| Commit message (Collapse) | Author | Age |
| |
|
|
|
|
|
|
|
|
|
|
| |
-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]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
replicas that must reach the same digests; the core's leaks of render, clock and host closed
The recorder (src/tape.zig) is null unless PARDES_RECORD=<prefix> names a tape (<prefix><pid>.tape); the local path pays a null check a call. It records a tree at the host boundary, both ways: each call the host makes into the core (`Input`: update, serveFs, pump, perform, the shell's acknowledgements and failures, a Restore, ...) with its arguments, and each call the core makes out -- a Host function (the vtable wrapped generically while recording) or a syscall of its own (`tape.site` for every old `turn.yield()`, `tape.value` for a clock or environment read) -- with its result; `done`/`ret` close each, a 9P request served while a step is out lands inside that step's syscall, and fields a host writes between calls (native_images, present_skipped, ...) are found by a diff at every crossing. After each call the host made at the top, the session's digest (src/walk.zig), one per field of the core, its stream and each pane, file and places.
`PARDES_REPLAY=<tape> pardes` (main.zig) replays it with the same build into three replicas: two at different addresses (the first's memory held while the second runs) and one that never draws. A host call answers what it answered, a site returns what it returned, nothing is performed and no syscall is made; `turn.yield` panics in a replica, so a syscall no site covers is a stack trace; a call the recording did not make, or a digest that differs, is a divergence that names the fields. The headless replica steps over the presents it does not make, applying what the host told the core inside them.
What the replicas found, fixed:
- Render wrote core state. The view's geometry (wrap map, context rows, body layer rows) is File.geometry, a lazy cache keyed by a hash of what it is made of (revision, scroll, size, tags, tables read, context declarations, the session's view: wrap, layered context, row metrics, rect); the render makes it afresh, a step reading it makes it on demand (wrapAt, wrapRow, ensureCursorVisible). Context declarations, notice bands and the header's reveal are the step's (sync), the tables' reading the tick's (a first look of 4096 lines, a count, wherever it is asked; the rest a frame's clock budget a tick), and a view's paint-ahead is asked of the host as the frame is drawn rather than queued as an effect.
- The clock: advance's wakes are the recording session's (a picture decoding or a fade adds no tick to a replica), as are the budgets of Places, Names and Tables (`tape.over`), a message's time of day (the stamp; msg_said_ms and the unattended expiry use the session clock), File.mtime, and the pump reads its clock whether or not it waits.
- The host: the session directory (getcwd, asked once at init), $SHELL and its existence, $HOME for `~`, `Shell` naming one that is not there, a missing interpreter, a directory refused above, a deleted file, a pointer seen through a sliding pane (Pardes.pointerCell).
- Worker handles (futures, the Io), the 9P server's state, the store file every window shares (a replica keeps its own in memory) and Recent read from it are authority-only; frames, animations and the render's caches local. Query rows' paints count apart from a view's.
Terminals replay from their taped pty bytes and are digested as their active screen, cursor and scroll; kitty images are left out until phase 5's screen-state terminals. tagnav.golden: the header's caret reveal now runs every step, as a pane's does, not once per frame drawn, so after `g l B B` typed in one burst the bar stays scrolled as a human typing it sees it.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
daemon included
Cross-compiled with -Dtarget=x86_64-freebsd.15.0 (15 ships libutil.so.10;
Zig's default 14.0 links .so.9) and run on the FreeBSD 15.1 box:
- 9p_io serves there: fstatat, not Linux's statx. madvise is Linux's.
- WINCH reaches the sigwait thread through a handler that does nothing:
FreeBSD drops a signal ignored by default as it is sent.
- A dead terminal answers a write with ENXIO, which std does not expect: a
Debug build traced it to stderr, that same terminal, and recursed until
the stack ran out (a session whose ssh dropped hung). No tracing there.
- waitid's FreeBSD numbers and 64-bit id: a command pane's exit is seen.
- No /proc: a shell's directory and program by sysctl, ttyTaken by the
foreground group, the foreground job's name by its program.
- File watching by FreeBSD 15's inotify.
- ghostty's kitty loader: off Linux a stat comes before a t=f open (a FIFO
hung it); FreeBSD's shm_open refuses O_NONBLOCK.
