Changes

I'm going to write a bit about web services -- which is the only technically noteworthy thing I've been doing since my last entry. But first, I'm going to record the events of the past couple of months. After all, this is supposed to be a chronicle of my experiences on this project.

We've had three departures from our team: one voluntary, one involuntary, and one reluctant. The reluctant departure was that of a valued colleague and a good programmer, who fortunately will still be available to me on a contract basis.

We've more-or-less committed to rejecting the financial and project accounting software that we were initially going to use in favour of a simpler package. The theory is that the simpler package will allow for a faster implementation; the more complex package, we've decided, isn't really required at this point since we've narrowed the scope of our project a bit: we're no longer aiming for a system that can be rolled out globally, but simply one that will handle the North American chunk of our major line of business. In other words, about 3,500 employees instead of 26,000. Right now, we're in the middle of three weeks' worth of meetings whose output will be a "model company" around which the system will be designed. Thankfully, I have no role in said meetings.

My own role in the project is a little vague right now. The main part of my job is application integration: defining the interactions between the various applications (HR, finance, project accounting, and project management) that comprise the overall system, and developing interfaces between them. I'm also consulting on the technical architecture. It's still pretty early in the implementation, so there's little actual work to be done: my team is down to myself plus one. At this point I'm investigating technologies, developing spike solutions, and thinking about business cases for solutions that my gut tells me are the right ones.

Within the next couple of weeks, I'll start requirements-gathering on the interfaces. I think the interfaces we require will be comparable to those we developed for the first (aborted) implementation. Hopefully I'll be able to get about 80% reuse out of the existing documentation.