summaryrefslogtreecommitdiff
path: root/test
Commit message (Collapse)AuthorAge
* A connection holds up to 128 reads, the next refused "too many reads ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | waiting: 128", and a mount beside 40 held event reads still answers Two limits met. pardes parked 32 reads per connection and refused the next with EAGAIN's C string, "Resource temporarily unavailable". 9ns kept 32 tags and 31 workers, each pinned by a held read. So 31 followers through one mount took every worker, and `cat layout` then queued for ever: the whole mount deadlocked. cloud9 (705be665, pushed to sr.ht and pinned here) now refuses in words past the cap. Its 9ns window is 256, so a mount always has workers past what pardes holds. pardes raises its cap to 128. fs.md documents the limit beside held reads. selfmount.py holds 40 event reads through the mount and reads layout beside them. It times out with the 9ns installed before this change and passes with the new one. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* A tty editor killed with SIGTERM or SIGHUP puts its terminal back before it diesGabriel Schneider4 days
| | | | | | | | | | | | | | A nested pardes killed in a pane died with the default action. That left the pane's terminal on the alternate screen, in raw mode, with mouse and paste reporting on, so the shell under it was unusable until a reset. The tty frontend now handles both signals: it writes the resets a clean exit writes (keyboard protocol, mouse modes, focus, bracketed paste, SGR, cursor, main screen) and restores the cooked termios, all async-signal-safe. Then it dies of the same signal, since SA_RESETHAND put the default back. fs.py kills a tty session with each signal and checks for the main-screen switch and cooked mode; it fails without the handler. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* pardes FILE in a pane whose session answers never falls back to a nested ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | editor: the refusal is printed, exit 1, and no stray pane is left When the environment named a live session that answered, several failures broke out of forwarding and started a whole editor inside the pane that asked: a refused name, a pane the session has not, a failed look write. Its screen was drawn over the shell. Now, once the session answers (Client.probe), every refusal is printed as `pardes: <file>: <why>`, in the session's own words (the Rerror, now kept by the client), and the launch exits 1, as acme's B does. A new name the session refuses deletes the pane made for it, so no empty +New is left. A missing environment or a session that does not answer still starts a separate editor. fs.py checks a refused name and a stale PARDES_PANE. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* The 9P monkey does not count a file watch's notice as a failed write's msgGabriel Schneider4 days
| | | | | | | | | | | | Seed 121 flagged a refused write to pane 2's dirty file because a `msg 3 <file> deleted on disk (Save writes it again)` arrived in the same window. That notice was from an earlier host rm of pane 3's file, said whenever the watch noticed it. It did not reproduce on replay, since it depends on when the watch fires. The one-failure rule now leaves out the watch's deleted-on-disk and changed-on-disk notices. Any other msg with a failed write still fails the rule. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Release Pardes 0.7Gabriel Schneider4 days
|\ | | | | | | Co-Authored-By: Claude Opus 5.5 <[email protected]>
| * lapis_plain: lapis on flat grounds, with its framed and shadowed tags and ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | its file-name shadow No dot grid under the text and no stripes behind the focused tag; every colour, the gold frames and vermilion shadows of the plaques, the rules, the scroll column's checker and the title shadow stay lapis's. It follows lapis in the ring and the Themes picker. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | pardes NEWFILE in a pane forwards even when its directory is missing, and ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Save makes the directories fs.md said FILE must already exist, while `pardes new.txt` in fact forwarded, making a pane named for it that Save creates. With a missing directory (`pardes notes/new.txt`) the launch gave up forwarding and started a nested editor inside the pane. Now the name is made absolute as written and forwarded the same way. A name written to a pane's name file that goes into a directory not there has its Save make the directories, as Config's pane does. A Save <path> into a missing directory still fails ENOENT. The Forwarding section now says what a new FILE does. fs.py launches one into a missing directory and saves it. