summaryrefslogtreecommitdiff
path: root/build.zig
Commit message (Collapse)AuthorAge
* fix: a build with the addons compiled in (any off Linux, the static step) ↵Gabriel Schneider11 days
| | | | | | | and -Dpdf=false did not compile, Addon.zig importing a pdf_addon the build had not made: the builtin_addons options say whether the pdf addon is built, and without it the image addon is compiled in alone; an aarch64-macos -Dpdf=false compile no longer names pdf_addon (its one error left is daemon.zig's darwin CMSG_CLOEXEC, a break of its own) Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #51, builtins-json lists the shipped addons' words, read off the Zig ↵Gabriel Schneider13 days
| | | | | | | kinds' `Word` lines (image.zig and pdf.zig, handed to the generator and read by the staleness test), each `"addon"` first so the docs test's search for a builtin never meets one and an addon word the docs name is checked against the sources (renaming PdfFit fails it); the glossary says whose addon a word is, and the changelog Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #50, macOS by pattern: off Linux every core module has the addons ↵Gabriel Schneider13 days
| | | | | | | compiled in (Addon.zig's loader is a Linux memfd, now refused elsewhere at compile time), the native shell registers them at start as the tui and gui do, and `-Dmacos-sdk=<path>` hands SDL's build the SDK's paths, so `-Dplatform=gui -Dtarget=aarch64-macos` configures (and `zig build core-test -Dtarget=aarch64-macos` compiles its test binary, MuPDF and both kinds in, without running it) Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #47, MuPDF and FreeType lazy: Zig 0.17's configure is handed the -D ↵Gabriel Schneider13 days
| | | | | | | options and never the steps asked for (the configurer's argv, and the configure cache keyed on it), so no step can fetch for itself; the guard is `-Dpdf`, on by itself once MuPDF is fetched (std.Build.package_map's `available`) and off Linux, which asks `b.lazyDependency` for MuPDF, and FreeType (with the zlib and libpng it brings) is lazy too and asked by it or a window: a fresh checkout's `zig build -Dplatform=tui` fetches none of them, and without `-Dpdf` the pdf, addons, static, mupdf-check, pdf-harness and bench steps say to add it; building.typ, install.typ and shared-libs.md say so Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #43, the PDF kind in Zig: addons/pdf/pdf.zig replaces pdf.c over the ↵Gabriel Schneider13 days
| | | | | | | same pdf_bridge.c (its header translated by build.zig, Zig 0.17 having no @cImport), one place a pane (a page and how far down it, which where says and address takes) laid out by each view its own way (a view with pictures its own strip, rasters and thumb by its id, a view of glyphs the page's text layer), the view of the last input the one whose window's top the place follows; the pages' sizes, the page's text and its hits, and the search over every page on the lanes as jobs (rows streamed through found, a stale search cancelled, its first hit gone to as a Look of its first row went), the tick asking and laying out only, open's one parse and close kept on the loop (0.4 and 0.8 ms; a truncated file stays a refused Look); a band at most PA_RASTER_MAX; the static build reads the module's pardes_addon, exported only from the library Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* Build: a word that prints this build into a +Build pane (SPC h b): the ↵Gabriel Schneider13 days
| | | | | | | version line, the commit, the jj change and the build time (stamped as the commit is: release builds and the ~/.local install, else "not stamped"), the zig that built it, its mode, target, platform and grammars, whether the addons are compiled in, the wire's semantics hash and the dependencies; build.zig stamps the change (jj log -r git_head()) and the time (date -u) through one command helper beside gitCommit; a T1 case reads the pane Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #35, the core's seam for ABI 3 (Kind.zig): each pane of a kind with ↵Gabriel Schneider13 days
| | | | | | | view told each open view's facts by id when they change (the slot's view table, a hash a view, a stopped view told cols 0, and a step woken for a view a frame's layout changed) and ticked after, the tick's pa_tick and its prefix (placed and retag gone: the art words go on every pane of a PA_PICTURES kind from the first frame, the same bytes whichever view's step lays the tag out, by the session's first view, which was the 30 ms soak's rollbacks), found's rows the search's or appended to its +Search after, cancel's tokens dropped on every core (on the tape), a picture over PA_RASTER_MAX dropped and said once, a record of at most PA_RECORD_MAX, a Debug call over 16 ms of the kind's own (its allocations and job copies the core's) said once a site and failing --addon-test (which gains view N on|off and prints the tag the core makes), and the record §17.5's: a Restore, an Addon load and a reload read the pane's file as a Look does, a file not there a placeholder saying <path> is not here, record kept, opened by the next Look or watch of it Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #33, the gui with the addons in it: `zig build static -Dplatform=gui` ↵Gabriel Schneider13 days
| | | | | | | makes a pardes-gui with image and pdf compiled in and registered at start as the terminal's musl static build has them (the gui linked dynamically, as SDL and the GPU need the display's libraries; the static step's module is a gui shell too, wired beside root_mod) Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #27, life: addons/life/life.zig is a second shipped Zig addon, ↵Gabriel Schneider13 days
