Skip to main content

maya-py

Python bindings for maya — a C++26 TUI framework with flexbox layout, a SIMD cell-diff renderer, and an Elm-style runtime. maya-py gives you a dead-simple Python API for building styled terminal UIs and interactive apps.

import maya_py as maya
from maya_py import card, field, b, hr

maya.show(card(
    b("maya-py").fg("sky"),
    hr(20),
    field("Status", "Online", value_color="green"),
    field("Region", "us-east-1"),
    title="service",
))

That's the whole program. Strings are UI — no manual element wrapping.

Documentation

A full reference manual lives in docs/:

The 30-second tour

Text: just style strings

from maya_py import T, b, i, dim_text, c

b("bold")                 # bold
T("hi").bold.fg("sky")     # fluent chain
c("warn", "orange")        # colored
T("x").bg("red").fg("white")

Colors accept names ("red", "sky", "gold"), hex ("#ff8800", "#f80"), tuples ((255, 128, 0)), or ints (0xFF8800).

Layout: stacks that take bare strings

from maya_py import col, row, card, field, hr

col("top", "middle", "bottom")          # vertical
row("left", "right", gap=2)             # horizontal
card("body", title="hi", pad=1)          # bordered box
field("Name", "Ada")                     # "Name: Ada"
hr(40)                                   # horizontal rule

col / row / card accept the full maya flexbox surface as keywords:

  • box model: pad, margin, gap (int or 1/2/4-tuples)
  • border: border ("round"/"double"/"bold"/"dashed"/...), border_color, border_sides=sides(top=..., left=...), title, and positioned border_text=("Title", Top, Center) / border_text_end
  • sizing: width, height, min_width/max_width, min_height/max_height — each takes an int (cells), "50%", a float like 0.5, "auto", or pct(50)/cells(20)/auto()
  • flex: grow, shrink, basis, align, align_self, justify, wrap ("wrap"/"nowrap"/"reverse"), overflow ("hidden"/"scroll")
  • style: bg, fg, style=maya.style(...)

Full layout power

Everything maya C++ can express, the Python API can too:

from maya_py import T, col, row, card, center, stack, grow, component, pct, cells, sides

# percent widths + a flex child that fills the rest
row(
    card("nav", width=pct(30), title="sidebar"),
    grow(card("main content", title="body")),   # expands to fill
    gap=1,
)

# center anything in a region
center("ready", width=cells(20), height=5, border="round")

# z-stack: layers paint on top; the first sets the size
stack(card("  panel  ", height=6), T("NEW").fg("red").bold)

# partial borders
card("footer", border_sides=sides(top=True, right=False, bottom=False, left=False))

# a size-aware widget — render_fn(width, height) draws to fit its box
def bar(w, h):
    filled = int(w * 0.4)
    return T("█" * filled + "░" * (w - filled)).fg("green")
col("Loading", component(bar, height=1))

Widgets: maya's native renderers

maya ships a library of ready-made widgets — they render through the same C++ renderer maya uses, then drop straight into any layout:

from maya_py import col, row, sparkline, gauge, progress, badge, table, bar_chart

col(
    sparkline([3, 1, 4, 1, 5, 9, 2, 6], label="req/s", color="sky", show_last=True),
    gauge(0.72, "load", style="bar"),            # "arc" or "bar"
    progress(0.55, "build", width=24, fill="lime"),
    row(badge("PASS", kind="success"), badge("SKIP", kind="warning"), gap=1),
    table(["Name", ("Score", 0, "right")],       # (header, width, align)
          [["Ada", 99], ["Bob", 7]], bordered=True, title="top"),
    bar_chart([("jan", 4), ("feb", 9), ("mar", 6)]),
    gap=1,
)

Available — every color argument takes a name / (r,g,b) / "#rrggbb" / Color, same as everywhere else:

  • charts & meters: sparkline, gauge, progress, bar_chart, line_chart, heatmap, flame_chart, waterfall, activity_indicator
  • controls (rendered in any state — pass checked/on/selected/cursor): checkbox, toggle, radio, select, slider, button
  • text & labels: badge, divider, spinner, callout, status_banner, breadcrumb, tabs, gradient, link, title_chip, model_badge, file_ref, markdown, html, error_block
  • structure & nav: table, tree, list_view, menu, disclosure, key_help, calendar, timeline, picker (bordered command palette), modal, command_palette, log_viewer
  • agent UI: thinking, todo_list, toast, inline_diff, tool_call, plan_view, phase_chip, context_gauge, context_window, diff_view, git_graph, git_status, shortcut_row, system_banner, popup, overlay, user_message, assistant_message, activity_bar, file_changes, api_usage, cost_tracker, phase_accent, checkpoint_divider, turn_divider, streaming_cursor, token_stream, token_stream_sparkline, search_result, changes_strip, welcome_screen
  • graphics: image (1-bit braille), canvas (color half-block grid), Canvas (imperative drawing surface: set_pixel/line/rect/fill)
from maya_py import col, row, checkbox, slider, todo_list, timeline, tree

col(
    checkbox("Ship it", checked=True),
    slider(0.6, "volume", width=24, fill="sky"),
    todo_list([("design", "completed"), ("build", "in_progress"), "test"],
              description="sprint", status="running"),
    timeline([("clone", "", "0.4s", "completed"),
              ("compile", "", "", "in_progress", 8)]),
    tree({"label": "src", "expanded": True,
          "children": [{"label": "main.py"}, {"label": "util.py"}]}),
    gap=1,
)