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | The Tutor pane is named +Tutor in the session's directory, not /TutorGabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | | | Every other generated pane is named `<dir>/+<Name>`: +New, +Errors, the outputs. Tutor alone was a bare `/Tutor`, a real path at the root, so a Save would try to write there, and /index listed it as if it were a file. A scan of the other generated panes found no other bare absolute name. The three snapshot goldens that show the Tutor tag were re-recorded by name (tutor, builtins, leader). Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | DumpConfig prints each setting as the ctl word that sets it, reads ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | unsupported for one the frontend lacks, and includes LocationsConfig The report meant to be feedable back as configuration, but several lines were not. Shell and Font each took three to six lines of "requested", "effective" and "pending". The transition read "Panel transition PanelSlide", with no Panel word. WindowOpacity and TaglineSize carried "(pending)" and "(build-time only)". A GUI-only toggle such as Ligatures was left out on the tty, and LocationsConfig was missing altogether. Now every line up to a blank one is a word and its value: - Shell <path>; - Font <name>:<size>; - one on/off line for each Panel word; - `unsupported` for a setting this frontend does not have; - LocationsConfig as it would be typed. After the blank line comes what is in effect but set by no word, phrased so that none of it starts with a ctl word. The test checks that each line before the blank one starts with a setting word. The GUI scripts that read the font lines (font_size.py, appearance_gallery.py) now read the new ones. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | An append to a body through 9P costs its own bytes, not five passes over the ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | whole body: 60 KB appends go from 35.7 to 2.1 ms each Round 25 measured bulk body writes at about 8 ms a write. There is no frame wait in it: a profile of open, write, clunk in a loop found the flush of each close walking the whole body five times. dotOf, setDot and showOffset turned the cursor between offsets and rows by counting every newline from the top; setContent found the line index's changed span by comparing old and new byte for byte, and hashed the new text to see whether it was back to the saved one. The rows now come from the file's line index by binary search (checked against the counting at every offset), a splice tells setContent the span it changed, and the text is hashed only when its length is the saved text's. Measured over 9P on a Debug build: 60 KB appends 35.7 to 2.1 ms, 8 KB 5.3 to 0.8 ms, 1 KB 1.3 to 0.9 ms. What is left is two copies of the text an edit (the splice's and the undo snapshot's). The perf gate gains body-appends, 64 appends of 8 KB on the 50k-line file, and its three baselines are recorded again. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | The GUI goldens take the workspace tag's new words, Dump Themes Config Debug ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | Exit pqqtzsrz changed the workspace tag's default text and recorded the tty snapshots again, but not test/gui-goldens.txt, so all 17 GUI frames differed on main. Built with the old words, every frame is byte-identical to the recorded ones; recorded again with the new words, from the standard Debug gui build. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | The 9P monkey lets a failed write's op log a msg when another of its Twrites ↵Gabriel Schneider4 days
|/ | | | | | | | succeeded A ctl line cut across Twrites runs a piece a Twrite: in seed 109, "Save Pl" ran Save and said so, and the next piece, "acem", was refused. one_failure_rule took the msg as the failed write's. It now flags a msg only when no Twrite of the op succeeded. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Release Pardes 0.6Gabriel Schneider4 days
|\ | | | | | | | | | | Merge: audit top-five cuts + tag sweeps + Config/DumpConfig (on the audit bug fixes C9, C2). Co-Authored-By: Claude Opus 5.5 <[email protected]>
| * Config opens the startup file, DumpConfig the settings report, and the ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | workspace tag reads Dump Themes Config Debug Exit Config (SPC f c) opens init, found as the settings report finds it (opts.startup_config_path: $XDG_CONFIG_HOME/pardes/init or ~/.config/pardes/init, and the macOS and Windows paths config.md gives). A pane that already holds it is gone to instead; a file on disk is looked at; with none, an empty pane is named for it, and Save makes its directory and any above it first (File.make_dirs, fs.makeDirs) before writing. Without a configuration directory it says so and opens nothing. The report Config used to open, the startup path and every live setting, is DumpConfig, in a +DumpConfig pane. No alias keeps the old meaning. The workspace tag's default words put Themes where NextColor was, NextColor staying a builtin, and Config beside it. A tag the user edited is kept as it was: a dump stores only an edited one (topbar_custom), and a restore puts back only that. theme.snap types NextColor onto the workspace tag to click it, and tagnav's walk back to Dump takes five long-word steps. The rest of the re-recorded goldens differ in the workspace words and their widths only. Co-Authored-By: Claude Opus 5.5 <[email protected]>