| | | | | | | Conway's Life in a pane written clean-room against src/kinds/addon.zig and the header alone (steps ten generations a second while focused, space pauses, r reseeds, hjkl and x edit, the prefix the generation and alive count, the body the board, dump and load the board), with its script and golden under addon-test and its own zig tests; the module's timing test and the dogfood that found Kind.step's stale tick clock Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #26, one static pardes: `zig build static -Dplatform=tui ↵Gabriel Schneider13 days
| | | | | | | -Dtarget=x86_64-linux-musl` (musl's, as glibc links only dynamically; never the default, and -Dstatic gone) links the terminal binary statically with both addons compiled in, image.zig as a module and pdf.c with MuPDF, its export renamed by a -D, which Addon.zig registers at start as if loaded (/addons says builtin, Addon reload refuses one, no library is scanned), and execveat and close_range go by their syscalls, which musl has no wrappers for; macOS's way to image and PDF panes too, say a comment, building.typ, shared-libs.md and the changelog Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #25, addons in Zig: src/kinds/addon.zig, over the translated header ↵Gabriel Schneider13 days
| | | | | | | and nothing of the core, makes a kind's pa_ops from a struct of the functions it has (its head from name, version, extensions and words; a public function of no kind's name a compile error), the pane's arena and a job's pa_out as std Allocators, typed keys, a Window to put text in, a Paint for the prefix and placements, tick's flags and a Deadline, dump and load as an extern struct's bytes in one line each, and body and ctl over a Writer; snake.zig and image.zig (stb_image in stb.c, static, beside any zstbi) replace snake.c and image.c with the same goldens, harness and perf cell, and addon-test runs their own zig tests Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #24, the addons as build steps: `zig build image` and `zig build pdf` ↵Gabriel Schneider13 days
| | | | | | | each build their library and install it in <prefix>/lib/pardes/addons (and ~/.local/lib/pardes/addons with the home install), `zig build addons` both, the default install the binaries alone and -Dmupdf gone, MuPDF compiled only for the pdf step and the tests that depend on its artifact as before (fetched at the first configure, which Zig runs alike for every step); install.typ, building.typ, reference.typ, shared-libs.md, the README and the changelog say so Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #13, pdf.so: addons/pdf/pdf.c is the PDF kind against ABI 2.1 (its ↵Gabriel Schneider13 days
| | | | | | | record the file's bytes, renders on jobs over two MuPDF lanes a cookie aborts, its strip laid out at `at` with crops and repainted rows for hover and selection, its text the body, search, links and PdfSections through look and text, P:F S places, PdfFit, PdfTint, PdfSections, its keys), pdf_bridge moves beside it, and the build makes it from position-independent MuPDF, FreeType and zlib exporting pardes_addon alone, installs <prefix>/lib/pardes/addons/pdf.so, -Dmupdf now builds it or not and mupdf-check runs its addon-test script Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #10, image-frame: the medium perf set gains a frame with a 640x400 ↵Gabriel Schneider13 days
| | | | | | | picture's pane (the image addon's) alone in the window; 163 us in Debug against render's 89 (the Debug paint-purity hash is over the kind's memory, so there it costs the picture's bytes: 1478 us at 2560x1600), 46 us in ReleaseFast against render's 81, of which the kind's clear and scan of its 4,366 cells are 2.2 us (2.7% of a text frame, under §16.1's 5%); test/perf-baseline-tui-debug.json and -gui-debug.json re-keyed to the harness and given the cell Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #9, the core's image code gone: src/image.zig (558 lines) and the ↵Gabriel Schneider13 days
| | | | | | | static table go, so no kind is compiled in (Registry.init, Kind.image, compiledIn, `static` in /addons and "is compiled in" with them) and a test of the core's side of a kind's pane opens Kind.test_kind (a picture of one pixel, no library); a placement's fit (NativeGeometry, containGeometry) moves to surface.zig and a PDF's fit modes to pdf_view.zig; a terminal's kitty PNG decodes in TerminalImages.zig through wuffs as before, its downscale through stb_image_resize, the one part of zstbi the core keeps (stb_image patched out of it, the wuffs_images option, tag_image and image_exts gone); the image cases load the built image.so by path: test/panes.zig's and fs_namespace's through Addon.perform, T1 through Addon load (a new behaviour case: PNG, JPEG and PPM through 9P, the art words, a refused corrupt PNG, a dump's placeholder and its picture back, which replaces the decode unit test and ctl.zig's corrupt-image case), the snapshot, process and image harnesses by linking it into each session's data directory; images.snap unchanged Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #8, image.so: addons/image/image.c (144 lines) is the image kind ↵Gabriel Schneider13 days
| | | | | | | against ABI 2 over stb_image as zstbi vendors it (every format it decoded, HDR out, at most 1280 pixels a side, 64 Mpixels refused before decoding), its record the file's bytes, its decode a job, one placement fitted, its prefix img; stb's memory is the pane's arena at open and the job's on a worker, so pa_out's alloc may be called as often as a job needs (an arena a job, freed after its job_done, the last allocation the answer: what the header found wanting); the build makes it with -Wall -Wextra -Werror and installs it to <prefix>/lib/pardes/addons/image.so (a bare build to ~/.local/lib/pardes/addons too); the compiled-in image kind is still there, so the start scan lists the library refused as a second image Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #6, the snake: addons/snake/snake.c (153 lines) against ABI 2 steps ↵Gabriel Schneider13 days