List/tuple inputs are flexible: list items accept "str", (label, description, icon), or {...} dicts; timeline/todo statuses accept strings ("completed", "in_progress", ...) or the exported enums (TaskStatus, TodoItemStatus, ToastLevel, ButtonVariant, ToolCallKind, ToolCallStatus, PopupStyle, BannerLevel, FileChangeKind, TurnRole, CursorStyle, SearchKind, SearchStatus).

Scrolling: viewport + scrollbar

Clip tall/wide content to a window and pair it with a live scrollbar. Scrolling just works with no handler code — exactly like maya, the run loop auto-forwards ↑↓ / PgUp / PgDn / Home / End and the mouse wheel + scrollbar drag to every on-screen scroll state:

from maya_py import App, col, row, T, scroll_state, viewport, scrollbar

app = App("log", mouse=True)            # mouse=True for wheel + thumb drag
s = scroll_state()                      # auto-dispatch on (the maya default)
app.state(s=s)
content = col(*[T(f"line {i}") for i in range(200)])

@app.on("q", "esc")
def quit(st): app.stop()

@app.view
def view(st):
    return row(
        viewport(content, st.s, height=14, grow=1),   # 14-row window, fills width
        scrollbar(st.s, 14, style="neon", thumb_color="sky"),
        gap=1,
    )

app.run()

That's the whole thing — no scroll handler. viewport(content, state, width=, height=, grow=) clips + scrolls (0 on an axis = fill; grow=1 expands to fill its row/column so a sibling scrollbar pins to the edge). scrollbar(state, size, axis="y"|"x", style=, thumb_color=, track_color=) draws the bar; style is a preset name — line, block, slim, heavy, double, dotted, dashed, braille, ascii, shadow, minimal, neon, retro, danger, pixel. The ScrollState exposes x/y/max_x/max_y, scroll_by, scroll_to, scroll_to_top/bottom, at_top()/at_bottom(), and viewport_bounds (the painted rect, for click hit-testing).

Need custom routing (several independent scroll regions)? Set state.auto_dispatch = False and call scroll_handle(state, ev) yourself inside @app.on_key / @app.on_mouse. Don't do both — that double-scrolls.

Apps: a class with decorators, no event loop

from maya_py import App, card, b

app = App("counter", n=0, quit_keys=("q", "esc"))   # state + quit in the constructor

@app.on("+", "=")
def inc(s): s.n += 1

@app.on("-")
def dec(s): s.n -= 1

@app.view
def view(s):
    return card(b(f"Count: {s.n}").fg("sky"), title="counter")

app.run()

Key names: single chars ("q", "+"), "up"/"down"/"left"/"right", "enter", "esc", "space", "tab"/"shift+tab", "home"/"end", "pgup"/"pgdn", "insert"/"delete", "f1"–"f12", and modifier prefixes "ctrl+s", "alt+x", "shift+a". A typo'd key spec raises with a did-you-mean hint instead of silently never firing. Handlers get the state object; the view re-renders every frame.

Text input

Real interactive Input widgets — type into them, Tab between them, read .value. Register them with app.focus(...) and drop them in the view:

from maya_py import App, text_input, col

app = App("search")
query = text_input("type to filter…")
app.focus(query)

@app.view
def view(s):
    return col(query, f"searching: {query.value}")

app.run()

text_input(password=True) masks, textarea() is multi-line. @query.on_submit fires on Enter. An animation tick is @app.on_frame def tick(s, dt): ....

Declarative state: For, bind, @app.derive

The view is a pure function of state, so the whole UI reads top-to-bottom. Three helpers kill the last of the glue — list comprehensions, widget read-back, and recomputed fields:

from maya_py import App, text_input, card, col, For

app = App("cart", items=[("Pen", 2.0), ("Mug", 9.5)], name="")

# 1) @app.derive — a computed field, recomputed on access (no stale cache):
@app.derive
def total(s): return sum(price for _, price in s.items)

# 2) bind=(state, field) — the input mirrors a state field both ways, so the
#    view reads s.name directly and never touches widget.value:
name = text_input("your name…", bind=(app.s, "name"))
app.focus(name)

@app.view
def view(s):
    return card(
        # 3) For(...) maps a list into a box — like SwiftUI ForEach / JSX map.
        #    A two-param renderer gets (index, item) for free.
        For(s.items, lambda it: col(f"{it[0]}  ${it[1]:.2f}")),
        col("Name:", name, f"hello {s.name}"),
        f"Total: ${s.total:.2f}",
        title="cart",
    )

app.run()
  • For(items, render, *, into=col, empty=None, **opts) — map a sequence into a container. render(item) or render(index, item) (auto-detected). empty= shows a fallback when the list is empty.
  • text_input(..., bind=(state, "field")) — two-way binding: the field seeds the input and every keystroke writes back. No .value plumbing.
  • @app.derive — install fn(state) as a read-only computed property on the state object. Works with the default state bag and model= objects.