| * A middle or right sweep over a column's or the workspace's tag runs or looks ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | for what it swept, as a pane tag's does Both headers acted on the press: a middle or right press ran, or looked for, the word under it (or the header's selection when it was on that) at once, so a drag had nothing left to do. A sweep of `echo SWEPT` in a column tag ran a bare `echo`. They have done so since the headers were drawn apart from the panes; it is no regression, and the -l/Restore allocator fix (ompukopz) does not touch it. Now a press on a header holds a HeaderSweep (the column by its serial, or the workspace), which the drag extends, held to the header's row and text as a pane tag's sweep is. It is painted in its button's tint meanwhile, and the release runs what it swept in that header's context; a release that has not moved runs the word or selection a click always did. The other of middle and right cancels it (acme 2-3/3-2); the select button is inert, as a header's command already takes the held selection as its argument. Running on the release, a builtin from a header is announced as one from a pane tag is: the press used to set the announcement and the release clear it at once. find, restore, theme, topbarexec, ttyhelp, tutor and windowops re-recorded for that badge alone. headersweep.snap sweeps the column and workspace tags with the middle button and a word part with the right one, and cancels a sweep; headersweep-restored.snap sweeps both after Dump and Restore. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | The PDF benches share their clock, fatal exit, hash, count parser and effect ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | drain, and lspprobe is lspbench's probe subcommand B4: test/bench_util.zig holds what pdf_bench, pdf_pointer_bench, pdf_scroll_bench and pdf_sections_bench each copied (nowNs x4, fatal x3 named per bench, mix x3, parsePositive/parseCount x3, drainEffects x2). They stay four entry points: pdf_bench links its own mupdf module (with its C source and c_heap) where the others link the core's, which one executable cannot import under one name, and their reports and identities are each bench's own. pdf_sections_bench had rotted since the pane's cursor and selections moved to its body and the render revision became next_raster_revision; it builds and runs again. C6: tools/lspprobe.zig is lspbench's `probe` subcommand (zig build lspbench -- probe <kind> <file> <line>:<col>), and the lspprobe build step goes; docs/lsp.md names the new command. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | Dumps and init files from before 0.0.4 load no more of their old tag ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | formats, and three orphan test scripts go A2: a restored pane's tag tail comes from the dump's tag_tail, which every dump since 0.0.4 writes; the 27 earlier default tails that were recognised to upgrade, restoreTail, tailClass, savedPrefix and restoreTailAt, pdf_view's and image's legacySavedPrefix, a terminal's TTY tag prefix and its raw output tail replay, and the ThemeSel, FontSel and ColumnTags words (init lines and dumped output names) go, with their tests; the CHANGELOG says so. A4: test/fs_soak.py, referenced nowhere (monkey9p.py does its work). A6: test/column_grip.py and test/look_placement.py, referenced nowhere; the six review scripts docs/ui-review.md names stay. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | Zig files go to zls as a child process through the protocol client, as every ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | other language does: the in-process ZLS backend goes lsp_zls.zig (1,522 lines), the zls dependency, build.zig's zls wiring and the import graph it baked into the options only for it, and main.zig's reference go. lsp_client's table gains a zls row (PARDES_LSP_ZIG overrides it); Hover, Rename, Diagnostics, WsSymbols, Lspinfo and gd answer through it (checked with lspprobe and the snapshots against zls 0.16.1-dev). What changes for a user: zls must be on PATH; a completion's rows are zls's candidates at the cursor, where the linked analyser gave each one's declaration; document symbols carry no signature. The Zig snapshots (lsp, lspcomplete, lspcompletemoved, lspdebug, lsprelpath) run the real zls now: lspdebug reads the client's explain and report, lsprelpath spells gr's rows where it spelled a completion's, and their goldens are recorded again (stable over three runs). Lspinfo still opens with its backend line. The ReleaseFast tty binary is 105.35 MB to 96.28 MB, the Debug one 296.5 MB to 245.9 MB. docs/lsp.md, web.md, design.typ and the README say so. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | The GUI embeds its shaders' GLSL from shaders/ always: the 14 prebuilt ↵Gabriel Schneider4 days