- vaxis's test terminal has no read off Linux: no input thread for it.
- test-install: the test-build programs installed into <prefix>/test, to
run where the build cannot.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Detach and --attach in place for terminals and windows, pty masters kept, Adopt
pardes in a terminal reaches (or starts) the daemon, opens /dev/tty and forks: the child is the session, in a session of its own (setsid), drawing on that terminal (tty.Term.fd, a raw vaxis Tty made from the fd); the parent stays as its stub, the job the shell waits on. Not the daemon spawning it from a passed fd, env and argv: the user's terminals each run in a 9ns user and mount namespace, which a session spawned by a per-user daemon would not share, and a fork costs one small process's page tables where the spawn cost an IPC and an exec. GUI windows and --detach sessions just register.
tty: Detach (and a terminal's end: EOF, EIO, a hangup the stub passes on, taken on the signal thread) puts the terminal back, drops it and the stub's stdio, tells the stub (which ends, exit 0), and leaves the session running with nothing drawn (Host.watched), its messages on the clock and `size` its own. pardes --attach[=name] hands its terminal through the daemon; the session enters it as the first one, sized, queried and drawn whole, and that process is its stub until it leaves. A hangup never ends a kept session; TERM and INT do, telling the daemon so the shells go with it (one async-signal-safe send).
gui: Detach closes the window (its swapchain or renderer) and keeps the device, atlas and pipelines; the loop sleeps on SDL's queue; pardes-gui --attach[=name] opens a window again. Closing a window still quits.
--detach is a session started detached: the tty host with no terminal yet, the GUI with a hidden window closed at once. forkShell sends the daemon a dup of every master and the hosts tell it what they close, so a session killed outright leaves its shells running; the new Adopt word takes them in as terminal panes (the Tty word's pane, its spawn taking the pty, a 1x1 size then the pane's to make it redraw).
The Attach word, which switched a session's own terminal to a detached session through the frames, goes, with its effect and request in the core. PARDES_DAEMON=off (the e2e harness's) runs a session as before. test/daemon.py: 50 detach/attach cycles at a level fd count with no stale socket, kill -9 and Adopt, and nothing left; --gui the same for a window. test/fs.py: a forked session's socket is found by its own pid, and a hangup leaves a kept session running.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
| |
in a new window (Handover.zig)
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
records read on use, the panes a session closes kept as one-pane dumps; Panes lists them (closed), Enter or a look at @pN brings one back beside the pane that took its rows
An append-only log under $XDG_STATE_HOME/pardes/store, framed records with
a CRC, flock on store.lock (never blocking a step), a stat on use, garbage
passed over and compacted away by the next writer, compaction by rename,
100 MiB cap. dump.dumpPane/restorePane lifted out of dumpState/initDump;
layout.giveBack; Options.recent_path is store_path. Host calls yield one at
a time; /recent refreshes only when no step is out. Tests: two windows and a
busy lock, the cap, garbage/torn/partial records, a closed pane's reopen
with the layout invariant; fs.py: two live sessions over 9P.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
the shells allocate from one too
Each arena (pardes, frame, lsp, tree_sitter, image, pdf and the new host)
maps its region once and tracks 4 KiB pages in a bitmap; a full region
spills to malloc and counts it. Freed runs of 64 KiB to 1 MiB go back at
once, the rest at memory.trim. In a release build init.gpa is
memory.host over libc's malloc. Measured against the old arenas: typing
in a 7.4 MB file 1.95 ms a key (was 2.0), a V8 Callstack page 142 MiB
(was 134) and 42 MiB after close (was 69).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
| |
ctl.failureErrno, dead switches and dumpHex out
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
bounds and resident-size fixes
Callstack walks while the server answers, counts on the message row past a
second, puts test callers last at each level, and ends a tree past one
query's memory (memory.limits.arena.lsp_query, a 256 MiB region taken and
given back per query) in a row whose Callstack shows the next page. Grep
reads whole files. New hard limits: undo text per file, image decode
pixels, the LSP snapshot of other open files, one paste. Safe builds no
longer fill worst-case buffers with 0xAA (memory.unfilled), big regions are
kept off transparent huge pages, and malloc_trim runs after big results.
