What's New in Rust's "Friday Devblog 10" Update

Friday Devblog 10 went up at the end of May 2014, back when Rust was a legacy game being rebuilt in public on the experimental branch. Garry opened it by admitting this was one of those weeks where he asked himself what he'd actually been doing - and by his own account there were almost no gameplay changes at all. What there was, though, was a set of infrastructure decisions that shaped the game for years: the UI moving wholesale to HTML, a completely different philosophy on error handling, and Easy Anti-Cheat arriving on experimental.

Moving the UI to HTML

The big one. All of Rust's UI was being changed to use an HTML renderer, and the reasoning was laid out in detail. It runs in a completely separate process, which instantly offsets all UI work onto another thread. It's webkit/chrome, so anything possible in a browser is possible in the UI. The UI was due for an overhaul anyway, so it was the right moment to switch.

Beyond the technical case there was a practical one: HTML is far more developer-friendly, and far more people can edit it. HTML has already solved every problem you're likely to hit - want tables, use tables; want syntax-highlighted code, use a syntax highlighter. It enforces an MVC pattern, makes it trivially easy for people to build their own skins (if Facepunch chose to allow it, which they expected to), and makes localization straightforward.

The only arguments against were that it might cost a couple of extra days and that it might be harder to port to consoles, if they ever chose to. At the time of writing the menu system and some in-game elements were already running on it, with the hope that the entire GUI system would be by the following week.

Error reporting

The old codebase's philosophy was that code will always fail somewhere, so catch everything. Garry gave the numbers: searching the old codebase for "try" returned 1275 matches; searching the new one returned 11.

The reasoning came from a C++ background - if code is doing something wrong, it should crash. Errors shouldn't be tolerated and shouldn't happen, because they're errors. The practical result is that when Rust throws an error it's now reported to Sentry, the team gets notified, and it gets fixed.

Easy Anti-Cheat

Easy Anti-Cheat went live on the experimental build. The long-standing decision had been to pay a third party to handle cheating so the team could concentrate on making the game rather than fighting people ruining it, with as little friction for players as possible.

Garry was upfront about the unavoidable part: you have to install a program. Anti-cheat needs elevated permissions and needs to look at all running processes - that's just how it works. But the install is as simple as clicking "yes" when prompted. Once installed it doesn't run continuously; the service starts when the game starts and stops when the game stops, using less than 1MB of memory while running and nothing at all when it isn't.

Grass and environment art

Andre brought grass back, on a system faster and more versatile than both Unity's and Rust's old one - fast enough that there wasn't even a way to turn it off. Petur and Bill continued on environment art, experimenting with different biomes and smaller detail work, while Alex spent the week fixing the character rig he wasn't happy with.

Concepts and animation

Xavier's concept work got a defence in the post: concepts should push boundaries and provoke a reaction, and a concept is a concept - it doesn't mean it's going to make it into the game. Minh (Goosey) was animating, and Dan and Tom's running contest over who gets more content into the Friday update went to Tom this week, though he was denied a trophy on the grounds that three months spent on the rocket launcher had ruined his average.

Where things were heading

The closing section was candid about the lack of progress and set the agenda for the following weeks. Building was the obvious gap, and Garry planned to focus on it. The team weren't fans of how the existing system had turned out - players carrying walls and foundations around wasn't great and wasn't sustainable - and it was something they'd been turning over for about six months. The plan was to spend the coming week experimenting with other approaches.

The other acknowledged gap was AI, which was going to Andre on the grounds that he was even more enthusiastic about it than he was about terrain. Also mentioned in passing: The Forest had just launched on Steam, and Garry recommended it to anyone who liked Rust.