|/ | | | | | | | copies, identical to their sources, go, and so does the label saying which was used shaders/prebuilt/*.glsl were byte-identical to shaders/X.glsl or to prefix.glsl + post/X.glsl (checked with cmp, all 14), kept only so a -Dprebuilt-shaders build could name them; EffectCode and Config then labelled the GUI's shader source as live or prebuilt. The GUI embeds shaderSource (a concatenation, no glslc) either way now; the prebuilt SPIR-V stays, and zig build shaders refreshes only it. Config's report and EffectCode lose their GUI shader source line. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Release Pardes 0.5Gabriel Schneider4 days
| | | | Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Terminal.zig is terminal.zig: a file of functions and no fields takes a ↵Gabriel Schneider4 days
| | | | | | | | namespace's lowercase name The last of the deferred renames, now that the tty and theme agents have landed: panes.terminal at its importers, the alias lines in Text.zig and File.zig and panes.zig's own uses following. dump.zig's Terminal struct, a dump record, is not this. The served sources list and docs/design.typ name the new file. test/perf.zig's references change, so the three perf baselines take its new harness id with their numbers as recorded. No behaviour changes. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Merge: renames/folds + tagline accents for every themeGabriel Schneider4 days
|\
| * acme tints a tag's file name, +Name and Tty word quietly, in its own ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | DMedblue #000099 acme tints nothing in a tag, but the tint is a pardes indicator the user wants in every theme. #000099 is acme's own (draw.h DMedblue, its dirty button): a navy that reads as ink until you look for it. Co-Authored-By: Claude Opus 5.5 <[email protected]>
| * A theme that names no tint for its tags' names takes one of its own hues, so ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | every theme draws a file's name, a +Name and the Tty word off the tag's ink The faithful ports (neobones_dark, dracula, tokyonight, the *bones, ...), every helix import and any theme file set no tag_name_fg, and ChromeTheme fell back to the plain ink: their tags lost the accent the pardes themes have. The user's neobones_dark looked tinted before only because the Sep 24 build had no such theme and fell back to orchard. The tint is now derived when unset: the theme's keyword colour, else the first of its ANSI accents, syntax, grip, selection or diagnostic colours with a hue, placed at the lightness nearest the ink where the hue holds, at least 4.5:1 on its ground (half a point under the ink's own where the ink has less to spare) and 0.07 off the ink in OKLab. Each ground, plain and focused, gets its own. A theme that names its tint keeps it exactly. Tests: every theme's chrome keeps both tints off the ink and legible; every theme draws a file's basename, +Errors and Tty+sh in a tint, focused or not. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | fs.py waits for a new file's pane to be named, not only madeGabriel Schneider4 days
| | | | | | | | | | | | pardes --wait NEWFILE reads pane/new and then writes the pane's name, two requests; the test checked /index for the name as soon as the pane showed, and a gate run caught it between them. It waits for the name now. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | Mini.zig is mini.zig: a file of functions and no fields takes a namespace's ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | lowercase name The naming the split agreed on, deferred until now; panes.mini is its name at its importers. File.zig keeps its alias line (const Mini = panes.mini) so the file another agent is working in changes by that line alone. The served sources list and docs/design.typ name the new file. No behaviour changes; test/perf.zig's two references change, and since the perf harness is keyed by its own text, its three baselines are recorded again (under other agents' builds, so a little slower). Co-Authored-By: Claude Opus 5.5 <[email protected]>
| |
| \
| \
| \
| \
| \
*-----. \ Merge: loose end + msize 64K + docs slimming + docs-audit fixes + monkey-gui ↵Gabriel Schneider4 days
