I’ve been working on NOVA because I’m genuinely exhausted by the current state of decentralized software. We’re building "distributed" tools that require 500MB of RAM just to render a search bar. It feels like we’ve gone backward.
NOVA is my attempt at a "Clean Room" implementation. It’s raw C++17. No runtimes, no heavy frameworks, just Win32 and GDI. I’m currently at v0.3 and using a terminal client called Onyx as the reference.
The hard parts:
Stripping the XP "Luna" theme to force a proper Whistler/Classic aesthetic without losing modern high-DPI support.
Using DeferWindowPos to handle atomic UI updates so it doesn't flicker like a 90s VB6 app.
Keeping the binary under 100KB while still supporting the metadata pipeline.
It’s definitely opinionated and probably "too low-level" for some, but I wanted to prove that you can still build extremely dense, high-performance tools that don't treat the user's CPU like an open buffet.
Curious what you guys think about the metadata fetch logic in the repo.
Option 2: The "Technical Deep Dive" (The Specialist)
Show HN: NOVA – A C++17 framework for decentralized data
I just pushed v0.3 of NOVA. I've spent the last few weeks specifically on the GDI owner-draw logic and the metadata pipeline.
The goal was to build a decentralized engine that behaves like a "Data Terminal" rather than a webpage. I went with a statically linked Win32 architecture specifically for the "xcopy-deployable" aspect—I want this to run instantly on any machine without a setup wizard or a dozen DLL dependencies.
What's inside:
A custom layout engine that avoids the usual Win32 resize-flicker.
A 5-step signal ramp drawn directly in the status bar partition via WM_DRAWITEM.
Sub-10MB idle memory footprint.
If you're into native Win32/C++ or low-latency P2P, I'd love some eyes on how I'm handling the thread-safe UI updates from the metadata streams.
I’ve been working on NOVA because I’m genuinely exhausted by the current state of decentralized software. We’re building "distributed" tools that require 500MB of RAM just to render a search bar. It feels like we’ve gone backward.
NOVA is my attempt at a "Clean Room" implementation. It’s raw C++17. No runtimes, no heavy frameworks, just Win32 and GDI. I’m currently at v0.3 and using a terminal client called Onyx as the reference.
The hard parts:
Stripping the XP "Luna" theme to force a proper Whistler/Classic aesthetic without losing modern high-DPI support.
Using DeferWindowPos to handle atomic UI updates so it doesn't flicker like a 90s VB6 app.
Keeping the binary under 100KB while still supporting the metadata pipeline.
It’s definitely opinionated and probably "too low-level" for some, but I wanted to prove that you can still build extremely dense, high-performance tools that don't treat the user's CPU like an open buffet.
Curious what you guys think about the metadata fetch logic in the repo.
Option 2: The "Technical Deep Dive" (The Specialist) Show HN: NOVA – A C++17 framework for decentralized data
I just pushed v0.3 of NOVA. I've spent the last few weeks specifically on the GDI owner-draw logic and the metadata pipeline.
The goal was to build a decentralized engine that behaves like a "Data Terminal" rather than a webpage. I went with a statically linked Win32 architecture specifically for the "xcopy-deployable" aspect—I want this to run instantly on any machine without a setup wizard or a dozen DLL dependencies.
What's inside:
A custom layout engine that avoids the usual Win32 resize-flicker.
A 5-step signal ramp drawn directly in the status bar partition via WM_DRAWITEM.
Sub-10MB idle memory footprint.
If you're into native Win32/C++ or low-latency P2P, I'd love some eyes on how I'm handling the thread-safe UI updates from the metadata streams.