Mouse: clicks, scroll, drag

Decorate handlers for mouse input — registering any of them auto-enables mouse reporting (no mouse=True needed):

from maya_py import App, card

app = App("clicker")
app.state(hits=0, pos=(0, 0))

@app.on_click("left")          # "left" / "right" / "middle" / "any"
def click(s, col, row):        # col, row are 1-based screen cells
    s.hits += 1
    s.pos = (col, row)

@app.on_scroll
def scroll(s, direction):      # -1 up, +1 down
    s.hits += direction

@app.on_mouse
def any_mouse(s, ev):          # every mouse event (press/release/move/scroll)
    ...

@app.on("q", "esc")
def quit(s): app.stop()

@app.view
def view(s):
    return card(f"clicks {s.hits}  at {s.pos}", title="clicker")

app.run()

Low-level predicates work on any event too: mouse_clicked(ev, button), mouse_released(ev), mouse_moved(ev), scrolled_up/down(ev), mouse_pos(ev) -> (col, row) | None, mouse_button(ev), mouse_kind(ev), is_mouse(ev). Mouse needs a terminal that reports it (xterm, kitty, iTerm2, Windows Terminal, or tmux with set -g mouse on).

Animation

import maya_py as maya
from maya_py import animate, card

frames = "⠋⠙⠹⠸⠼⠴⠦⠧⠇⠏"
n = 0
def render(dt):
    global n; n += 1
    return card(frames[n % len(frames)] + " working")
animate(render, fps=30)   # maya.quit() to stop

The Elm Architecture: Program (pure, testable, no mutation)

When you want the full discipline — an immutable Model, a closed set of Msg, a pure update, and side effects described as data — maya-py exposes maya's native MVU runtime as a Program. This is the same run<P> loop the C++ side uses, so the effects and subscriptions you return from Python are interpreted by maya's real runtime, not a reimplementation.

import maya_py as maya
from maya_py import Cmd, Sub, Program, card, T

class Counter(Program):
    title = "counter"

    def init(self):                       # -> model | (model, Cmd)
        return {"n": 0}, Cmd.set_title("counter")

    def update(self, m, msg):             # pure: no I/O, no mutation
        if msg == "inc":   return {**m, "n": m["n"] + 1}
        if msg == "dec":   return {**m, "n": m["n"] - 1}
        if msg == "quit":  return m, Cmd.quit()
        return m

    def view(self, m):                    # model -> Element
        return card(T(f"count: {m['n']}").bold)

    def subscribe(self, m):               # model -> event sources
        return Sub.on_key(lambda ev:
            "inc" if maya.key(ev, "+") else
            "dec" if maya.key(ev, "-") else
            "quit" if maya.key(ev, "q") else None)

Counter().run()

update never touches the terminal, so it stays a pure state transition you can unit-test in isolation. Effects are Cmd values — Cmd.quit(), Cmd.set_title(...), Cmd.after(ms, msg), Cmd.task(fn) (async work), Cmd.batch(...), clipboard + scrollback commands. Event sources are Sub values — Sub.on_key, Sub.on_mouse, Sub.on_resize, Sub.on_paste, Sub.every(ms, msg) (a timer), Sub.on_animation_frame(msg), Sub.batch(...). There's a plain-function form too: run_program(init, update, view, subscribe).

The whole architecture is headlessly testable. Because update is pure, Program.test() hands you a ProgramPilot that runs the real init/update/view with no terminal — you dispatch messages (exactly like elm-test), thread the immutable model, and assert on it and the rendered view:

p = Counter().test()
p.send("inc", "inc", "inc", "dec")        # dispatch messages in order
assert p.model == {"n": 2}
assert p.cmds                              # Cmd.set_title was emitted at init
assert "count: 2" in p.view_string(40)

Purity is enforced, not just encouraged. The pilot runs in strict mode by default: it snapshots the model before every update/view and raises ImpureUpdateError if a hook mutated it in place instead of returning a new one. So the moment someone writes the tempting m["n"] += 1; return m, their tests fail loudly with a pointer to the exact hook — the beautiful, pure architecture is the verified one, and the subtle stale-view bug never ships:

class Sloppy(Program):
    def update(self, m, msg):
        m["n"] += 1        # in-place mutation — the TEA sin
        return m
    ...

Sloppy().test().send("inc")   # raises ImpureUpdateError: update() mutated the
                              # model in place. Return a NEW model ({**m, ...}).

The check costs nothing at runtime (it lives in the test pilot), and there's an escape hatch — prog.test(strict=False) — for an intentionally huge or un-deep-copyable model.

See examples/counter_program.py and examples/stopwatch_program.py (the latter uses Sub.every timers, Sub.batch, and Cmd.batch, and ships a --test self-check that proves every transition with the pilot).

Examples

Every maya C++ example is ported 1:1 to Python — 34 of them, plus extras. Most are full apps you can drive; run any with PYTHONPATH=src python examples/NAME.py. A headless examples/smoke_all.py renders one frame of each (CI-friendly, no TTY needed).