|\ \ \ \ \ | | | | | | | | | | | | | | | | | | + tty ssh acked frames
| | | | * | Over a slow link the tty keeps one frame in flight: it asks the terminal to ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | | | | | answer after each, and draws the latest state when it does
| | | * | | The 9P monkey allows a look's misses to each Twrite that succeeded, not to ↵Gabriel Schneider4 days
| | | | |/ | | | |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | none when one failed A look line written a byte a Twrite is a look a byte, each its own write logging its own miss; when its last byte was a control character that one write failed, and the oracle dropped every miss allowance on the path, so seed 51 failed with 21 errs for 1 refused write. Allowances are now counted over the Twrites that succeeded (and an hwrite's only once it has). Co-Authored-By: Claude Opus 5.5 <[email protected]>
| | * / | The GUI monkey lives in test/ and runs as zig build monkey-guiGabriel Schneider4 days
| | |/ / | | | | | | | | | | | | | | | | | | | | It was a scratch tool (.scratch/render/tools/gui_monkey.py) that the gates rely on and that imported fs.py from a workspace since removed. It is test/gui_monkey.py now, importing from its own directory, and zig build monkey-gui -Dplatform=gui runs it on the built window, the same hidden test windows; it exits 1 when a combination fails, and its outdir is optional. Co-Authored-By: Claude Opus 5.5 <[email protected]>
| * / / The 9P server offers 64 KiB frames, not 8 KiBGabriel Schneider4 days
| |/ / | | | | | | | | | | | | | | | 8192 was kept for round 24 because 64 KiB seemed to hang fs-test's restore detached space; that was the batch-wake overflow fixed in kmqtwqsm, which any msize could hit, and fs.py passes at 64 KiB with it. A client asking for more now gets it, so a write up to 64 KiB less 24 is one Twrite, and round 24's no-newline cutoff (write_room, the negotiated msize less 24) follows: fs.py checks a 20000-byte ctl line from a 64 KiB client is taken whole. Each of the 16 connection slots holds its buffers at the larger size. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* / / pardes NEWFILE in a pane opens a pane named it in the session, as acme's B ↵Gabriel Schneider4 days
|/ / | | | | | | | | | | | | | | does, and a start with no terminal says so A name that did not exist yet failed resolveOs, so the launch fell through to starting an editor of its own, which on a pty with no controlling terminal (a detached session's) ended in a NoDevice error trace. A new name's directory is now resolved and the name kept: the launch reads pane/new and writes that path to the pane's name, so Save creates the file, and --wait waits on that pane (the git commit case). A standalone start whose /dev/tty will not open prints why and exits 1. fs.py covers both; the known-wrong line in docs/divergences.md goes. Co-Authored-By: Claude Opus 5.5 <[email protected]>
| |
| \
*-. \ Merge: 9P round 24 + acme compat + fuzzer fixes + --wait + tty perf fixes + ↵Gabriel Schneider4 days
|\ \ \ | | |/ | |/| | | | 9P monkey
| | * The 9P monkey's one-failure rule is a write's: a refused open, create or ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | | | | | remove, or a write to pane/new, must log no err The coordinator's reading of fs.md, now documented (slnyzzps): only a write or truncation logs its reason; a refused open (pane/new's and an OTRUNC one included), create or remove is its Rerror alone, and so is a write to pane/new. The oracle allowed an err or none for those, and required one for a refused pane/new open. Co-Authored-By: Claude Opus 5.5 <[email protected]>
| | * The 9P monkey counts the lines a write runs as the tree does: a short ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | | | | | write's unended tail is a line of its own Since round 24 (uwzuqzxp), a write to a lines file shorter than a Twrite holds is whole, as acme takes each write, so a look line written a byte a Twrite is as many looks as bytes, each missing with its own err. one_failure_rule still split the data on newlines whole, allowed one err, and failed the fixed smoke seed (20260929 step 607, size 20 6 written in bytes to look: 7 errs). served_lines models the tree's write (tree.zig) and has a self-check run at start. Co-Authored-By: Claude Opus 5.5 <[email protected]>