| | | | | | | only while its pane has the keyboard and asks for nothing otherwise (19 lines of boilerplate, 13 by §16.7's measure: the head, word and ops literal, the api kept, an alloc check and two of prefix; the body pair and open's unused arguments are the rest), built by the build with -Wall -Wextra -Werror and not installed, and addon-test runs it against addons/snake/snake.out: its moves, h j k l turns and no turning back, food, the ctl read and body, blur with settle asking for no ticks and the snake still, focus resuming, a dump into a fresh pane, death on the wall and a restart Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #5, ABI 2 and the seam: src/kinds/pardes_addon.h is ABI 2 as ↵Gabriel Schneider13 days
| | | | | | | architecture §16.9 has it (pa_window 24 bytes and no art bytes, pa_placement 80 with at, from and dirty_y/dirty_h and no stride, pa_word 32 with no size and PA_WORD, tick(st, now_ns, flags) with PA_FOCUSED, PA_OK and the conventions, the dump and load rules, pa_put with attrs, pa_copy and PA_POD, whose fragment gcc -Werror refused as written and is ternaries here) and fake.c and image's table speak it; tick is called at its deadline and once after open, a taken key or click, a word, a job's answer, a load and a change of PA_FOCUSED (the pane has the keyboard in some view, its window focused), nextWake folds the deadlines, and a Debug build hashes the kind's allocations across paint and panics naming it; paint runs once a frame, the slot keeps its prefix and whether it placed a picture, the tag is that prefix, the core's art words when it did and the path, and a frame that changes it owes a step to lay the tag out again; Addon load and reload are an addon effect the host performs (memfd, BLAKE3, dlopen, the head checks and init, the bytes a process holds shared and counted) answered by an addon event carrying what the library brings as data, whose update does the collisions, the reload's validate-then-swap at a new generation and the waiting panes deterministically, while unload is the update's alone and lets the library go by effect; a kind's job is Effect.job run on the host's workers (host_io.KindJob: tui and gui on Io.concurrent, macOS and tests on the loop) answered by Event.job, stale generations dropped and a library closed after its last job; each core has its own Registry (16 kinds, their words, leaders and extensions, digested and dumped; their code a local side array a replica has none of), so a test session no longer unloads the world and Restore and -l keep the process's kinds; every call into a kind whose answer the core acts on is taped as an Outcome (its answer, refusal, message, Look, text pane) and the prefix a step reads is taped too, so a replica's core, which runs no kind's code, steps alike; a pane of a kind on a replica is a Shadow drawn from the authority's frames, painted for the peer's view and sent by Screens as row hashes with the rows that changed and each raster once a key (screen_state.Shadow), its keys going to the authority; the tail, reload and the Addon word and event are Addon.zig's, the slot's record dump.zig's, a tail word's run exec.zig's; --addon-test gains focus, blur, dump, body, ctl and the window line, its jobs done as a host does them; fake.out gains its window line, loses a body row once the tag carries the art words and lands FakeJob at once; tests rewritten: the image dump test (the replica-style load keeps the record), the Petscii/Palette/Ascii tag test in ctl.zig and test/panes.zig's image tag test (a picture decoded and shown first), fs_namespace's decode test (the job performed as a host would), Kind.zig's registry test and Addon.zig's crash-mark test, the addons suite's ABI numbers; new: a job under way across a reload (T1), a Shadow's frame codec, and a replica with no code for a kind showing its pane and passing its keys; gates: core-test 958/958 and behaviour-test 249/249 on both, unit-test 1490/1491 tui and 1533/1534 gui, pane-test 171/172, snap all 105 ok, process-test 0 of 37 failed on both, addon-test ok, hxdiff 19 and hxparity 6 mismatches by the base's names, hxgolf 0, replicas, gui-golden, monkey-gui, daemon.py main and window-clients, soak local pass, perf medium within 1% of the base Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #4, the fake addon and the addon test: addons/fake/fake.c is the ↵Gabriel Schneider13 days
| | | | | | | template, every function of src/kinds/pardes_addon.h in a few lines (a counter its keys, clicks, words, ticks and jobs move, a 2x2 picture, a body, an address, a ctl read, a dump and a load), built by the build with -Wall -Wextra -Werror four more ways for the loader to refuse (another ABI, no pardes_addon, a word that is the core's) or reload (no record of its own); `pardes --addon-test <lib> <script>` loads a library into a session of its own (80x24 unless `size`, `seed N` for the panes' seeds, jobs at once, a clock moved only by `tick` and `settle`, which ticks until the kind asks for no frame) and prints the pane's tag, cells and placements after each line of keys, clicks and words; the addon-test step (and behaviour-test) runs it over the fake against addons/fake/fake.out; T1 (test/behaviour/addons.zig): Addon load, /addons with its path, BLAKE3 and panes open, the words in /commands and Help, a word refused away from its kind, unload refused while a pane is open, a compiled-in kind not unloadable; a library of another ABI, one with no export, one whose word is the core's and a file that is no library each refused and listed with why; a pane of the kind taking keys, a click, a word, a job, its body and ctl read, Esc leaving it; a dump restored with the addon, without