Games & toys (half-block pixel rendering via examples/_halfblock.py):

  • snake.py — playable Snake (arrow keys / WASD).
  • breakout.py — Breakout/Arkanoid with bricks, comet trail, particles.
  • life.py — Conway's Game of Life with heat-aging palettes + patterns.
  • sorts.py — five sorting algorithms racing side by side.
  • maze.py — watch a maze carve itself (recursive backtracker) then a BFS flood-solve it; click to set the start cell, right-click the goal.

Graphics & sims (animate / App render loops):

  • doom_fire.py — the classic Doom fire effect, 3 palettes.
  • fluid.py — advection fluid / plasma with curl-noise velocity.
  • particles.py — a gravity particle fountain.
  • space.py — warp-speed starfield. space3d.py — rotating 3D wireframes.
  • raymarch.py — a real-time raymarched SDF scene (sphere + plane).
  • fps.py — a Wolfenstein-style textured raycaster with a minimap.
  • mandelbrot.py — a zooming coloured Mandelbrot. matrix.py — digital rain.
  • canvas.py — the Canvas drawing surface: lines, rects, a live plot.
  • clock.py — a live analog clock drawn on the Canvas (hands, ticks, arc).
  • gravity.py — an n-body gravity sandbox; mouse-hover crosshair, click to spawn an orbiting burst at the exact pixel.
  • boids.py — Reynolds flocking; the mouse leads or scatters the flock (precise, viewport-bounds hit-testing — no hardcoded offsets).

Dashboards & data:

  • dashboard.py — full-power layout (sidebar, z-stack, size-aware bars).
  • sysmon.py — live system monitor (sparklines, gauges, log feed).
  • stocks.py — stock ticker with sparklines + a gainers board.
  • spectrum.py — a faux audio spectrum analyzer. music.py — a player UI.
  • hacker.py — a "hollywood hacker" terminal of fake breaches.

Apps & agent UI:

  • deploy.py — a CI/CD pipeline dashboard with a live stage timeline.
  • chat.py — a chat client with bubbles + typing indicator.
  • messenger.py — multi-channel chat with unread badges + a composer.
  • ide.py — a VS Code / Zed-style mini IDE (tree, tabs, diagnostics, git).
  • agent.py / agent_session.py — Claude-Code-style agent sessions: thinking → tool cards → todo plan → streaming markdown answer.
  • widgets.py — one-shot widget showcase. widgets_gallery.py — live version.
  • markup.py — a GFM markdown render that flows into scrollback (pipe to less -R).
  • inline_progress.py — inline print + live (no alt-screen takeover).

Primitives & basics:

  • hello.py static card · counter.py / stopwatch.py / todo.py App basics · paint.py mouse painter · live_spinner.py inline animation.
  • counter_program.py / stopwatch_program.py — the pure MVU Program (Elm Architecture): immutable model, pure update, Cmd/Sub effects, and a headless --test self-check.
  • form.py — the declarative trio (For, bind=, @app.derive) in one cart.
  • scroll.py / scroll_clip.py / scroll_2d.py / scroll_slice.py / scroll_styles.py — every scrolling pattern (clip, two-axis, million-row slice, and all 15 scrollbar styles).

Run any of them:

PYTHONPATH=src python examples/agent_session.py

Install

The wheels are standalone — no compiler needed. maya-py ships precompiled binary wheels for CPython 3.9–3.14 on Linux (x86_64; aarch64 configured), Windows (x64), and macOS (Apple Silicon / arm64), so on a normal machine you just:

pip install maya-py

That's it — the wheel already contains the compiled extension, and pip picks the right one for your OS / architecture / Python automatically.

Then scaffold a runnable app and go:

maya new myapp     # writes myapp.py — a working counter app
python myapp.py    # run it (q / Esc to quit)

(maya new, maya demo, maya version also work as python -m maya_py ....)

Terminal note. maya speaks ANSI/VT escapes + truecolor + UTF-8. On Linux and macOS any modern terminal works. On Windows use Windows Terminal (the default on Windows 11; free on Windows 10) or any VT-capable host — the legacy cmd.exe console works too (maya enables ENABLE_VIRTUAL_TERMINAL_PROCESSING on startup), but Windows Terminal gives the best Unicode + color fidelity.

Installing straight from a GitHub Release (no PyPI)

If you'd rather not use PyPI (or want a specific build), install from the release assets. Let pip pick the right wheel for your Python:

pip install --find-links \
  https://github.com/1ay1/maya-py/releases/expanded_assets/v0.2.7 \
  maya-py

Or install a specific .whl by direct URL:

# e.g. CPython 3.13 on x86_64 Linux
pip install https://github.com/1ay1/maya-py/releases/download/v0.2.7/maya_py-0.2.7-cp313-cp313-manylinux_2_28_x86_64.whl

The Linux and macOS wheels work even on machines with a very old C++ toolchain (or none at all), because:

  • the wheel is built inside a manylinux_2_28 container, so it targets glibc 2.28 (2019) and runs on that-or-newer distros;
  • it statically links libstdc++/libgcc, so it doesn't need the host's (old) C++ runtime — it depends only on baseline libc/libm.

