#vibejam update day 11!
Another challenging day, but good progress.
Holomap is now full 3d.
The big issue was rotation and this was a conceptual hurdle for me. In the original implementation I was pushing forward, we would clip the planets out when zooming/rotating. This made it frustrating to zoom at all because you couldn't rotate to see more of the battlefield - the planets just weren't there to see.
The end point was a second scene being viewed through a parallel rig, which got trapped as depth in the table, and eventually we got free.
Still a lot to go, but this is the centerpiece for playing this game so it's got to feel good. Next will be making selection cleaner and billboarding of text/flags. Then it's moving the combat into the 3D plane.
Planned features:
- Visual pass of the holo table. Better models, found a nice holographic shader.
- Tighten up combat, make it more visual and interesting.
- Add ship "postures" and more expressive AI for the "command at high level" feel.
- Ship Captain dialog and reports, and eventually, requests to change orders which the commander can approve or deny (I cannot wait for my guys to panic at the sight of a capital ship and request to retreat only to tell them "belay that!")
Tone@InOneProjects
#vibeJam update Day 10! Yesterday precipitated an app wide refactor and serious adjustment toward better quality code that's easier for both me and the agents to reason about. Implemented ESLint (I don't know why I waited) based on the advice in this excellent article: @albro/eslint-as-ai-guardrails-the-rules-that-make-ai-code-readable-8899c71d3446" target="_blank" rel="nofollow noopener">medium.com/@albro/eslint-…
All files now below 500 LoC, params are simpler or have options objects, constants are cleaned up, scenes broken up into logical chunks, dead code removed, etc. Process: - Claude Opus started things off with a plan for the major refactor in phases. I let it run until my Pro plan ran out of usage over night and spent about $6 extra before it finished phase 1, which was a huge lift. - Over the course of the day I nudged Codex to implement each phase, and as I got more impatient I had it do so with subagents as an orchestratot. This level of refactor would have taken six months in my work paradigm, and realistically would never have been prioritized. Result - the task I could not get done last night is NEARLY done, but needs some polish but my fiancee is complaining I don't spend time with her anymore because of this game. She's right. I'm having so much fun. So no screens of half done stuff tonight. Don't sleep on ESLint in your project. Here's a pic of GPT 5.4 codex max thinking but as a cute puppy English