Also: Look's git show pages into a +Diff, and dropped marks in File.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
ghostty, libvaxis and uucode move to ghostty's 0.17 pins, cloud9 to its 0.17
main, SDL to castholm's release-3.4.x (3.4.18), zstbi and tree-sitter to
forks with 0.17 build files (see build.zig.zon). A bare zig build copies the
binaries into ~/.local/bin (-Dhome-install=false leaves it out): 0.17 keeps
the prefix from build scripts. A terminal's edited copy goes when the program
clears the screen. tinywin.snap is gone; perf baselines are re-recorded.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
| |
edits, context rows that hold still under a scroll, and the pardes skill
Def and Refs (gd, gr, Ctrl-click; chorded 2-1 at the selection) for a file
a language server speaks for; hunks.zig stacks a unified diff as a diff
viewer does, its changed words on a tinted ground as delta's do; the
tree-sitter context rows stay put while the body glides under a scroll;
.agents/skills gains the pardes skill beside pardes-9p.
|
| |
|
|
|
|
| |
session's socket holds (a plain session is named by its pid) says so plainly, and any other failure of it in words, never a Zig error trace
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
| |
exit 1, instead of drawing a screen of its own in the asker's terminal
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
| |
written after its directory with ghostty-vt, SGR becomes spans drawn over the plain text, every other escape is dropped, and PagerColor off pages it plain
|
| |
|
|
|
|
|
|
|
|
|
|
| |
or a script's EDITOR) asks that session and waits there, not a screen of its own
Forwarding asked for $PARDES_PID and $PARDES_PANE, which only a pane's
shell has, so a client that talks to a session from outside its panes
got a full editor drawn in its terminal. With --wait and $PARDES_9P
alone, the file is looked at through the session's root look and the
wait is for that pane's del.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
since the Dump is said changed, its first Save asking before overwriting it
A Restore put back the dump's copy of every file pane and armed the
watch with the disk as it now was, so a file changed on disk after the
Dump came back as its old text, marked clean, and the next Save wrote
the old text over the new without a word.
A clean file pane's text is now not in the dump: Restore reads its file
(an older dump's copy is read past too, and used only when the file is
gone). An unsaved pane keeps its text once, base64, the plain copy beside
it dropped, with the hash of the file it was read against
(`disk_hash`); a Restore that finds another there says the file changed
on disk, and Save asks once before overwriting it, as for a change seen
while it was open. file_watch.zig's own tests were never collected (no
test root imported it): main.zig's test block does now, and the restored
watch test says what a clean restore now reads.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
| |
two\nlines.txt: a file name is one line...`, not `a file name: a file name ...`
The refusal's subject was the words "a file name", and its reason
began with them again. It now names the name given, its newline shown
`\n`, as /log and /index show one.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
not be written, permission denied; `pardes -` from such a directory pages into the session's
`/tmp/../etc` named a +Pager `/tmp/../etc/+Pager`, a second one beside
`/etc/+Pager`, and `~/x` was refused as relative. The directory is now
home-expanded and resolved before it is checked, and one that may not
be written is refused as a file's name there is (fs.deniedAbove).