The Windows wheel links the UCRT present on every Windows 10+ machine. Nothing on your machine is compiled at install time, so your system compiler is irrelevant.

No matching wheel? Build from the sdist

If no prebuilt wheel matches your platform/Python, install the source distribution; this compiles the extension locally and needs a C++23-capable toolchain and CMake ≥ 3.28:

  • Linux: GCC ≥ 14 or Clang ≥ 18
  • macOS: Homebrew GCC (brew install gcc) — AppleClang is rejected (it misses C++23 library bits maya needs); CMake auto-picks the gcc-* driver
  • Windows: Visual Studio 2022 ≥ 17.10 (MSVC cl ≥ 19.40), "Desktop development with C++" workload
pip install \
  https://github.com/1ay1/maya-py/releases/download/v0.2.7/maya_py-0.2.7.tar.gz

The compile pulls maya in via CMake FetchContent and takes ~1–2 minutes (it builds maya from source). If your compiler is too old, the build aborts early with a clear, actionable message rather than a wall of template errors.

Development build

To hack on maya-py, build the extension in place (uses a sibling ../maya-src checkout if present, else clones maya):

cmake -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build -j
cp build/_maya*.so src/maya_py/
PYTHONPATH=src python examples/hello.py

Wheels are produced by cibuildwheel via .github/workflows/wheels.yml — push a vX.Y.Z tag to build them and attach them to a GitHub Release automatically.

Low-level API

The friendly layer above is built on a thin primitive surface. You rarely need it, but it's there when you want raw control. box/text take explicit Style/Color objects:

maya.box(
    maya.text("Hello", maya.bold | maya.fg(255, 128, 0)),
    border=maya.Round, padding=1,
)
  • text(content, style=None, wrap=...), box(*children, **opts), vstack/hstack/zstack, center, blank(), nothing().
  • box opts mirror maya's BoxBuilder 1:1: direction, wrap, gap, padding, margin, border, border_color, border_sides, border_text, border_text_end, bg, fg, style, overflow, grow, shrink, basis, align, align_self, justify, and width/height/min_*/max_* (each accepts an int, "50%", "auto", or a Dimension).
  • component(fn, grow=..., width=..., height=...) — a lazy element whose fn(width, height) runs once layout allocates a size and returns the tree to fill it.
  • scroll_state(), viewport(content, state, width=, height=), scrollbar(state, size, axis=, style=), scroll_handle(state, ev) — the scrollable-window + scrollbar pair (maya's ScrollState + scrollbar_y/x).
  • Styles compose with |: maya.bold | maya.fg(255, 128, 0), or maya.style(fg=(80,220,120), bold=True).
  • render_to_string(element, width=80), print(element), live(render_fn, fps=30), run(event_fn, render_fn, ...).
  • Event predicates: key(ev, "q"), key_special(ev, SpecialKey.Up), ctrl(ev, "c"), alt(ev, "x"), any_key(ev), resized(ev).

Notes

maya's compile-time DSL (t<"...">, type-state pipes) is resolved by the C++ compiler and can't cross the Python boundary, so maya-py routes everything through maya's equivalent runtime builders. The element trees produced are identical — anything the DSL can express, the runtime API can too.

The widget functions wrap maya's own widget classes (Sparkline, Gauge, Table, ...) and return the Element they build, so what you see is maya's native rendering, not a Python reimplementation. Interactive controls that need maya's Program runtime + focus + signals (Input, TextArea, full List/Tree navigation) aren't wrapped — drive interactivity from the App class instead.

Performance

Honest numbers from examples/bench.py and examples/bench_live.py (30-row dashboard, this machine — yours will differ):

Rendering to a string (one-shot output). Here a tuned pure-Python renderer still edges ahead — building the element tree in Python and crossing pybind11 costs more than maya's native render saves:

path per render
maya-py (build + render) ~145 µs
pure-Python equivalent ~70 µs

~80% of maya-py's time is the Python tree construction + boundary crossing, not maya — the native render itself is only ~24 µs. So if you only render static output once, a pure-Python lib like Rich is still a touch faster (now ~2×, down from ~5×).

Live redraw to a terminal (what maya is built for). When you redraw a frame and only part changed, maya's SIMD cell-diff emits only the changed cells. A string renderer must re-emit the whole frame:

path bytes written / frame
maya-py (diff) ~112 B
re-emit whole frame ~1238 B

maya writes ~11× fewer bytes to the terminal. On a real tty — especially over SSH or a slow connection — bytes-on-wire is the bottleneck, and this is a decisive win. This is the scenario maya was designed for.

Making it fast

Three levers close most of the Python-side gap:

  1. Fluent styling is free. T("x").bold.fg("sky") accumulates state in pure Python and makes a single boundary crossing when the element is built (4.8× faster than the naive one-call-per-.bold approach).

