The project is mostly working, so it's time to rewrite it 3 or 4 times.

My recording project has a mascot character now.
More projects need mascot characters.

Away for the weekend, but there's always time for testing.

Making some progress. Two here are on the same network, one is on 5G tethering - no ports forwarded. Relaying the UDP traffic correctly has been a painful experience, but necessary for matching chat and pacemaker graphs to work.

It would be nice if IIDX ARENA was more accessible. I suppose that'll be my next project?

Spent most of the weekend rewriting some card reading code that everything else in the project relies on - all to support the new cardless QR login feature that I bet nobody is going to use. A little nervous about this release... but fingers crossed!

All done now. No more on-site maintenance for a while... I hope.

One month already.

Begrudgingly working on a project that I've been trying to avoid, but I suppose you can't ignore errors like this forever. Backing up as much as I can remotely (since Monday, 1.36TB so far, two more days to go) then swapping the drive on Friday.

Only a few years late, but I finally got around to finding the real values in memory. Some more porting and testing to be done before it'll make its way into a release. Won't make a difference to any users of the site, but it helps me sleep at night.

The correct entry IDs here should actually be 25007, 18032, 16020, and 12004. Fortunately, the endpoint on my score tracker for handling class course results only looked at the course ID before validating against its own set of known charts, so it didn't actually cause any issues.

Guess there was no point in keeping that up, so everything from 9th Dan onwards in IIDX 32+ is just 5.1.1. (N). Anyway, getting back to the point of this post, I started my score tracker project before this was a thing, so I had just been copying from the local file when saving class results.

Moving forward to IIDX 31, where the real courses were later set network-side, they had some fun with the local file and included some fake courses instead. Single Kaiden here was TITANS RETURN (L), SAMURAI-Scramble (L), n/a (A), and I suppose it'd be too obvious if it didn't end with 冥 (A).

Ever since IIDX 31, class courses can be overriden using values provided by the network, presumably in case they needed to make corrections without pushing an OTA update, but also to prevent people from leaking the locked courses ahead of time. Here's Kaiden from the IIDX 30 local course data file.

This is what the chapter appending stuff looks like. The events from each cabinet are stored on disk for a few hours, so newly created recordings just need to query the last 5 minutes of events and remap the relevant ones into chapters.

GitHub - aixxe/retrack: Redis event publishing hook for BEMANI games Redis event publishing hook for BEMANI games. Contribute to aixxe/retrack development by creating an account on GitHub.

Released retrack, a component of REPLAY. It publishes network messages and scene transitions from BEMANI games to Redis. Currently only used for appending chapters to videos and auto-submitting scores to events. Lots more that can be done with it, but I don't know if I'll ever get around to it.

I hope things get better soon.

Back from a much needed break.

Moving on.

GitHub - aixxe/eacproxy: Eacnet network interception framework Eacnet network interception framework. Contribute to aixxe/eacproxy development by creating an account on GitHub.

Released a project at the end of 2025 for intercepting INFINITAS network requests. I've been using it for years to get my arcade scores to show up in-game alongside my home ones. The example that injects a Python interpreter into the game to reverse your DJ name was my favorite to write.

Quick update to this post: we've exceeded our entrant record by far with over a week to go. Thank you to everyone for participating!

Regardless of how it turns out, I'm proud of everything that has gone into it. Extremely grateful to all those who have helped out with ideas behind the scenes, and of course, everyone participating in the event itself. I'll be around grinding for a good score today.

These are just the backend changes too. I'd love to write another thread in the future about the suite of Python scripts I wrote to streamline creation of the final video that ends up on YouTube. But it's getting late, so perhaps another time.

I also found out that it's possible to edit webhook messages after they've been posted, so the event leaderboard is now contained in a separate Discord channel, which updates automatically whenever a new submission is posted.

You don't even need to be at the arcade to register, either. While the recording site is still local intranet only, I've built a separate mini-site running from the same codebase solely for handling event participation, which is publicly accessible through cloudflared.

Now, after a one-time linking step with the network web UI, everything for the player is completely automated. Once you're signed up to the event, you only need to play the chart and the rest is handled automatically - including score improvements. No interaction with REPLAY necessary.

REPLAY, the aforementioned recording system, has seen a lot of changes since then. Most notably, the network and game scene introspection features, which opened up a lot of possibilities. Embedded chapters in videos was only scratching the surface. It was possible to fully automate event submission.

Outside of those usability issues, there were plenty of other areas that could've used some serious work. As participation started to fall off, I decided to shelve the whole thing until they could all be addressed in a way I'd be happy with. I feel like we're finally there now.