`pardes -` run in such a directory (`git log` under /usr/src) asks for
the session's +Pager instead, so its text is still paged.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
permission denied, and a Save there says so, not no such directory
`pardes noperm/a/b/c` and `pardes /proc/1/root/x` took the missing
directory for one Save would make, opened a pane and exited 0; its Save
then said "no such directory". fs.deniedAbove finds the nearest
directory there and asks whether it may be searched and written:
forwarding refuses such a name, exit 1, "permission denied", and a
failed Save says "permission denied", which the writer's 9P error
carries as EPERM (9ns: EACCES).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
| |
It said `no such option: --`. What follows `--` is a name, one at most,
as for any other positional.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
one found again by name in /index
The forwarded look was a write and nothing more, and --wait then
looked its file up in /index. /index shows names escaped (`\\` for a
backslash, `\xNN`), so a file named with one matched no pane and
--wait exited 0 at once, as if the pane were already closed. The look
is now asked on one open (Client.ask) and its answer is the pane
waited on; /index by name stays only for a look that answers none.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
another client's exec meanwhile can never take the text into its pane
pardes - wrote `pager <dir>` to /ctl, then read /exec on a fresh open,
which answers the session's last answer: a client that exec'd in
between had its pane overwritten by the paged text. /pager takes the
directory and answers the +Pager's serial on the open that wrote it,
as /pane/new answers its own open; the ctl verb is gone. Client.ask
writes and reads on one open. The README's look recipe reads its
answer on its own open too.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
8 MiB arrives whole, not as its last chunk
Each 8 MiB chunk went in by its own Client.write, which opens a body
with OTRUNC, so every chunk replaced the one before and a long page
kept only its tail, with exit 0. The text now goes in through one open,
truncated once and written at rising offsets; a long write is timed by
its progress, each 9P write answered within 30 s rather than the whole
transfer within 2. fs.py pages 11 MiB of numbered lines and compares
the body byte for byte, and pages past the 256 MiB cap and checks the
first 256 MiB and the cut note.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
log record; streams the text in bounded writes; never loses it; handles \r, NUL and BEL as a terminal does
Each paged command made another +Pager until placement failed, and
then git's text was lost with exit 0; a 130 MB page timed out half
written and dirty; a directory whose name held a newline left a stray
+New; and the log said new +New then rename. The root ctl's `pager
<dir>` now makes the directory's +Pager (named before it is announced,
placed without the keyboard) or empties the one there; the text goes
into it a line-ended 8 MiB chunk a write, one undo step, up to the file
limit and its note, and is left clean. A directory a ctl line cannot
name pages into the session's. Any failure prints the text to stderr
with why and exits 1. A carriage return keeps a progress line's last
state and CRLF is a newline; NUL, BEL, SO and SI go; empty stdin makes
no pane.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
GIT_PAGER, Pager pardes|off to opt out; command panes page through cat
A paging command in a terminal, or run through its pty/run, sat in
less, and a pty/run never answered. `pardes -` now reads stdin to its
end, strips terminal escapes (colour, OSC, overstrike), and shows it:
inside a session as a clean <cwd>/+Pager pane made through pane/new and
back at once, outside one as a new editor whose first pane it is, as
`vim -` does. A terminal's shell gets PAGER and GIT_PAGER set to this
pardes's own path and `-` where the user's environment sets neither;
`Pager off`, a setting in /ctl, init and DumpConfig, leaves terminals
started after it to the environment. A command pane's one-shot shell
keeps cat, its output being a pane already.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
| |
write is refused with words, not a pane that only fails at Save
A new name forwarded into a directory that is there but locked opened
an empty pane named for it, and the failure came only at its Save,
long after the launch. The directory is checked first: one that cannot
be read, written and entered is refused, exit 1, no pane made.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
| |
session has no such pane", exit 1 and no pane made; one it has is looked at
@p999:1 was taken as a new file's name, so a pane named <cwd>/@p999 was
made and the launch succeeded; an @pN the session did have was the same
stray pane. A pane address is now checked against the session: one it
has is a look at it, one it has not is refused with the reason and exit
1, as a refused name is.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
| |
"a file name is one line", never NotOneLine
The forwarding refusal printed the error's name for anything the
session itself did not answer: NotOneLine, NotAFileName,
NoWorkingDirectory. Those are now sentences, a dial failure is said as
the message row says one, and any other error is its name's words.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |\
| |
| |
| | |
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
pane closed before it looks is done, not a failure
fs.py's `pardes --wait 'wait new.txt'` failed two runs in three on the
0.0.8 merge. The stress case repeats the test's steps: 4 of 50 failed
under load, each with "pardes: --wait: no pane shows that file". A new
name's launch makes its pane, names it, then found it again by path in
/index. The script had already seen the name, saved and removed the pane
by then, so the lookup missed and --wait exited 1.
Now the launch hands --wait the serial pane/new gave it. After a look,
finding no pane on the file means that pane is already closed, which is
the end --wait waits for: exit 0. The follow with its /index recheck
covers a close in between, as before. Under the same load the stress
case went 0 of 100.
The same load showed a test race too, at fs.py's second Restore. The
exec write is answered before the hang-up, but the client's clunk after
it can meet the connection already cut. The test now requires the
answer and tolerates the reset.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
that takes one, through one helper
Only DumpDir (dump.zig) and ShaderBuild expanded `~/`, each its own way.
