GeneRally 2 Update (Q1 2018)

It’s been ages since the last update on the progress of GeneRally 2. As we do realize that the total radio-silence gives way to rumours, here are some musings about the current state of the development, as well as reasons behind the lack of updates.

Even though there’s been lots of progress, at least from our perspective, the current state is a bit of a mess by “software development project” standards, but fortunately/unfortunately, this isn’t a normal project. It really ties in to why we’re incredibly terrible at writing posts about progress – because I think we’re now on maybe the 6th or 7th iteration of the track loader/importer; the 4th or 5th version of the car physics; the 3rd lighting model; maybe the 4th or 5th implementation of the way skidmarks are generated (it has a surprisingly high performance impact); etc. We’re rewriting so many things, just because the pace of development is slow… something gets out-dated, or performs really badly when integrated with something else (keep in mind, we’re aiming for really low specs again), and the cycle repeats.

If you know you’re going to have 16GB of RAM, 4GB of VRAM, and a 4GHz processor to throw at your game, you can do whatever you like in almost any way you like – but we’re trying to keep to the mantra of GR being comfortably playable on older machines. Probably some of this falls into the realm of premature optimisation – but both of us find these iterations quite fun, which leads to us getting drawn into them perhaps a little too often. This is something we’ve realised recently, and we’ve made some changes to ensure we don’t spend all our time rewriting things we already did.

I’d say we’re not a huge distance from something broadly playable – but the time estimate for actually getting from “okay, everything’s a mess but we’re getting there” to “playable” is just not something I want to hazard a guess at. We’ve missed pretty much every estimate we’ve set so far, so it’s clearly not something we’re great at (and somewhat normal, considering the amount of free time really fluctuates).

Why don’t you just make more, small update posts?
Because it’s really just not fun for us. That’s not a good reason, it’s just the reason it’s been the case so far. I guess this is something that can be seen from two perspectives: posting updates about rewriting things for the nth time just leads to dissatisfaction and complaints about not making progress; and not posting updates about rewriting things for the nth time just leads to dissatisfaction and complaints about not making progress and the lack of information.

Neither Markku nor I are really the social media types – and we’re both very private individuals – so sharing “this is what I did today” is not really something that comes naturally. Most of the blog posts we’ve written have taken hours upon hours to write, tweak and perfect – so it feels like when you sit down one weekend to make some progress that your choice is: “I can improve the way the AI works” or “I can write about how little progress we’ve made.” I think you can guess which wins time and again.

Fortunately, things are about to change as someone has come forward with a really good suggestion about improving updates with little-to-no effort required from myself and Markku on that front. In fact, he’s already written his introductory blog post, which will be coming in the following weeks. So, dear readers, you can read this as a promise – it won’t be another 18 months before the next update.

– James