Maccy 3 Plan

[p0deje]

Hey everyone,

Maccy 2.0 was released two years ago and brought a complete rewrite using SwiftData and SwiftUI. That came with its own challenges, particularly around memory management, but turned out to be a good move. It fixed lots of AppKit-specific issues and unblocked tons of features that weren't possible before (e.g., scrollable history, a slide-out preview pane, drag & drop, inline editing of items, etc.). Check out #790 if you want to learn more. At this point, I would consider Maccy 2.x stable.

The memory issues didn't go away, though a lot of optimizations have been made. Hats off to @weisJ for working on that 🫡. However, it's not possible to fully solve these problems without major changes to the storage layer (e.g., external storage of blobs and item fingerprints). These changes won't be backwards compatible, so I can't make them in 2.x and need to increment the major version.

With that said, here's a rough path for Maccy 3:

1. Merge some of the PRs I've been planning to merge, then release a final 2.x version.
2. Freeze the current master branch and stop releasing new 2.x versions unless there are critical bugs.
3. Focus on a new 3.x branch and start doing beta releases.

My current plan for 3.0 includes the following, though it may change:

• As mentioned above, memory improvements and a redesign of the storage layer
• iCloud support
• An iOS companion app, syncing data via iCloud
• A stdio MCP server
• Switching to a different Apple account for signing the application
• Apple Silicon only as Apple ends Intel support

I want to clarify the final point a bit more. Historically, Maccy has been distributed through both GitHub and the App Store. The two versions have always been functionally similar, but they were signed using different Apple Developer accounts. For 3.0, I'd like to start using the App Store account to sign OSS releases as well, which would allow me to use the same iCloud container for data sync. Otherwise, iCloud sync wouldn't work seamlessly between the GitHub and App Store versions. I don't expect any issues when upgrading to 3.x, but I'll need to test that properly.

One other thing I'd like to call out is the number of issues and PRs in the repository, which has grown a lot over the last year. In the past, I did my best to keep the number of issues under 25 so they would all fit on a single GitHub page. But with a full-time job, other open source projects I work on, and AI making it extremely easy to file bugs and submit PRs, none of the maintainers can keep up with everything.

Every bug report and PR needs to be checked, but their trustworthiness is very low. There are, for example, lots of PRs that aim to improve memory usage. Some of them contain genuinely good ideas, but lack proper testing and are written poorly enough that it's easier for me to rewrite them myself than to work with the author (or, more frequently, the author's agent) to fix them.

I'm not taking it as far as completely disabling external PRs like Sindre has, at least not yet, but I will experiment with automatically marking issues and PRs as AI-generated. Those will be a lower priority for me, as I'm already struggling to find enough time to review genuine human work. This doesn't mean AI-assisted PRs are bad, as #1448 demonstrates. I just don't want to spend time reviewing superficial work, which, in my experience, makes up the majority of fully AI-generated PRs.

[jaspermayone]

Can we get my pr (unlimited history) in for v3?

[p0deje]

Potentially, it would be a nice feature to have, but it needs thorough testing as the memory issues are real and I don't want to give users a footgun.

[jaspermayone]

so true @p0deje! I'll take another stab at stuff tonight but I do think I addressed some of the issues.