RFC: Technical Architecture - Tauri + Rust + React

[cuevaio]

Request for Comments: Technical Architecture

We've drafted a comprehensive technical plan for Open ScreenStudio and want your input before we start building!

Read the full plan: docs/TECHNICAL_PLAN.md

---

Summary of Key Decisions

Decision	Choice	Rationale
Framework	Tauri 2.0	Small bundles (~10MB), excellent performance, Rust backend
Backend	Rust	Performance for video processing, native API access, memory safety
Frontend	React + TypeScript	Largest ecosystem, most contributors familiar with it
Styling	Tailwind CSS	Rapid development, consistent design
State	Zustand	Lightweight, TypeScript-friendly
Min macOS	13.0 (Ventura)	Modern ScreenCaptureKit APIs
Min Windows	11	Stable Windows.Graphics.Capture

---

We Need Your Input On

1. Stack Choice: Tauri vs Electron

We chose Tauri for performance and bundle size, but Electron has a larger contributor pool.

What do you think?

• Is Rust a barrier to contribution?
• Is the performance benefit worth the learning curve?

2. Platform Support

We're targeting macOS 13+ and Windows 11 only.

Questions:

• Is this too restrictive?
• Should we support older versions even if it adds complexity?
• Linux support - how important is it for v1?

3. Project File Format

Options:

• A) JSON + assets folder (human-readable, debuggable)
• B) Custom binary format (smaller, faster)
• C) SQLite database (queryable)

Which do you prefer and why?

4. Recording Storage Strategy

Options:

• A) Record to memory, save on stop (simple, RAM-limited)
• B) Record directly to disk (handles long recordings)
• C) Hybrid approach

What's your experience with similar tools?

5. Implementation Priorities

The plan outlines a 14-week roadmap. Should we adjust priorities?

Current order:

1. Screen capture
2. Audio capture
3. Cursor tracking
4. Automatic zoom
5. Cursor smoothing
6. Basic editor
7. Styling
8. Export

---

How to Participate

1. Comment below with your thoughts on any of the above
2. React with emoji to show support for others' ideas
3. Open separate issues for detailed technical proposals

---

Timeline

We'd like to finalize the architecture within 1-2 weeks so we can start building. Please share your thoughts soon!

---

Related Documents:

• VISION.md - Why we're building this
• FEATURES.md - Feature specifications
• ROADMAP.md - Development phases

Looking forward to your feedback! 🚀

[Enriquefft]

Linux is a must

[cuevaio]

look a this screen studio project format: https://github.com/crafter-station/ship-project-screenstudio/tree/main/recording

[DaveLee-fun]

Nice RFC. I record coding tutorials and have used several tools in this space, so wanted to share some feedback on the priorities:

On #5 (Implementation Priorities):

From a user perspective, I think cursor tracking + smoothing should come before automatic zoom. Here's why: auto-zoom is the hardest part to get right and is where ScreenStudio/FocuSee get the most complaints. The auto-zoom often zooms into the wrong thing, or the timing feels off. If the zoom is bad, people turn it off entirely.

One alternative approach worth looking at: instead of auto-zoom in post-production, some tools take the "live zoom" route where the user controls zoom manually during recording. ZoomShot on macOS does this, for example. The user presses a hotkey + scroll to zoom in/out in real-time, and it renders as a window overlay that the recorder captures. It trades automatic behavior for user control, which avoids the "zoomed into the wrong thing" problem entirely.

For open-screenstudio, you could support both: a basic manual zoom during recording (low effort, high value) and then tackle auto-zoom as a polish feature later.

On #1 (Tauri vs Electron):

Tauri is the right call for video processing workloads. The memory difference matters when you're processing 1080p/4K frames. Rust's performance for the video pipeline is worth the contributor learning curve.

On #4 (Recording Storage):

Definitely go with B (direct to disk). Any tutorial longer than 5-10 minutes will eat through RAM fast, especially at high resolution. I've had OBS running for 2+ hour recording sessions and disk-based storage is the only thing that scales.

Hope this is useful. Looking forward to trying the first release!