it (the placeholder, a word refused there), and the addon loaded again reopening it in place with its count; a reload refused for a half-built library with the old one running on, one taking the state across by dump and load, one with no record coming back from its words' last values; a Look by an addon's extension; a test session lets go of its addons at its end (the registry is the process's); gates on the head: core-test 961/961 and unit-test 1494/1495 tui, 1538/1539 gui, behaviour-test 250/250, pane-test 171/172, snap all 108 ok, process-test 0 of 37 failed on both, addon-test ok, hxdiff 19 and hxparity 6 mismatches (the base's), hxgolf 0, replicas, gui-golden, monkey-gui, daemon.py main and window-clients, soak local pass Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* kinds #2, a pane of a kind is one slot and image is a kind compiled in: ↵Gabriel Schneider13 days
| | | | | | | src/kinds/pardes_addon.h is the kinds' C ABI (the cell, the window a kind paints, placements and the tag prefix it hands back, keys, the mouse, pa_api of six functions over the pane's arena, words, the head and pa_ops of paint and sixteen optional functions), translated once by the build into the core's records, never mirrored; src/Kind.zig is the core's side: the kinds this process has (a registry with the checks a load makes: the head's size and abi major, a paint, a name, words, leaders and extensions none else has), a pane's Slot (the kind's state and arena, its path, the art bytes Petscii, Palette and Ascii set, its tick deadline and jobs) and the addon arm at the sites: Look and the initial file by extension, draw (cells into the grid, placements keyed by serial and the kind's key), the tag prefix from a paint over no cells, nextWake and the step (ticks, jobs' answers landing on the loop), close, Recent, paneDir, /index's kind word; a dump writes a slot by hand (kind name and version, path, where, art, the kind's own record) and loads it through the kind's open and load, or as a placeholder that keeps the record when the kind is not here; image.zig's pane is now a static pa_ops (open refuses what it cannot read, the decode a job whose answer lands at a step instead of a thread polled by draw, one placement fitted by the frontend, its record the file's bytes), Pane.image is Pane.addon, and the three art words act on the slot; no golden changes (images.snap and tagbottomimage.snap pass, the tag prefix and the placement key are the same bytes); tests: the image dump test, the Petscii/Palette/Ascii ctl tests, the image tag-prefix test and fs_namespace's decode test rewritten on the slot (the decode now lands at the step after the open), plus a registry test; gates: core-test 960/960 on both, unit-test 1487/1488 tui and 1531/1532 gui, behaviour-test 244/244, pane-test 171/172, snap all ok, process-test 0 of 37 failed on both Co-Authored-By: Claude Fable 5.1 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* build: xitdb, the database pardes's disk state is to go through ↵Gabriel Schneider14 days
| | | | | | | (architecture §14), vendored as every dependency is: `zig fetch --save=xitdb` of the tarball of github.com/xit-vcs/xitdb at its pinned commit 6b93af1a (Zig 0.17, MIT, no dependencies) into build.zig.zon and zig-pkg, and imported by every core module (wireCore); pure Zig, so there is no C library to share and no allocator hook to set (it allocates through the allocators it is given), its row in docs/shared-libs.md; nothing uses it yet Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* effects out, the theme crossfade: a theme change takes the chrome at once, ↵Gabriel Schneider14 days
| | | | | | | as it always took the content; colors.ChromeAnimation (its OKLab steps, its retarget, the ticks and the frames it asked of nextWake) goes, Pardes.chrome is the theme's chrome, set as the theme changes, and the `animate` flag goes from loadThemeFile and applyThemeEffect with it, as does the build's -Dtheme-animation option (and its place in the wire's fingerprint); no frontend fades a theme of its own (a window crossfading every cell's colour for 160 ms is a copy of the last frame kept and blended, for one keystroke's change); the fade's tests go, the PDF test loses its ticks through a fade, and `core animation keeps wall time at 60, 120 and 144 Hz and across a stall` goes with the one animation that ran it; the perf baselines lose the option from their build identity; GUI goldens 16-debug and 17-terminal change, the golden session having faded its themes by default: a fade's frames kept the Debug word's announcement up in the picture's pane, which gave up a row to it under the Debug box; the tree before this change built with -Dtheme-animation=false draws exactly these frames, all sixteen byte for byte Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* effects out, the post chain: Crt, Bloom, Vignette, Grain, Shader and ↵Gabriel Schneider14 days
| | | | | | | ShaderAnimation go with src/gui/Post.zig, src/ShaderBuild.zig (the run-time glslc spawn and its watched files), shaders/post/, post.vert, bloom-down and bloom-up and their prebuilt .spv, the window's scene texture and its post-chain-only redraw (level A), the disable_scene_effects tape site and Pardes.disableSceneEffects; EffectCode goes with src/effect_sources.zig, which only embedded the effects' sources (/virtual keeps src/); the corner-click test keeps its point without a chain, the window monkey loses its Bloom axis; macOS's post-processor (ScenePostprocessor.swift, crt.ci.metal, pardes_scene) is a change of its own Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* targets: the DOM web shell goes (src/web.zig and src/web/, its harnesses ↵Gabriel Schneider14 days