  2. Pass styled cells as tuples to skip the T objects. row/col accept (text, fg[, bg[, attrs]]) tuple specs directly — in a hot redraw path (tables, lists, dashboards) the throwaway T per cell is the dominant build cost, and the tuple form flattens the whole row in one boundary crossing. Byte-identical output, same API:

    from maya_py import row, T, c, DIM
    
    # reads nicer for one-off UI:
    row(T(name).fg("sky"), c(status, color), T(latency).dim, gap=2)
    # faster (tuple cells, zero T allocations — use in per-frame loops):
    row((name, "sky"), (status, color), (latency, None, None, DIM), gap=2)
    

    You can mix both, and any built Element / nested box / component child transparently falls back to the general path — so row(sidebar, grow(main)) still works. (trow/tcol are kept as back-compat aliases; new code can just use row/col.)

  3. memo caches unchanged sub-trees. In a live app, wrap builders whose inputs rarely change — the hot frame then does no Python tree construction, just maya's native diff:

    from maya_py import memo, card, col, b
    
    @memo
    def header(title, count):     # rebuilt only when title/count change
        return card(b(title), f"{count} items")
    
    def view(s):
        return col(header(s.title, len(s.items)), body(s))
    

Bottom line: building a UI in Python always costs Python calls, but the thing that actually matters in a terminal — the per-frame redraw — is now faster than a hand-tuned pure-Python renderer: maya's native layout + paint renders the cached tree in ~25µs vs ~70µs for the bespoke string-builder in examples/bench.py, while still doing real flexbox, wrapping, and a partial-frame diff. memo + tuple-cell row/col let the steady-state frame skip Python almost entirely.

License

MIT (same as maya).

Metadata

Release files for maya-py 0.2.8

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for maya-py 0.2.8
File Size Uploaded
maya_py-0.2.8.tar.gz 421.3 kB Details

Built distributions (wheels)

Table of built distributions (wheels) for maya-py 0.2.8
File
maya_py-0.2.8-cp314-cp314-win_amd64.whl CPython 3.14 CPython 3.14 Windows x86-64 Details
maya_py-0.2.8-cp314-cp314-manylinux_2_28_x86_64.whl CPython 3.14 CPython 3.14 Linux glibc 2.28+ x86-64 Details
maya_py-0.2.8-cp314-cp314-macosx_11_0_arm64.whl CPython 3.14 CPython 3.14 macOS 11.0+ ARM64 Details
maya_py-0.2.8-cp313-cp313-win_amd64.whl CPython 3.13 CPython 3.13 Windows x86-64 Details
maya_py-0.2.8-cp313-cp313-manylinux_2_28_x86_64.whl CPython 3.13 CPython 3.13 Linux glibc 2.28+ x86-64 Details
maya_py-0.2.8-cp313-cp313-macosx_11_0_arm64.whl CPython 3.13 CPython 3.13 macOS 11.0+ ARM64 Details
maya_py-0.2.8-cp312-cp312-win_amd64.whl CPython 3.12 CPython 3.12 Windows x86-64 Details
maya_py-0.2.8-cp312-cp312-manylinux_2_28_x86_64.whl CPython 3.12 CPython 3.12 Linux glibc 2.28+ x86-64 Details
maya_py-0.2.8-cp312-cp312-macosx_11_0_arm64.whl CPython 3.12 CPython 3.12 macOS 11.0+ ARM64 Details
maya_py-0.2.8-cp311-cp311-win_amd64.whl CPython 3.11 CPython 3.11 Windows x86-64 Details
maya_py-0.2.8-cp311-cp311-manylinux_2_28_x86_64.whl CPython 3.11 CPython 3.11 Linux glibc 2.28+ x86-64 Details
maya_py-0.2.8-cp311-cp311-macosx_11_0_arm64.whl CPython 3.11 CPython 3.11 macOS 11.0+ ARM64 Details
maya_py-0.2.8-cp310-cp310-win_amd64.whl CPython 3.10 CPython 3.10 Windows x86-64 Details
maya_py-0.2.8-cp310-cp310-manylinux_2_28_x86_64.whl CPython 3.10 CPython 3.10 Linux glibc 2.28+ x86-64 Details
maya_py-0.2.8-cp310-cp310-macosx_11_0_arm64.whl CPython 3.10 CPython 3.10 macOS 11.0+ ARM64 Details
maya_py-0.2.8-cp39-cp39-win_amd64.whl CPython 3.9 CPython 3.9 Windows x86-64 Details
maya_py-0.2.8-cp39-cp39-manylinux_2_28_x86_64.whl CPython 3.9 CPython 3.9 Linux glibc 2.28+ x86-64 Details
maya_py-0.2.8-cp39-cp39-macosx_11_0_arm64.whl CPython 3.9 CPython 3.9 macOS 11.0+ ARM64 Details

Total release size: 46.5 MB

Release files / maya_py-0.2.8.tar.gz

Download URL maya_py-0.2.8.tar.gz
Size 421.3 kB
Tags Source
SHA-256 checksum
How to use checksums
2302f3a95327f96c0835de1ef861f2476502e521d0473975671dae9223953b15
BLAKE2b-256 checksum
How to use checksums
6eb3c6252f792d40956624441543c4bedc110c5ee8c785e279be9c90e7115db7
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp314-cp314-win_amd64.whl