| | * A 9P monkey drives random requests at a throwaway session and checks the ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | documented rules test/monkey9p.py starts pardes --detach with its own HOME, XDG dirs and runtime dir, speaks raw 9P (test/ninep.py) with a timeout on every reply and Tflush for held reads, and after each seeded operation checks that pardes is alive, the one-failure rule over a marker-delimited log window (with and without a follower), /index against /pane, and chunked against whole reads. Every run writes a replayable ops log; a failure is saved with its seed and step, shrunk by delta debugging over replays, and the run goes on in a fresh session. Teardown makes the fuzzer a subreaper and kills every process carrying the session's token. zig build monkey-9p -- --seed N --steps M | --replay F | --shrink F; a 150-step fixed-seed smoke run is part of fs-test. Co-Authored-By: Claude Opus 5.5 <[email protected]>
| * | Snapshots leave out a blank cell's ink, which shows nowhere and the tty now ↵Gabriel Schneider4 days
| |/ | | | | | | keeps from the frame before
* | pardes --wait (-w) in a pane returns only once the pane its file landed in ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | is deleted, so it can be $EDITOR A launch inside a pane forwards its file to the session's look and returned at once, so fish's Ctrl-O (edit_command_buffer), git commit and crontab -e read their still-unedited file back and went on. With --wait, as acme's E against B and read the way plan9port's E reads acme's log, it finds the pane /index names the file by (the one already showing it, if open), follows /log on one connection with 'follow new', reads /index once more to catch a Del that came first, and then blocks with no deadline until that pane's del record: exit 0, or 1 when the connection ends with the session. The 9P client's one-shot requests keep their 2 s deadline; its new follow takes it for the setup only. Outside pardes nothing changes. --help, docs/fs.md, the README and the 9P skill say to set EDITOR='pardes --wait', which GIT_EDITOR follows. Tested in fs.py (it returns within 50 ms of the Del, a second -w waits on the same pane, one whose session is killed exits 1) and end to end: fish 4.8 in a detached session, echo hi, Ctrl-O, the waiting launch idle in poll, the line edited over 9P, Save, Del, and fish ran echo edited. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | Mini with no argument is wrong #args, EINVAL; Find, Grep and Mini say their ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | names capitalised Mini needs a file but did not say so, so a bare Mini ran and failed EIO. It now requires its argument, refused wrong #args (EINVAL) as every such word is. Its failures, Find's and Grep's start with the builtin's own name (Mini:, Find:, Grep:), as the others' do. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* | An open's writes in a row to one place go in as one edit: a 10 MB body write ↵Gabriel Schneider4 days
|/ | | | | | | | is linear Each body or data write copied and hashed the whole buffer, so a write the mount cut in 8 KB pieces was quadratic: 10 MB took 50 s and a 1 MB insert into 10 MB 11 s, holding the editor's turn. An open's appends to body, or inserts going on at data's address, are now held and put in as one splice (one copy, one undo step, one line-starts pass) before any other request, the close, or the editor's step once the writes pause 20 ms. fs.py-driven: 2 MB 2.19 -> 0.16 s, 10 MB 50.17 -> 0.78 s, 1 MB data into 10 MB 11.49 -> 0.26 s. A body-2m case (2 MB in 256 KB writes on one open) joins the perf gate: 84023 -> 15785 us; baselines re-recorded. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Delcol on the last column leaves the window, empty but for its workspace ↵Gabriel Schneider4 days
| | | | | | | | tag, as acme's does The last column's Delcol ended the session. The window now stays, with only the workspace tag; Newcol, and New (which makes the column it goes in), start it again. A word run with no pane at all runs with a stand-in that is no pane, the root ctl takes session words, and pane/new makes its scratch in the session's directory. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Not-found failures say it once, in plain words: grep: pattern not found, ↵Gabriel Schneider4 days
| | | | | | | | callers not found The ENOENT that 9ns reads from a failure's words came from a tacked-on suffix, grep: no match, not found, definition: nothing found, not found, no shell "x", not found. They now say it once: grep: pattern not found, find: name not found, Callers: callers not found, definition: not found, shell "x" not found (...). Co-Authored-By: Claude Opus 5.5 <[email protected]>
* A ThemeFile that cannot be read or is no theme fails its write, and a bad ↵Gabriel Schneider4 days
| | | | | | | | path says why The theme is read when the host performs the request, after the write had been answered, so a missing file or a bad theme only showed a message; and a path not ending .zon said only invalid theme path. Load failures are now the waiting write's (late, as a Save's), and the path refusal says a theme file is a .zon. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* A buffer whose file was deleted on disk is dirty, and Del asks firstGabriel Schneider4 days