| | | | | | | web_harness, web_snapshot, web_driver_test and web-snapshots/, tools/web_fixture.py, -Dplatform=web, -Ddump, and the web, web-harness, web-snap, web-e2e and web-driver-test steps), and with it what only a freestanding build needed: `hosted` and every guard on it (every build has a host now), memory's stack-fallback arenas and reduced_target sizes, the emscripten main, the tree-sitter libc shim and the stb stub, lsp's has_client, the web arms of the platform switches (-Dplatform is tui, gui or macos); the macOS files pass zig ast-check; docs/web.md is a paragraph on what it was Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* targets: the ESP32-P4 board port goes (src/esp32p4*.zig and src/esp32p4/, ↵Gabriel Schneider14 days
| | | | | | | -Dplatform=esp32p4 with its pinned riscv32 target, -Desp32p4-cols and -rows, the Peek, Poke, Hexdump and Gpio words and the Board behind them, the board's 9P engine options, the gpio_toggle host slot), and with it every guard and limit only the board needed: terminal_panes and terminal.enabled (ghostty is in every build), memory.limits' board branching (the desktop values stay), MAX_PANES 64 and MAX_COLS 16 fixed, the board's boot buffer and its embedded boot, the pager's and the terminal's emulator-less paths; tests: the board's GPIO, ring and Board ones go, the boot-OOM and transition-grid ones stay under their own names; the perf harness loses the board's fixture, so the Debug baselines of both frontends are recorded again (--only medium --reps 10) Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* test: T1 asks the language words' servers, the suite next-tests.md said was ↵Gabriel Schneider2026-11-05
| | | | | | | missing: a Session made by initLanguage performs each lsp effect as a host does (host_io.Lsp, on the session's thread and at once; host_io is pub in the core for it), a .rs file's server the scripted one (test/lspmock.zig, built for the session's module as for the snapshot scripts), and test/behaviour/callstack.zig pins the guide's Call hierarchy through it: the tree +Callstack shows, and CallstackDepth bounding the walk Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* build: a unit-test program is wired by one Units.add (test-build, unit-test, ↵Gabriel Schneider2026-11-05
| | | | | | | its one filter check, and a step of its own with its own check), where pane, behaviour, kitty, deck, shell, the freestanding roots and the perf and fs bench tests each spelled it out Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* build: vendoredThemes lists vendor/themes once, into a list it sorts, where ↵Gabriel Schneider2026-11-05
| | | | | | | it read the directory twice (to count, then to fill); the comment on a file read while configuring moves to configureRead, which it is about Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* build: the 9p-io-test and fs-unit-test modules are two rows of one loop, ↵Gabriel Schneider2026-11-05
| | | | | | | where each copied the core's link objects and imports by hand Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* build: src/pdf.zig's module over MuPDF is made in one loop for the core and ↵Gabriel Schneider2026-11-05
| | | | | | | for pdf-bench at ReleaseFast, where it was spelled out twice, and the four PDF benches are rows of one table, where each wired its executable, options and run by hand Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* build: the perf and fs bench programs' tests, which compile no editor, run ↵Gabriel Schneider2026-11-05
| | | | | | | with unit-test, and their perf-test and fs-bench-test steps go; lspbench-test and hxdiff-test go, with lspbench's three tests, which only that step ran, and hxdiff's --self-test (hxdiff's own four tests stay: test/panes.zig runs them in pane-test) Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* build: four steps nobody runs go (lspbench-check, which is `zig build ↵Gabriel Schneider2026-11-05
| | | | | | | lspbench -- --check`; syntax-bench, which is `zig build syntax -- --bench`, the syntax step rerun each time now; unit-profile; history-test, test/history.zig's tests still run by unit-test) Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* wire.zig: the replica's wire codec moves from src/detached/, a directory of ↵Gabriel Schneider2026-11-05
| | | | | | | one file since the phase 3 frontends went, to src/wire.zig; its imports and the paths that name it follow Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* 9p_io, build: 9P over QUIC is gone: src/9p_quic.zig, the Listener's poll ↵Gabriel Schneider2026-11-05
| | | | | | | path on the editor's thread (its accept, fill, flush and watcher thread), the Client's quic! dials, -Dquic with its OpenSSL linking, the quic-test step that ran no test, Namespace.quic_address and /listeners' quic row, the QUIC process case, the perf baselines' quic key and every mention in the docs; main.zig still parses --9p-quic until the processes change takes it out, so 9p_io.quic_enabled stays, false, until then Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* test: the window's own tests in Zig (T4, test/process/window.zig), ↵Gabriel Schneider2026-11-04
| | | | | | | gui_monkey.py and gui_golden.py gone: the monkey drives the hidden window's own input (Detached.typeKeys into its pty, Detached.mouse's OSC 777 events) with clicks, drags, wheels, hovers and keys, and columns, looks, Joincol, Collapse, Delcol and themes over 9P, then the window must settle, alive, with no crash record (two combinations of 30 steps in process-test, all 40 Motion x Lift x Bloom with zig build monkey-gui -Dplatform=gui); the goldens hash the 17 scenes' frames (Detached.capture, a transition's frame 6 too) against test/gui-goldens.txt, regenerated for the offscreen window and twice reproduced, run when asked (zig build gui-golden -Dplatform=gui, PARDES_GUI_GOLDEN=update); Detached's init file, settle and capture; no Python step left in the build Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* rename: the terminal frontend is the tui (-Dplatform=tui, src/tui/tui.zig, ↵Gabriel Schneider2026-11-04