Download URL maya_py-0.2.8-cp314-cp314-win_amd64.whl
Size 1.3 MB
Tags CPython 3.14 Windows x86-64
SHA-256 checksum
How to use checksums
3b670c2e2f6681c605c4f661f2cae5baba94a8eb0ab5cb0000b85c3d783dee51
BLAKE2b-256 checksum
How to use checksums
6b8045b8e48d6f6e2862a49c3b69d088244d291be7914b86bf08928de1428fe6
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp314-cp314-manylinux_2_28_x86_64.whl

Download URL maya_py-0.2.8-cp314-cp314-manylinux_2_28_x86_64.whl
Size 4.1 MB
Tags CPython 3.14 Linux glibc 2.28+ x86-64
SHA-256 checksum
How to use checksums
cb423caeb047c40236b508ae5edf9ada12ce4d698689c5ead29695ea9ebd17f3
BLAKE2b-256 checksum
How to use checksums
d1b98bad92199c54ea4b3d4709a234cf28cc69a85d8c88041842a3f8a4aa6591
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp314-cp314-macosx_11_0_arm64.whl

Download URL maya_py-0.2.8-cp314-cp314-macosx_11_0_arm64.whl
Size 2.3 MB
Tags CPython 3.14 macOS 11.0+ ARM64
SHA-256 checksum
How to use checksums
421f6421dc252a536b4659127c414eb14599a19c54e8d741355ada9e39a9b84d
BLAKE2b-256 checksum
How to use checksums
a6fb7bd3511ff04f54bac162a6a14bcd0e6e2e285e758d87120a6b8c0e6219ce
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp313-cp313-win_amd64.whl

Download URL maya_py-0.2.8-cp313-cp313-win_amd64.whl
Size 1.3 MB
Tags CPython 3.13 Windows x86-64
SHA-256 checksum
How to use checksums
78d1217f49f7da09420599ac9242922096555122f75d1fb830f6f8d36007d012
BLAKE2b-256 checksum
How to use checksums
5a8123b9e0afe2da3933d76f2c23429885f0df6eaddb54f75cdf5ea11611bbbf
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp313-cp313-manylinux_2_28_x86_64.whl

Download URL maya_py-0.2.8-cp313-cp313-manylinux_2_28_x86_64.whl
Size 4.1 MB
Tags CPython 3.13 Linux glibc 2.28+ x86-64
SHA-256 checksum
How to use checksums
bf3197294db60381a3f0892011805b7110d3ea9b8e6ca26c66410ca04a4c1f84
BLAKE2b-256 checksum
How to use checksums
a75c735e623f83130299407013815081efb72b695551c42be082ee68b5d2f718
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp313-cp313-macosx_11_0_arm64.whl

Download URL maya_py-0.2.8-cp313-cp313-macosx_11_0_arm64.whl
Size 2.3 MB
Tags CPython 3.13 macOS 11.0+ ARM64
SHA-256 checksum
How to use checksums
40437d7f6fdca4f84204498d973621e375ee0d784ebd437e1b4a4a0d165bf176
BLAKE2b-256 checksum
How to use checksums
dc94ceaf518ebc72835a7b367a6f04fb9ed20b90271673a479572b7ecda6f923
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp312-cp312-win_amd64.whl

Download URL maya_py-0.2.8-cp312-cp312-win_amd64.whl
Size 1.3 MB
Tags CPython 3.12 Windows x86-64
SHA-256 checksum
How to use checksums
39c4637ea0ec31fad0442aa51467f839189b3da63fa6e9cd67f7b21bd649e178
BLAKE2b-256 checksum
How to use checksums
e4c1d822323252a65ea03ed2785808b61d4ee2f43ea1ddb24f91ee3ec506ecd9
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp312-cp312-manylinux_2_28_x86_64.whl

Download URL maya_py-0.2.8-cp312-cp312-manylinux_2_28_x86_64.whl
Size 4.1 MB
Tags CPython 3.12 Linux glibc 2.28+ x86-64
SHA-256 checksum
How to use checksums
4bc378631f9b4f2358901bd494b0b8b66150d8ac21d3acfc49bfb121331814bd
BLAKE2b-256 checksum
How to use checksums
e35aa8f3fc6b71b4f3b8814f8f001c975f5c12487fa3b103282836cddf5d74b4
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp312-cp312-macosx_11_0_arm64.whl

Download URL maya_py-0.2.8-cp312-cp312-macosx_11_0_arm64.whl
Size 2.3 MB
Tags CPython 3.12 macOS 11.0+ ARM64
SHA-256 checksum
How to use checksums
cd4d072f52bb3bc9c22f1298855abd81bd2e25de089ae0be0888c9721a3eded0
BLAKE2b-256 checksum
How to use checksums
1ede0a10c133384a76869f0cad99c546f3dc96ee05436a75ff165ae1f31952ae
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp311-cp311-win_amd64.whl