So a look at `~/notes.txt`, `name ~/x`, `Save ~/x`, ThemeFile, Restore and
a quoted `pardes '~/x'` all took `~` as a directory of that name under the
pane's. There is now one rule, filesystem.expandHome. A leading `~`,
alone or before `/` or `:`, is $HOME, else the passwd entry's home;
`~user` is that user's (getpwnam). As in a shell, it applies even beside a
file named `~`, and `./~` names that file. Every entry point runs its
typed path through it:
- look, B3 and 9P, before any parsing, so `~/x:12`, `~/x:12:3` and
`~/x:/re/` all work;
- name, and a tag's rename;
- Save <path>;
- ThemeFile, DumpDir, Restore and ShaderBuild's shader files;
- the forwarding launch.
Recent rows are not shown with `~` and need nothing. How names are shown
is unchanged.
Tests: expandHome's cases, and HOME unset (passwd), with HOME passed in
rather than set, since setenv moves the environ a spawned child reads.
fs.py covers Save into missing directories under `~`, `name ~/x`, a look
at `~/x:3`, and a forwarded `pardes '~/x'`.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
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]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
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]>
|
| |
|
|
|
|
|
|
| |
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]>
|
| |
|
|
|
|
|
|
| |
map, vaxis_input.zig, for the tty, the gui and the board
The vaxis-to-Event key mapping was copied in tty.zig, gui.zig and esp32p4.zig, and the copies had drifted: the gui's (its stdin feed, which test windows and a gui started in a terminal read) had 9 arms against the tty's 26, without the function keys, Insert or the keypad, so they reached a terminal pane as private-use codepoints. vaxis_input.zig holds effCp, mapKey and keyEvent once; the gui posts through postKey, and a test runs F5, Insert and the keypad's Up through it into a terminal pane.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
| |
Locations setting and cache beside what uses them
Config (75 lines) was re-exported from locations.zig already, and Cache (213) served only its format; locations.zig is 683 lines with them. The cache's own Result, which would be ambiguous beside locations' Result in one file, is Cache.Hit, what its comment already called it. pardes.zig, dump.zig and main.zig import locations.zig. No behaviour changes.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
| |
the post chain's builds) takes a type's name
The naming the split agreed on: a file with fields is a type, TitleCase, as Messages.zig and Layer.zig are. Its importers bind it as ShaderBuild; comments and docs/render-pipeline.md name the new file. No behaviour changes.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
| |
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]>
|
| |
|
|
|
|
|
|
| |
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]>
|
| |\
| |
| |
| | |
cursor + faithful themes
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
chain
shader_build.zig takes the Shadertoy compile out of the GUI's Post: the
process that holds the core compiles each chain file behind the prefix on a
thread, watches its directory (file_watch's new shader slots), and on a
change reads and hashes each file, compiling only one whose bytes moved. A
failed compile keeps the last good SPIR-V and is said once; the same bytes
are never compiled or reported twice. A file is read with std, not fs.zig's
readFile, whose turn hand-off a worker thread does not hold.
A detached session compiles for its frontends and sends them the chain
(wire post: each pass's scene and level or the file's SPIR-V, and
ShaderAnimation) on attach and on every change, so an attached GUI runs the
same passes, levels and animation as a local one while still reading no disk
and running no program. The attached GUI describes its frames to the chain
from the session's chrome and redraws an animating chain on its own.
|
| |/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
closed file reopens it
A file closed by accident, its path forgotten, could not be found again:
the jumplist dropped closed panes. Every file opened (a look, a Save of a
scratch, a Restore, a rename) goes first in a list of 200, kept once, its
dot noted when it closes, kept across sessions in $XDG_STATE_HOME/pardes/
recent. `Recent` (SPC f r) shows it in a reused +Recent, each row
`path:line:col open|closed` a look reopens; /recent reads `open|closed
<path>`. The jumplist keeps a closed file's entries, +Jumps marks them
`(closed)`, and Back to one opens the file at its place. acme has no
such thing; its dump and Load are the nearest.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
build too
The tty, gui and detached shells allocate from std.start's gpa, which is a
DebugAllocator in Debug and, because we link libc, libc's malloc otherwise.
The macOS shell hardcoded std.heap.smp_allocator in every mode, so its Debug
build checked nothing it allocated outside memory.zig's subsystems.