| | | | | | | --tui with --tty kept as its old spelling, the process tests' backend .tui and --backend=tui, docs/typ/builtins-tui.json, the Settings report's platform), and the builtin Tty is Term (its registry entry, SPC n t, every default tag, the docs, tests and snap scripts), Tty still run as Term through exec.acme_words, Tty+fish too on a tag and a ctl; tty as a pane's raw mode and as the OS's terminal unchanged Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* test: the process tests run on each backend (the user: an effect's test is ↵Gabriel Schneider2026-11-04
| | | | | | | its backend's): zig build process-test -Dplatform=tty drives pardes, -Dplatform=gui drives pardes-gui hidden (PARDES_TEST, SDL_VIDEO_DRIVER=offscreen, no display of the user's touched) on a pty of its own and a detached session's 160x50 grid, its frames and status in a capture directory of its own, Detached.frame() waiting for it to say it is idle; the runner told its backend (--backend, from the build), each line marked [tty] or [gui] with its time, a case that skips said so; the terminal frontend's and a detached session's own cases (pardes -, a killed tty editor, a signalled detached session, a message expiring with nothing attached) process.only(.tty), a detached session's size checks and pty pixel cells the tty's alone; the canary tty-only for now (on the window its Newcol is refused after the terminal steps, docs/next-tests.md); tty 36 of 36, gui 31 and 5 skipped Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
*-. merge the Python port (fsa, fsb, mky): fs.py's 38 cases in Zig, 36 process ↵Gabriel Schneider2026-11-04
|\ \ | | | | | | | | | | | | | | | | | | tests (test/process/: tree, terminals, root_ctl, edit, flags, mounts, builtins, repls, look_and_exec, other_files) and a few at T1, monkey9p.py as test/monkey9p.zig at T1 (monkey-session, its smoke in behaviour-test), each checked against breaks by its porter and a reviewer; the runner's helpers from both fs.py halves joined (one replace, one newPane, the look that takes a source); the fs-test, fs-discovery-test, monkey-9p, monkey-9p-smoke and agent-session-test steps gone, fs.py kept as the developer tools' helpers; behaviour (221), process (36), core, fs-unit and pane tests, the gui build and all 113 snap scripts pass Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
| | * test: the 9P monkey at T1 and the agent driver's interaction checks at T2, ↵Gabriel Schneider2026-11-04
| |/ | | | | | | | | | | | | for the Python ports: test/monkey9p.zig (was test/monkey9p.py) drives an in-process Session with random 9P requests, every op drawn from the seed alone (panes, columns, stale serials and held files as tokens resolved when the op runs), so --seed/--steps replay a run, --only replays any subset of its steps and a failure shrinks by delta debugging over them; each run is a child of the program in a throwaway root (its HOME, XDG directories and dump), so a panic or a hang is a failure reported and shrunk like the rules' (one failure rule, close runs only line files, /index matching /pane, chunked reads equal whole, the log marker); the 9P engine's own refusals and OTRUNC's truncation made by the monkey, Session doing no effects; steps monkey-session and monkey-session-smoke, the smoke part of behaviour-test; Session's mky helpers walk, ask (a raw request answering errno, held or bytes) and ended; test/process/terminals.zig, three cases of test/agent_session_test.py (a prompt that settles, input shown and ^U taking the screen back; history past a screenful in the body; a raw-mode control byte through pty/data), the driver's own argument, output, timeout and cleanup checks left in Python; monkey9p.py deleted; each checked to fail against a break: serveFs not noting a pane file's err (one failure rule), noting it twice (one failure rule's upper bound), a msg beside a refused pane write (its msg clause), /pane leaving out serial 3 (index matching pane dirs), stagedReply skipping a byte past offset 0 on small reads (chunked equals whole), a panic in pty/run's write (a crash, shrunk 377 -> 3 ops), the root ctl taking an unknown word (the log marker), pty/data dropping a lone control byte (raw key), a 4 KiB terminal scrollback (history, raw key), pty/data dropping a write's last byte (all three terminal cases) Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* / Tty9p and its pardes-v9fs helper removed (the user: a one-off), a terminal ↵Gabriel Schneider2026-11-04
|/ | | | | | | with the session kernel-mounted through Linux v9fs: the builtin and its SPC n 9, the spawn's mounted path and the pane's flag, the host's helper launch, src/linux/v9fs.zig and its build and install, test/v9fs.py, v9fs_driver_test.py and v9fs_terminal.py with their three steps, the docs (the build notes keep how to mount a session by hand); the builtin lists regenerated; core, fs-unit, pane, behaviour and process tests, the gui build and all 113 snap scripts pass; ~/.local/bin/pardes-v9fs, installed before, is left Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* test: review of the core migration's tests: the Pet test takes its branch ↵Gabriel Schneider2026-11-03
| | | | | | | from the build's platform (pardes_config, imported into the behaviour module, its options alone), not from whether Pet was taken, so a terminal's view taking Pet now fails it; New's absence checked in a file pane's tag and a scratch's made by New, not only in 9P's +New twice; the emptied column's tag shown to have the keyboard (typed into); the pane's tag drawn on the row under a one-row column tag; each checked to fail against a break (Pet without in_window, New in the file pane's tag, an emptied column's focus given to the workspace tag) Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* test: the behaviour-test step and test/session.zig, T1 of ↵Gabriel Schneider2026-11-03