Download URL maya_py-0.2.8-cp311-cp311-win_amd64.whl
Size 1.3 MB
Tags CPython 3.11 Windows x86-64
SHA-256 checksum
How to use checksums
78dfcb4f41be16cc2d45575736093ecb76d67f36d7f10daa8ed99e9566c5e756
BLAKE2b-256 checksum
How to use checksums
6d81272140d2388555594815b85a50ebf5855b3eeda0d06fefc6148d617256dd
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp311-cp311-manylinux_2_28_x86_64.whl

Download URL maya_py-0.2.8-cp311-cp311-manylinux_2_28_x86_64.whl
Size 4.1 MB
Tags CPython 3.11 Linux glibc 2.28+ x86-64
SHA-256 checksum
How to use checksums
01f73ba3cbb5b62ccd7ecd81a09cc752ef589adf63cfe7f090d1d63bd58ed23a
BLAKE2b-256 checksum
How to use checksums
0e2ce09f2984a7767cfb52a33a95ba80e39c8d68b8a1245897f832e22fe2eb17
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp311-cp311-macosx_11_0_arm64.whl

Download URL maya_py-0.2.8-cp311-cp311-macosx_11_0_arm64.whl
Size 2.3 MB
Tags CPython 3.11 macOS 11.0+ ARM64
SHA-256 checksum
How to use checksums
0eb56a2685e964deb4a156a23d6ed13798850bf8e1d1c6e703e3ee50e6be1084
BLAKE2b-256 checksum
How to use checksums
cbcc477d0536dde820b9e873e2f52d53c7786ea57bfa1cbe6cbc4f4aacb17a8c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp310-cp310-win_amd64.whl

Download URL maya_py-0.2.8-cp310-cp310-win_amd64.whl
Size 1.3 MB
Tags CPython 3.10 Windows x86-64
SHA-256 checksum
How to use checksums
1db03039c910e923975b605bdd5b396622dfb089b7f9d6c688de67b073b96c91
BLAKE2b-256 checksum
How to use checksums
0cb3b9c131e9300cc9851c7bc996236cd9a652ecd11af844f7fe8effdc4e6e22
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp310-cp310-manylinux_2_28_x86_64.whl

Download URL maya_py-0.2.8-cp310-cp310-manylinux_2_28_x86_64.whl
Size 4.1 MB
Tags CPython 3.10 Linux glibc 2.28+ x86-64
SHA-256 checksum
How to use checksums
02b401246ad16bcb158090ff541d011a2e7f68db87f5d0c446323ad8d6cec645
BLAKE2b-256 checksum
How to use checksums
4a861059a711c16e965600f0a4cc735dbc9ebae235f57b571fd82756b3b3e5b7
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp310-cp310-macosx_11_0_arm64.whl

Download URL maya_py-0.2.8-cp310-cp310-macosx_11_0_arm64.whl
Size 2.3 MB
Tags CPython 3.10 macOS 11.0+ ARM64
SHA-256 checksum
How to use checksums
214b31ee671a657a6219bd9e93784782042b0d3692fd5fa5ef885c8b79ed0928
BLAKE2b-256 checksum
How to use checksums
7d260b1251929f295463e6c3b7802fca5a4a8798291095ef72be77b323426245
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp39-cp39-win_amd64.whl

Download URL maya_py-0.2.8-cp39-cp39-win_amd64.whl
Size 1.3 MB
Tags CPython 3.9 Windows x86-64
SHA-256 checksum
How to use checksums
69f98e4a6c09023b3a664a18cdce2cd12807d5226484b71575ab7e57c2ddc473
BLAKE2b-256 checksum
How to use checksums
17de5061b97786f8fd30a074c47aabf7d91a4a307bdeb69b4741bf92452111fa
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp39-cp39-manylinux_2_28_x86_64.whl

Download URL maya_py-0.2.8-cp39-cp39-manylinux_2_28_x86_64.whl
Size 4.1 MB
Tags CPython 3.9 Linux glibc 2.28+ x86-64
SHA-256 checksum
How to use checksums
06e8ab38aa477d830e8a0708ceb34f3a486e37e533795f3d083626611c2c1f65
BLAKE2b-256 checksum
How to use checksums
ce20a418ce03419c4053f2bdd6a85251491465179ae3f94edc36f5f471120b4d
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release files / maya_py-0.2.8-cp39-cp39-macosx_11_0_arm64.whl

Download URL maya_py-0.2.8-cp39-cp39-macosx_11_0_arm64.whl
Size 2.3 MB
Tags CPython 3.9 macOS 11.0+ ARM64
SHA-256 checksum
How to use checksums
0eef887efd26eae37556d3c7ebdbb88fa523d6ee2dd7ea8502734ebdbac8ec16
BLAKE2b-256 checksum
How to use checksums
a0eba951d842f86114e67ee6453c2fc966c3ee5ccdb02eaac1977a53f4a34002
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/6.1.0 CPython/3.13.12

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Jun 20, 2026.

Transparency log

Release history Release notifications | RSS feed

0.3.1

19 release files

0.2.9

19 release files

This release

0.2.8 This release

19 release files

0.2.7

19 release files

0.2.6

19 release files

0.2.5

13 release files

0.2.3

19 release files

0.2.2

19 release files

0.2.1

19 release files

0.2.0

19 release files

0.1.6

19 release files

0.1.5

19 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page