wasm-gui: a WebAssembly Linux desktop (M1–M6) rendered by the native Rust client#104

Open

NathanFlurry wants to merge 28 commits into

mainfrom

wasm-gui-desktop

Commits on Jun 21, 2026

wasm-gui: M6 Xft + M6-INTERACTIVE desktop + M7 JWM + PNG proof (full M1-M7 suite green)

NathanFlurry
and
claude
committed
kernel: add open_pty_split (M6.3 PTY-spawn foundation)

NathanFlurry
and
claude
committed
sidecar: __pty_spawn stdio mode + __pty_read/__pty_write handlers (M6.3)

NathanFlurry
and
claude
committed
execution: wasm host_net pty_spawn/pty_read/pty_write imports + bridge bindings (M6.3)

NathanFlurry
and
claude
committed
wasm-gui M6.3: sustained interactive PTY shell session (echo/ping/exit cycles)

NathanFlurry
and
claude
committed
wasm-gui M6.3: real terminal emulator (suckless st 0.9.2) cross-compiled to wasm

NathanFlurry
and
claude
committed
wasm-gui M6.1: working X keyboard in the wasm server (precompiled keymap, no xkbcomp)

NathanFlurry
and
claude
committed
wasm-gui: correct the st live-typing root cause in SPEC (X-event-delivery, not deadlock)

NathanFlurry
and
claude
committed
wasm-gui M2.3: concurrent libX11 init works (concurrent launch mode + test)

NathanFlurry
and
claude
committed
wasm-gui: record st-typing investigation findings (keyboard path bypasses DeliverFocusedEvent)

NathanFlurry
and
claude
committed
wasm-gui: st-typing — eliminate 3 central dix delivery functions from the keyboard path

NathanFlurry
and
claude
committed
wasm-gui: st-typing root cause CORRECTED — timing/load issue, not routing (+ tracing caveat)

NathanFlurry
and
claude
committed
wasm-gui: st-typing isolated to window/connection specifics (+ xpoll-target diagnostic)

NathanFlurry
and
claude
committed
wasm-gui: st-typing crystallized — window events delivered, device events NOT (XI2/core synthesis)

NathanFlurry
and
claude
committed
wasm-gui: st-typing bisection — Xft init and locale eliminated; narrowed to terminal machinery

NathanFlurry
and
claude
committed
wasm-gui: st-typing — X layer DEFINITIVELY ruled out; cause is st terminal-core/run-flow

NathanFlurry
and
claude
committed
wasm-gui: st-typing isolated to xinit font/color loading (split tests; X path fully ruled out)

NathanFlurry
and
claude
committed
wasm-gui: st-typing — no single op is the cause (bisection-by-addition exhausted)

NathanFlurry
and
claude
committed
wasm-gui: st-typing — X-call layer exhaustively excluded; only protocol-trace remains

NathanFlurry
and
claude
committed
wasm-gui: st-typing DEFINITIVELY localized to CLIENT-SIDE (server delivers KeyPress correctly)

NathanFlurry
and
claude
committed
wasm-gui: st-typing refinement — forced read doesn't help; boundary is flush/wire/recv

NathanFlurry
and
claude
committed
wasm-gui: st-typing pinned to client-side host_net RECV (server flushes/sends correctly)

NathanFlurry
and
claude
committed
wasm-gui: st-typing full recv-chain trace; pinned to host_net net.poll/refill for st's connection

NathanFlurry
and
claude
committed
wasm-gui: st-typing complete stack trace — every component verified; X-server flush-timing frontier

NathanFlurry
and
claude
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

wasm-gui: a WebAssembly Linux desktop (M1–M6) rendered by the native Rust client#104

wasm-gui: a WebAssembly Linux desktop (M1–M6) rendered by the native Rust client#104
NathanFlurry wants to merge 28 commits into
mainfrom
wasm-gui-desktop

Commits on Jun 20, 2026

wasm-gui: M1-M6 desktop recovery — all 26 libs rebuilt, Xvfb/twm rebuilt, M4b/M5 tests pass

wasm-gui M6.4: fix sync-RPC fairness (net.poll cap 50ms->3ms)

wasm-gui M6.4 DONE: robust multi-app desktop (twm + xclock + libX11 window)

wasm-gui M6.1 (input half): host-driven input via XTEST proven

Commits on Jun 21, 2026

wasm-gui: M6 Xft + M6-INTERACTIVE desktop + M7 JWM + PNG proof (full M1-M7 suite green)

kernel: add open_pty_split (M6.3 PTY-spawn foundation)

sidecar: __pty_spawn stdio mode + __pty_read/__pty_write handlers (M6.3)

execution: wasm host_net pty_spawn/pty_read/pty_write imports + bridge bindings (M6.3)

wasm-gui M6.3: sustained interactive PTY shell session (echo/ping/exit cycles)

wasm-gui M6.3: real terminal emulator (suckless st 0.9.2) cross-compiled to wasm

wasm-gui M6.1: working X keyboard in the wasm server (precompiled keymap, no xkbcomp)

wasm-gui: correct the st live-typing root cause in SPEC (X-event-delivery, not deadlock)

wasm-gui M2.3: concurrent libX11 init works (concurrent launch mode + test)

wasm-gui: record st-typing investigation findings (keyboard path bypasses DeliverFocusedEvent)

wasm-gui: st-typing — eliminate 3 central dix delivery functions from the keyboard path

wasm-gui: st-typing root cause CORRECTED — timing/load issue, not routing (+ tracing caveat)

wasm-gui: st-typing isolated to window/connection specifics (+ xpoll-target diagnostic)

wasm-gui: st-typing crystallized — window events delivered, device events NOT (XI2/core synthesis)

wasm-gui: st-typing bisection — Xft init and locale eliminated; narrowed to terminal machinery

wasm-gui: st-typing — X layer DEFINITIVELY ruled out; cause is st terminal-core/run-flow

wasm-gui: st-typing isolated to xinit font/color loading (split tests; X path fully ruled out)

wasm-gui: st-typing — no single op is the cause (bisection-by-addition exhausted)

wasm-gui: st-typing — X-call layer exhaustively excluded; only protocol-trace remains

wasm-gui: st-typing DEFINITIVELY localized to CLIENT-SIDE (server delivers KeyPress correctly)

wasm-gui: st-typing refinement — forced read doesn't help; boundary is flush/wire/recv

wasm-gui: st-typing pinned to client-side host_net RECV (server flushes/sends correctly)

wasm-gui: st-typing full recv-chain trace; pinned to host_net net.poll/refill for st's connection

wasm-gui: st-typing complete stack trace — every component verified; X-server flush-timing frontier