| | | | | | | docs/next-tests.md: a Session drives the core as a 9P client does (walk by path, open, read or write, clunk, held lines run at the clunk as src/9p_io.zig runs them), the core opaque and the step's module importing session alone, so a behavioural test cannot reach past the 9P tree; keys (kbdin), click (on text found on /screen, mousein), screen, focused, refused and the refusal's words; test/behaviour/{other_files,panes,edit}.zig, a file a section of the reference: kbdin and mousein, a collapsed pane's ctl, name's leading blank, Putall, Edit blocks, moved out of src/ninep/ctl.zig (an Edit block left open now checked as a client sees it, an err in the log at its clunk), each checked to fail against a break (collapsed left off the ctl read, Putall saving nothing); the plan doc says the layout, the decisions and the progress Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_016Y9MNAtZenFv4weoMD3cge
* c_heap: a heap's allocator read as its bytes are, through a volatile ↵Gabriel Schneider2026-10-30
| | | | | | pointer: std's stateless allocators (c_, page_, smp_, wasm_allocator) leave `ptr` undefined, and one never written outside a test (replica.zig's zstd heap over c_allocator, web.zig's over wasm_allocator) is a constant to a release build, which folded `known.ptr == current.ptr` to false: every malloc named a new allocator and the 65th panicked (the box's session aborted at about its 16th far join; the core's table held 64 copies of {ptr 0, c_allocator's vtable}); its test runs built ReleaseSafe in unit-test too, where it panicked before Co-Authored-By: Claude Opus 5.5 <[email protected]>
* replica: a far client's content cache and zstd: the session lists its join's ↵Gabriel Schneider2026-10-30
| | | | | | blobs by BLAKE3 hash first (wire 16 `blobs`), the client finds each in the core it leaves or on disk ($XDG_CACHE_HOME/pardes/blobs, hashed again on load, a bad file a miss, the least recently used evicted past 1 GiB) and asks for the rest (`want`), and everything after the welcome goes in zstd envelopes (level 6 over 8 MiB for the join, an undo snapshot being a near copy of its text, then level 3), so `ssh -C` leaves the default command; a far client (through ssh, or PARDES_REPLICA_FAR) keeps what it was sent and, at its end, what it made; a hangup ends a remote client the normal way; through the box a client holding the blobs is drawn in 2.4 s where main took 7.0 to 11.9, and a first join of a session with undo in 7.1 s where main took 11.9; libzstd 1.5.7 vendored, its allocations on c_heap, a second copy refused by shared-libs-check; tests for the cache's verify-on-load, its eviction and a far join's blobs sent once; test/daemon.py's far clients Co-Authored-By: Claude Opus 5.5 <[email protected]>
* web: a dump is a .dump by name everywhere (the download, the embedded ↵Gabriel Schneider2026-10-29
| | | | | | fixture, -l's help); the touch check resizes a pane by its grip, there being no rule between panes Co-Authored-By: Claude Opus 5.5 <[email protected]>
* web: the browser shell builds again and loads a dump of a native build: a ↵Gabriel Schneider2026-10-29
| | | | | | | | | | theme by its name, a field compiled out passed over, no disk asked; touch.dump replaces the hand-written ZON fixture Fixed for wasm32-freestanding: counts as usize, no threads, stb_image, tape or host stat there. touch.mjs follows the renderer's 13 tag-layer fields, helix's retuned colours and the tag column. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* dump: one walker dump of the whole session, by name across builds, for ↵Gabriel Schneider2026-10-29
| | | | | | | | | | | Restore, -l, Handover and the store; dump.zig's hand-written schema gone A dump is walk.zig's walk of the core with a table of every type it reaches (fields, enum and union names, number widths), read back by name; terminals are ghostty's snapshot (screens, history, kitty images); a failed load frees what it made. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* phase 5 part A: the speculation and rollback engine, in one process: a ↵Gabriel Schneider2026-10-29
| | | | | | | | | | | | | | | | walker copy put back where it was, replicas that apply their own input at once and take the authority's tape as the truth, and a harness of typists over delayed links src/rollback.zig: the replica core's allocator (`Epoch`) knows every allocation, and while a copy is held puts off frees of what was there before and refuses to grow it in place, so what the core held stays where it was. The copy (`Copy`) is the digest's walk saving the bytes of every allocation it reaches, field by field (the derived caches and this process's own too: a safe build's free writes over what it frees, and the frame arena points at what the epoch then frees), but the queues a step drains and the frame statistics; a file's text, its line index and its undo snapshots are pinned instead (made anew at each edit, never written; now freed raw everywhere), checked unchanged in a safe build. Put back, each allocation's bytes go back over it, the epoch's allocations are freed and its put-off frees forgotten; let go, they are made. Tree-sitter's trees, the C heap's, are held the same way (syntax.held_trees). src/replica.zig: an `Authority` records its core's every input to a tape in memory and hands each client's input back as a unit (origin, sequence number, its tape); a `Replica` applies its own input at once, as the authority will (update, then pump), recorded by its own tape (tape.zig's speculation mode), and for each unit in order drops its pending head when the authority recorded the same (inputs, host