Which allocator a release build should use was measured rather than
assumed, in ReleaseFast, with an experimental gui that chose at startup
between glibc's malloc and smp_allocator, with and without memory.zig's
stack-fallback buffers, plus an A/A pair; ten interleaved rounds, compared
round by round. The gui frame-cost harness (idle, scroll and typing in
src/pardes.zig, terminal spew, wheel-scrolling docs/design.pdf, first paint)
saw every candidate within the A/A pair's noise. Twelve rounds of the
allocation-heavy paths themselves, highlighting all of src/pardes.zig (305k
tree-sitter allocations) and pardes-pdf-bench's MuPDF renders, split them:
smp_allocator was 2 to 6% slower on MuPDF's page-sized rasters (slower in 9
to 11 of 12 rounds), no better on tree-sitter, and the stack-fallback buffers
changed nothing either way. So glibc's malloc it is, with the buffers kept.
macos.zig now takes a DebugAllocator in Debug, deinitialized in
pardes_deinit with leaks logged the way std.start treats the others', and
libc's malloc otherwise. main.zig says why init.gpa is kept.
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The registry sweep could delete a live socket, anywhere on the filesystem. A
reviewer reproduced it: a socket that is bound but has not reached listen(2)
answers ECONNREFUSED exactly like a dead one -- that window is every server's
startup -- and the sweep then followed the entry's symlink and unlinked
whatever absolute path it named. It now follows a target only into the
directory our own sockets live in and only to a `pardes-9p-*.sock` name, it
re-probes immediately before deleting rather than trusting a probe that is by
then several syscalls old, and a readlink that exactly filled its buffer is
treated as the truncation it is. The test grew a case for an entry whose
target is not ours: the entry goes, the file does not.
Ctrl-V in raw tty mode was a black hole when the yank register was empty --
neither typed nor forwarded -- so vim's visual block, readline's quoted-insert
and every other program's Ctrl-V simply vanished. With nothing to paste the
chord belongs to the program again.
The lone-ESC flush added earlier was dead code. vaxis already returns Escape
for a one-byte 0x1b (`Parser.parseGround` asserts `input.len == 1`), so the
carried byte it waited for can never exist; a reviewer showed a 3 ms gap and a
60 ms gap behaving identically. Removed rather than left to imply a guarantee
it never provided.
A shell whose editor is gone can start one again. Naming a live but
unreachable session made `pardes <file>` exit 1, which let a stale environment
variable lock someone out of their own editor; it falls through to an ordinary
session, as it did before the variable existed.
Also: the macOS ABI check for `pardes_topbar_pane_border_px` had been replaced
by a duplicate of the line above it; `--startup` now fails on a leak the way
every other measurement in that file does, and stops calling its maximum a p95
below twenty samples; the served README and the skill no longer tell you to
write to `data` with `>`, which truncates the whole body before the write
lands; `docs/v9fs.md` described the allocate-on-walk design that was rejected;
and `test/fs.py` keys nesting off `PARDES_PID`, so its forwarding case stops
passing only when the runner happens to be inside a live pardes.
fs-test now reaches its one documented pre-existing failure instead of dying
early. Suite 778/783 with the two known crashes.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A 9P round trip on a local Unix socket cost 9.7 ms against the SDL shell and
0.758 ms against the terminal one. The server was not slow and the wake was
not broken: instrumenting the path showed every request waking the loop early
(wakes=210, woke_early=212, timed_out=38 over 250 ticks) and being answered on
that same pass. The cost was that `pump` answers 9P at one point in a loop
that then renders and presents unconditionally, so a client's next request
landed while the main thread was blocked on the display, and each round trip
therefore cost a whole frame. The frame rate was governing something that has
nothing to do with drawing.
`Pardes.needs_frame` starts true, is set by every event except a tick with
nothing animating and a filesystem request that only reads, and is cleared
once a frame is presented. `pump` returns before render and present when it is
false and nothing is animating. An idle editor answering reads now draws
nothing at all.
9P read_fid, one RPC: gui 9.7 ms -> 0.056 ms (173x)
tty 0.758 ms -> 0.062 ms (12x)
Verified the shells still paint rather than going quiet: the rendered screen
carries the opened file, a write through 9P redraws within the frame, and
`fs-discovery-test` passes over the real wire. Suite unchanged at 778/783 with
the two pre-existing crashes.