| | | | | | A file deleted under an open buffer left dirty 0 and let Del, Exit and the rest throw the only copy of its text away unasked. It now counts as dirty, as acme's does, and the refusals ask about it. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* A path with a part over 255 bytes is refused, never a panicGabriel Schneider4 days
| | | | | | std's statFile takes the kernel's ENAMETOOLONG for a bug (errnoBug), so a name, look, DumpDir or Save path with a part over 255 bytes panicked the editor (exec.kindOf via writeName, recentKeeps, dumpFailed). Every non-test statFile now goes through fs.statPath, which refuses such a name as NameTooLong first; fs.py drives long, looping, not-a-directory and not-ours paths through name, look, DumpDir, Dump and Save. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Typing tells g. where its edit ended, so no edit scans the whole file twice ↵Gabriel Schneider4 days
| | | | | | | | to find it setEditText found the last edit's end by comparing the old and new texts from both ends and counting rows to it, twice over the file on every keystroke: edit-char grew with the file (+135% at 300k lines). Typing, deleting, Enter and Tab now pass the end they know (setEditTextEnd); the comparison stays for the edits that do not say (paste, pipes, s). perf edit-char: 50k lines 16661 -> 15286 us, 300k lines 30507 -> 22291 us (render alone ~13.6 ms of each, the render agent's). Co-Authored-By: Claude Opus 5.5 <[email protected]>
* The theme's chrome is worked out once, not per grapheme; still post passes ↵Gabriel Schneider4 days
| | | | | | | | | | | | | | | | | | | | | | | | | | | | let the GUI rest; a perf gate somtrmsz's contrast floors made ChromeTheme.fromTheme run its focus-tint and separator searches (pow calls each), and Output's RowDecoration.styleAt called it for every grapheme of every highlighted row: the 50k-line file's render went from 0.11 ms to 13 ms. Pardes.bodyChrome now keeps the theme's chrome, worked out again only when the theme differs; recolorSyntax asks for it once a decorated row, and a plain row (nearly every row of a file) never asks. ReleaseFast, medium fixture, median us, before -> after (main): render 13150 -> 76 (111), key-down 14658 -> 79 (110), wheel 13589 -> 76 (109), open 14891 -> 2095 (2055), edit-char 16703 -> 3006 (1251; the rest of that gap is editing and tree-sitter, not this). Post.animating asked for frames whenever the window had focus and any pass was ready, so a still pass kept the GUI drawing at the display's rate (Bloom: 49% of a core idle, in a hidden test window). A pass now says whether it moves on its own: the CRT (its hum and dither) and a Shadertoy file whose source reads iTime, iFrame or iDate (shader_build.readsTime; the flag rides the wire's post message); Bloom, Vignette and Grain are still. Bloom idle: 49% -> 1.2%, as with no pass. zig build perf-gate: the 50k-line file's gestures, each's fastest sample within 3x of the recorded baseline's (test/perf-baseline-<platform>- <optimize>.json, recorded from this build), run with every unit-test. On the regressed code it fails at 180x for render.
* The workspace tag's Restore word names the last dump, and the dump a Restore ↵Gabriel Schneider4 days
| | | | | | | | came from Once the workspace tag was edited, its own text kept the Restore <path> of the dump at that time, stale after the next Dump; and a Restore dropped the path altogether (the new session had no last dump). A new dump now swaps the old path in the tag's own text, and a Restore points the word at the dump it read. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* A language server question that finds nothing, cannot be answered or has no ↵Gabriel Schneider4 days
| | | | | | | | room fails, and so does a write to an image's or PDF's body Callers or Callees with nothing found, Hover or Lspwhy with no room for its pane, and a question asked of no file's pane said nothing, the 9P write succeeding; a body write to an image or PDF pane went nowhere. A write that asks a language server now waits for its answer (turn.lsp_answers), and what fails there -- nothing found, no answer, a backend's @none why, a placement with no room (ENOSPC) -- fails the write and logs its err. No diagnostics is said, not failed. Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Recent goes on across a Restore, and is kept after itGabriel Schneider4 days
| | | | | | A Restore makes a new session, which never loaded the kept list, and saveRecent gives up until it has: Recent was empty after a Restore and nothing opened afterwards was kept. The new session now takes the old one's list (a boot from a dump reads the kept file). Co-Authored-By: Claude Opus 5.5 <[email protected]>
generated by cgit v1.3 (git 2.53.0) at 2026-10-04 21:13:23 +0000