects makes me feel proud, their ultimate lack of success notwithstanding.

Recently, I brought some of the Web-based projects back to life. These prototypes no longer worked, mainly because the PHP library that they used to access a MySQL database was deprecated for security reasons. Replacing it with a modern alternative is straightforward, but it had to be done "by hand", to avoid subtle traps.

Humidification systems - product catalog and load sizing

In the 1990s, I developed a comprehensive product catalog and load sizing application for a local manufacturer or large humidification equipment for HVAC systems. Then in the late 2000s, I was asked to build a Web-based replacement. For reasons I never quite understood, this project came to a halt after most of the development work was completed.

The software, though not fully debugged, mostly works. And it still looks nice, I think. The product catalog speaks for itself.

The load sizing component computed the humidification load of the system in various configurations, warning the user of pitfalls, such as humidification exceeding 100% that would result in condensation. It also led directly to a product selector, allowing the user to pick the right combination of equipment for their application.

The project is approximately 27,000 lines of source code (HTML+JavaScript, utilizing the Dojo library for front-end UI; PHP with MySQL; some back-end shell scripts.)

Medical device registration service

A couple of years later, I was asked to develop a prototype for a fictitious Canadian medical device registration system for licensed medical devices. I think the idea was that our tech might be sold to a foreign government. The sale didn't materialize, but before the project fizzled out, I did develop a fairly comprehensive system, complete with a registration database.

The software also produced printable PDF certificates, complete wit QR code.

The QR code resolved to a URL that permitted real-time verification of the license. As I recall, this feature was specifically suggested because apparently, some unscrupulous shippers used medical device registration documents for entirely different equipment in attempts to deceive customs authorities.

This prototype was built with approximately 4,000 lines of HTML+CSS+JS source code, with a PHP+MySQL backend.

LTE small network planner

Several years later I was tasked with another project, part research, part a possible production application, a planner for LTE networks. The idea was to plan for, and optimize, locations of transmission towers to build reliable, efficient LTE mobile coverage. The software had the ability to carry out detailed analysis.

It could also produce printed PDF reports on demand.

This project consists of just over 5,000 lines of code, mostly HTML+CSS+JavaScript, a PHP+MySQL backend, along with some shell scripts.

RF spectrum management

The there was PRISM. I gave it that name: Program for Remote Integrated Spectrum Monitoring. It was supposed to be a Web-based alternative for software I developed, and maintained, for Industry Canada over a period spanning almost 20 years. I thought that with a working prototype I'll stand a chance: perhaps Industry Canada might be interested, perhaps one of their foreign partners, responsible for radio frequency spectrum management in their respective countries. Well, I thought wrong. Still, the software I developed, even in its prototype form... what can I say, it speaks for itself I think. And yes, it still works.

This prototype had a front-end built using mostly HTML+CSS+JavaScript, with the back-end code using C#; a total of approximately 4,000 lines of code.

Litigation analysis

How about this beautiful project? Utilizing a user interface that guided the user through an in-depth interview, the software estimated the likely outcome of litigation. I developed this prototype for an overseas client back in the mid-2010s. Eventually, they decided against launching the project. I have no doubt that they made the right business decision but that does not mean I don't feel a bit sad that this solution never "made it to market". I mean, come on, just look at this summary page at the conclusion of the interview and analysis.

This prototype was implemented using about 9,000 code lines; much of it was in HTML+CSS+JavaScript with a PHP+MySQL back-end, but a significant chunk of the logic was, in fact, implemented in the form of stored formulas in the database proper.

Package management

This nearly complete project was meant to be a Web-based service, managing package uploads and subscriptions. The packages could be anything: collections of files, software distribution, whatever users desired to share (within reason and within applicable legal constraints, of course.)

This project, more than anything, made me realize what an enormous gap exists between developing software and providing a service on the InterWebs, and also the importance of making my clients realize this and understand what it takes to operate a revenue-generating enterprise on the Web.

The software, amounting to some 20,000 lines of code, was written mostly in HTML+CSS+JavaScript, with a PHP+MySQL back-end; some external components, including code managing cloud-based storage and the payment processor, were written in other languages.

Boulder Dash

Back in 1993, I was inspired to reimplement the classic game, Boulder Dash, for Windows, using Microsoft's then still fairly new Visual C++ development environment.

Unfortunately the title is subject to a trade mark. The rights holder was not terribly interested and in any case, they criticized the fact that in my implementation, the entire play area is visible at all times instead of just a scrolling view. So the project remained unfinished, but it was still loads of fun to write, and play. A few years later, I revisited the implementation and made sure it compiles on 32-bit platforms, which is why I can still run the executable on my current Windows 10 desktop computer.

These are just some of my "forgotten projects": projects I worked on that never quite made it to "production", for whatever reason. Is it inappropriate for me to take pride in projects like that? I hope the screenshots above tell otherwise: Whatever the reasons were for these projects' lack of commercial success, the stuff I built stood the test of time, and I am still happy to showcase my work.

As a side note, I also feel compelled to observe just how difficult it is to translate an idea that has robust technical merit, into a viable public-facing, revenue-generating service. Operational issues, customer support, payment processor integration, system security and monitoring... the demands, not to mention the costs, of these aspects often exceeds (significantly) the costs of initial development.