Also from the adversarial review of the previous commits:
`pardes --tty FILE` silently discarded the file, and `--tty MISSING` silently
discarded the error pane. main.zig named the boot layout before the positional
was resolved, and naming one short-circuits `Boot.of`. The choice now happens
after the argument is known, and only when there is no file and no missing
word. macos.zig names the same layout, so the app no longer boots a different
one from the terminal and SDL shells.
`pre_close_last_pane_tail` was transcribed from the NEW default rather than the
old one, so the upgrade path it was added for did not exist: a workspace dumped
before the tagline reorder came back with the old default welded on as a custom
tail. It is now the string it claims to be.
A pane two rows tall lost its message and, worse, its prompt and the cursor
with it. The notice cap keeps the LAST notices now, because the prompt is last
and a prompt you cannot see is one you type into blind.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The 9P tree stops being a command language wearing a filesystem. /new
created a pane as a side effect of a *read*; it is now Tcreate in /pane,
with Tremove to close, which cloud9's engine has always supported and the
editor never declared: tree.zig now says
`features = .{ .create = true, .remove = true }`. Eleven pane ctl verbs
become files that can be read as well as written -- dot, limit, dirty,
mark, scroll, look, exec -- leaving ctl with `get`, the one verb no file
would say better. Root /ctl splits into a read-only /status and the
/look and /exec files whose write IS the click. stat carries real sizes
where it used to answer 0, and qid versions track a pane's revision, so a
client can poll for change without re-reading the body.
Commit a624a56 moved raw-tty keys to an early-return branch that knew only
Ctrl-B and bare Escape, and in the same edit deleted the paste branch below
it. That cost Shift-Escape (the unconditional way out of tty mode) and both
paste chords: Ctrl-V and Ctrl-Shift-V reached the child as keystrokes, so an
agent CLI running in a pane took Ctrl-V for its image-paste binding and
answered "No image found in clipboard". Both are restored, with tests.
Nested detection was not subtly broken but deleted: 60367d8 removed
nested.zig's process-ancestry walk and left "am I inside pardes" derived from
PARDES_FORWARD_LOOK, which read "0" both for --nested and for "the listener
did not come up". PARDES_PID now answers that question on its own, checked
with kill(pid, 0); PARDES_9P and PARDES_PANE answer how to reach it; the flag
is gone. The posted-9P registry also self-heals now -- a session that aborts
cannot unlink its own socket, so posting sweeps entries whose target refuses
a connection, symlinks only and on a definite ECONNREFUSED only.
Elsewhere: tty scrolling is sticky-bottom, following new output only from
the last row, with typing and entering raw mode snapping back to live; the
boot layouts are a Boot enum instead of a chain of ifs, and the bare tty
startup (Boot.tty, which main.zig names) opens an empty text pane under the
shell while tests keep Boot.tty_shell; builtins announce themselves on the
message row under a Verbose setting that is on by default; Config prints
each setting the way you would type it back, so WindowOpacity 70 rather than
"WindowOpacity: 70%"; LocationsConfig opens its window only when called bare;
every tagline puts the word that closes the thing last, and a column now
outlives its panes -- closing the last one leaves an empty pane, and only
Delcol, newly on the column tagline, takes the column away.
Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The served tree loses the self/ level: /index /ctl /new /log /screen
/listeners /pane/<n>/... /os, with /src only in -Dembed-sources=true builds
(default off, on for esp32p4). ctl speaks the editor's own language with two
lowercase verbs, look TEXT and exec TEXT, plus acme's addr verbs; the new/
factory directory becomes one clone file; cons is gone (exec Msg); name and
sel are files; stats report real lengths, modes and mtimes; /log streams
pane new/del/rename/save events. The tree code lives in src/ninep/
(tree, pane, ctl, addr, pty, events, screen, sources); fs.zig keeps host
access, mounts, resolution and find/grep. Same engine and transports.
README (fs-help.txt) and docs rewritten; tests updated and extended.
Co-Authored-By: Claude Fable 5.1 <[email protected]>
|
| | |
|
| | |
|