npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

qipgui

v0.2.1

Published

Prototype: run QIP GUI components in terminals with the Kitty graphics protocol

Readme

qipgui (prototype)

qipgui runs one QIP GUI component in a terminal that supports the Kitty graphics protocol and SGR pixel mouse reports (kitty, Ghostty and WezTerm). It is a prototype for testing whether this approach is viable. It requires Node.js 22 or newer.

npx qipgui ./gui/calculator.wasm
npx qipgui qip.dev/gui/calculator.wasm
npx qipgui qip.dev gui/calculator.wasm
node npm/qipgui/qipgui.mjs gui/calculator.wasm
node npm/qipgui/qipgui.mjs --stats --transport zlib gui/snake.wasm
node npm/qipgui/qipgui.mjs --size fill gui/paint.wasm
node npm/qipgui/qipgui.mjs gui/klondike.wasm \
  image/svg+xml/svg-rasterize-to-ktx2-r8g8b8a8-srgb.wasm -u background_color_rgba=0xffffffff

Transform chains

Components after the first transform each frame in order, as with qipdo, and only the last one has to output KTX2 R8G8B8A8_SRGB. This lets SVG components run through a rasterizer. -u sets a uniform on the component before it. Before any component, it applies to the first one.

  • Only the first component receives keys, pointer events, and update calls. Later components must be content transforms: they take input and export no begin_update_at, finish_update, key_event, or pointer_event.
  • Each transform's declared input type must match the output type of the stage before it, when both declare one.
  • Transforms are assumed only to scale the frame. A click is taken as a fraction of the final image and multiplied by the first component's own size, which comes from its KTX2 header or its SVG root's numeric width and height, or its viewBox. For other formats, qipgui uses the final image size.

With svg-rasterize-to-ktx2-r8g8b8a8-srgb.wasm, the repository's 9 SVG GUI components rasterize in 0.4 ms (qip-logo-creature) to 48 ms (klondike) per frame. qipgui allows each component up to 1 GiB of memory (qiptui allows 256 MiB), so the -simd and ThorVG rasterizers also run, but on klondike they are slower: about 220 ms and 3 s per frame.

It shares qiptui's loader, which supports hosted host/path.wasm downloads and leading hosts (qipgui qip.dev gui/calculator.wasm) that supply relative components missing locally. Like qipx and qipok, each component tries the hosts in order, so one pipeline may mix components from different hosts, along with its Strict Wasm Profile checks, -u uniform parsing, and key decoder. That code lives once in npm/shared/qip-host.mjs and is copied into each package as qip-host.mjs by make npm-shared; edit only the source. Only presentation and pointer input are new:

  • The final render() output must be KTX2 R8G8B8A8_SRGB. Each frame's level 0 pixels go to the terminal as an f=32 image in 4096-byte base64 chunks. Frames alternate between image ids 1 and 2: each new image is placed before the old one is deleted, inside synchronized output (?2026), to avoid flicker.
  • Mouse reporting uses ?1003 (all motion), ?1006, and ?1016 (pixel coordinates). The left, middle, and right buttons map to QIP masks 1, 2, and 4. Moving off the image, or losing focus (?1004), sends pointer_event(0, -1, -1). During a drag, a pointer outside the image reports the nearest edge pixel. The wheel is ignored.
  • A component exporting uniform_set_is_focused draws its own focus state, such as an inactive window. qipgui leaves it at 1 at launch, then sets it to 0 or 1 from the terminal's focus reports (?1004) and runs an update and a render on each change. Terminals without focus reporting keep it at 1.
  • Keys are delivered like qiptui's: a key-down and a key-up together, because terminals do not report key releases. Ctrl-C exits and Ctrl-Z suspends.
  • At startup, qipgui queries for graphics support, the window size in pixels (CSI 14 t), and the cell size (CSI 16 t). It fails if the graphics query gets no answer within one second. Each resize sends the size queries again.

Options

| Option | Effect | | --- | --- | | --size auto | 1:1 pixels. An image larger than the window shrinks to fit. This is the default. | | --size native | Always 1:1 pixels. | | --size fill | Scale up or down to fit the window, sized in whole cell columns. | | --transport direct | Base64 RGBA. This is the default. | | --transport zlib | Deflated RGBA (o=z). | | --transport file | Writes a temporary file for the terminal to read (t=t). Only works when the terminal runs on the same machine. | | --stats | On exit, prints to stderr the frame rate, render and encode times, bytes per frame, and how long each input took to reach a flushed frame. |

Backpressure

When stdout.write returns false, qipgui renders no more frames until drain and then renders the latest state once. A slow terminal therefore drops frames instead of queueing seconds of stale ones. --stats reports how many renders were deferred.

Measured encode cost

These are render(0) frame sizes for the repository's gui/ components on an Apple Silicon Mac:

| Component | Size | direct | zlib | | --- | --- | --- | --- | | calculator | 220×220 | 252 KiB, 0.2 ms | 6 KiB, 0.4 ms | | snake | 400×280 | 584 KiB, 0.2 ms | 22 KiB, 0.7 ms | | sudoku | 762×522 | 2.0 MiB, 0.3 ms | 40 KiB, 1.6 ms | | god-rays | 640×360 | 1.2 MiB, 0.4 ms | 756 KiB, 22 ms | | openai-anthropic-arr | 1640×1080 | 9.0 MiB, 4.5 ms | 155 KiB, 7.6 ms | | photo-light-table | 1600×1040 | 8.5 MiB, 2.0 ms | 2.2 MiB, 54 ms |

Flat interface frames compress 20 to 60 times for a few milliseconds of work. Photographic and shaded frames compress poorly and cost tens of milliseconds, so direct or file suits them better.

qipgui presents 47 of the 57 gui/*.wasm components on their own. The other 9 SVG components need a rasterizer after them (see above), and gif-player needs GIF input (-i file.gif).

Still to verify in a real terminal

  • Whether SGR pixel coordinates are 0-based or 1-based in each terminal. qipgui treats them as 0-based. Click a corner pixel to check.
  • Whether pixel reports and the CSI 14 t window size use the same units on HiDPI displays.
  • Pointer accuracy after resizing with --size fill.
  • Frame rate and input latency over SSH, where --transport file is not available.