calls and answers: the fast path), else puts the copy back, replays the units since and this one (every digest checked) and applies its pending inputs again. A host call or a syscall stops a speculation: it is answered harmlessly, undone when it returns, and the input waits for the authority's answer; a clock or environment read is a guess the comparison checks; an emulator is the authority's (terminal.zig's writers stop a speculation; a safe build asserts none changed). The view's events (a resize, the pointer) wait. Messages are bytes: no transport yet. Each client has a view of its own (src/views.zig): the authority goes into it before the client's input (`toView`, taped), the replica's speculation the same, and the copy walks the views not live too (`parked`, each pane's `pv`). test/replicas.zig (`zig build replicas`): one authority and N replicas, a view each (`--shared` puts them in one), each client typing monkey.zig's mix at 50-200 ms over its own link (half the RTT plus 10% jitter, in order), the authority's pty echoing 2 ms later; gates: no replica diverges, every replica's digest is the authority's at quiescence; reported per RTT: inputs applied at once, stopped, waiting, fast paths, rollbacks (mispredicted, foreign), and what a speculative apply, a copy and a rollback cost. 3 clients x 100 inputs at 30, 80 and 207 ms: no divergence over 12 Debug seeds and 20 ReleaseFast ones with a filled heap; ReleaseFast p99 a speculative apply under 0.2 ms, a rollback under 1.6 ms. `zig build rollback-bench` times the real copy now (take, release, an edit, restore). What the replicas found, fixed: a release build reads what the heap had where Debug's fill made two cores agree: `col_panes` and `restore_buf` digested undefined (zeroed now), `terminal.State.reply_dropped` never set at create, `Text.sels` and `PendingPipe.ranges` digested past their length (a walker `len` cuts a slice too). PARDES_REPLICA_FILL fills a replica's allocations to show what is left. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* rollback-bench and fork_stall.py: what a replica's rollback copy is (a walk ↵Gabriel Schneider2026-10-29
| | | | | | | | | | of ~100 KB once file texts are pinned) and what a fork stalls the authority for, which settle open points 3 and 5 `zig build rollback-bench -Dplatform=tty -Doptimize=ReleaseFast` (test/rollback_bench.zig) builds a typical session (a shell with 1 MB of scrollback, three 60 KB files and a 3 MB one, 40 edits) and a big one (14 panes, 18 MB, 4x3 MB and 8 small files, 8 MB of scrollback, 200 edits) in process, then measures the state the digest walk visits, the part of it that is a file's text and its undo snapshots (immutable once made: an edit puts a new buffer in place, a snapshot is a copy, so a copy pins them), and the walk again with those taken out, plus a memcpy of what is left. Typical: 36.3 MB walked, 80 KB to copy, about 0.02 ms; big: 282.5 MB, 96 KB, about 0.045 ms. Then a fork of the big session (431 MB RSS: 3.4 to 8.3 ms) and 64 MB written by the parent under the child (~28 ms of copy-on-write). test/fork_stall.py does the fork alone, on any system with a Python (the FreeBSD box): N MB written, then fork() timed and 64 MB written under the child, with Linux's THP as set and with 4 KiB pages. 1 GB: 13.3 to 19.6 ms here, 20.4 ms on FreeBSD 15.1; the copy-on-write after it ~1.3 us a page on both. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* the replica socket: the daemon listens on pardes-replica.sock, a stream `ssh ↵Gabriel Schneider2026-10-29
| | | | | | | | | | | | -L` can forward, refuses another build's hello in words and hands the connection to its session, which welcomes it and answers pings The daemon binds `pardes-replica.sock` beside its own (SOCK_STREAM, so ssh can forward it; the SEQPACKET socket that carries fds stays local) once it holds its lock. A replica's first message is a hello (detached/wire.zig, now version 14): the protocol version at offset 0, the client's semantics hash and the session's name. The daemon reads exactly that much and no further, so whatever the client sends next is the session's, keeps up to 16 hellos still on their way (ponytail: one that never comes holds its slot until it hangs up), and refuses in words another protocol, a session nobody has, or another build: every session now says its own hash in its registration (daemon protocol 2), so a mismatch never reaches it. An accepted connection goes to the session over its link (SCM_RIGHTS) and the daemon keeps no copy; the session welcomes it with its pid and name on a thread of its own and answers each ping with a pong, until bye or a hangup. ponytail: that stub is all a session does with a replica until phase 5 proper; it is what times a round trip through the handoff and `ssh -L`. The semantics hash (wire.semantics) is phase 1's undone item: the protocol, the release and commit, the platform, the build options, every dependency the manifest pins by its content hash (grammars, ghostty, the Unicode tables: build.zig's new `deps` option) and the walker codec's assumptions (pointer width, byte order). Not the OS: a Linux client of a FreeBSD session is the main case. ponytail: an unstamped build has no commit, so two dev trees of one release with the same options are not told apart. `pardes --version` prints it on a second line. The hello lost its grid (a replica's view takes a resize like any other input). test/daemon.py: a hello of this build welcomed by the session itself (its pid, one fd more in the session and none in the daemon, and back once it hangs up), 50 pings, another build's hello and an unknown session's refused, and `--ping <socket> <session> <pardes> [n]`, which times a hello and n pings on any replica socket (a forwarded one, say). Co-Authored-By: Claude Opus 5.5 <[email protected]>