summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--src/pardes.zig38
-rw-r--r--src/term_pane.zig118
-rw-r--r--test/perf.zig244
3 files changed, 382 insertions, 18 deletions
diff --git a/src/pardes.zig b/src/pardes.zig
index e9cc746d..f95b4e69 100644
--- a/src/pardes.zig
+++ b/src/pardes.zig
@@ -4484,6 +4484,9 @@ pub const Pardes = struct {
surface: Surface = .{},
/// per-update scratch (paneCursorLines, selection text); reset each update
scratch: std.heap.ArenaAllocator,
+ /// the terminal motion surface, memoized against the pane it was built
+ /// for — see term_pane.RowsCache for the lifetime rule
+ shell_rows: term_pane.RowsCache = .{},
pub fn init(gpa: std.mem.Allocator, opts: Options) !*Pardes {
const image_gpa = opts.image_allocator orelse gpa;
@@ -4563,6 +4566,7 @@ pub const Pardes = struct {
if (p.last_dump) |d| gpa.free(d);
if (p.chord_arg) |a| gpa.free(a);
if (p.pipe_wait) |*wait| wait.deinit(gpa);
+ p.shell_rows.reset(gpa);
p.scratch.deinit();
gpa.free(p.surface.cells);
if (p.surface.images.len > 0) gpa.free(p.surface.images);
@@ -4594,6 +4598,7 @@ pub const Pardes = struct {
pub fn deinitPane(p: *Pardes, pane: *Pane) void {
std.debug.assert(pane.payloadValid());
+ p.shell_rows.dropPane(pane);
if (pane.image) |*iv| {
p.image_gpa.free(iv.path);
if (iv.rgba.len > 0) p.image_gpa.free(iv.rgba);
@@ -4808,6 +4813,10 @@ pub const Pardes = struct {
}
pub fn update(p: *Pardes, ev: Event) void {
+ // Free a motion surface that went bad during the LAST update, before
+ // anything in this one can ask for it. Nothing frees it mid-update:
+ // handlers hand `rows` around for the length of a single update.
+ p.shell_rows.sweep(p.gpa);
// A transient message is exactly as old as your last input: touch the
// keyboard or the mouse and it is gone, on every pane, because a
// message is a report and you have just proved you are back. Only
@@ -4862,6 +4871,7 @@ pub const Pardes = struct {
const pane = p.panes[o.pane] orelse return;
pane.appendTtyReplay(o.bytes);
pane.stream.nextSlice(o.bytes);
+ p.shell_rows.markStale(pane);
// follow new output to the bottom only in tty mode; in
// normal/insert the user is reading scrollback and the view
// holds still while output streams in. Tag edits hijack
@@ -6048,14 +6058,23 @@ pub const Pardes = struct {
return .{ .lines = ls, .row0 = 0 };
}
const rows = try term_pane.shellRows(p, pane);
+ // No edit buffer: the loop below would copy `rows` into a second array
+ // one slice at a time and return exactly what it was given. On a
+ // scrollback of any size that copy is the keystroke.
+ if (pane.ovl == null) return .{ .lines = rows, .row0 = 0 };
// the buffer may sit below the dump's last row (the grid's trailing
// blanks are trimmed off), so run the loop to whichever ends later
var last = rows.len;
if (pane.ovl) |o| last = @max(last, @as(usize, @intCast(@max(0, o.row + o.rows))));
var count = last;
if (pane.ovl) |o| {
+ // The split below writes ONE line for an empty buffer, but
+ // modal.lineCount("") is 0 — the same floor `n - 1` needs at the
+ // paste site. Without it, killing a whole line (`A<C-u>`, `d%`)
+ // sizes `ls` one short of what the loop writes, and on a buffer
+ // covering the last row that is a length of zero.
if (o.row >= 0 and @as(usize, @intCast(o.row)) < last)
- count = count - @min(@as(usize, @intCast(o.rows)), last - @as(usize, @intCast(o.row))) + modal.lineCount(o.text);
+ count = count - @min(@as(usize, @intCast(o.rows)), last - @as(usize, @intCast(o.row))) + @max(1, modal.lineCount(o.text));
}
const ls = try arena.alloc([]const u8, count);
var n: usize = 0;
@@ -6097,6 +6116,13 @@ pub const Pardes = struct {
fn flatSurface(p: *Pardes, pane: *Pane, pl: PaneLines) ![]const u8 {
if (pane.file) |f| return f.content;
if (hasPdf(pane)) return p.ensurePdfText(pane);
+ // The memo already holds these rows joined by '\n' — that is what its
+ // `text` IS. The identity check is the same argument hxOff makes below:
+ // `lines` came out of the cache untouched (no edit buffer stood in for
+ // any row), so the join cannot have drifted from it.
+ const c = &p.shell_rows;
+ if (!c.stale and c.pane == pane and
+ pl.lines.ptr == c.rows.ptr and pl.lines.len == c.rows.len) return c.text;
const arena = p.scratch.allocator();
var total: usize = if (pl.lines.len > 0) pl.lines.len - 1 else 0;
for (pl.lines) |ln| total += ln.len;
@@ -6127,6 +6153,15 @@ pub const Pardes = struct {
/// differential and 561 parity cases say so; the arithmetic below is
/// hxOff/hxPos transcribed, with line i's terminator read off idx[i+1]-1
/// instead of searched for.
+ ///
+ /// ponytail: a terminal's motion surface is NOT one of the things with no
+ /// index any more — shell_rows.rows are slices into shell_rows.text, so
+ /// their offsets ARE the line starts, for free. Claiming them here needs
+ /// `p`, which means threading it through paneOff/panePos/paneLineStart and
+ /// their callers (paneRange, setPaneRange, the three motion helpers). Left
+ /// undone deliberately: measured, that is a terminal key-down of 615us
+ /// against 140us, and 0.6ms is not a keystroke anybody feels. Do it the
+ /// day one of those scans shows up next to something that matters.
fn indexOf(pane: *Pane, text: []const u8) ?[]const usize {
const f = if (pane.file) |*ff| ff else return null;
if (text.ptr != f.content.ptr or text.len != f.content.len) return null;
@@ -12563,6 +12598,7 @@ pub const Pardes = struct {
// the size so bodyText renders the right number of rows
if (pane.isTerminal()) {
pane.vt.resize(p.gpa, .{ .cols = cols, .rows = rows }) catch {};
+ p.shell_rows.markStale(pane); // reflow moved every row
p.emit(.{ .resize_pty = .{ .pane = @intCast(id), .cols = cols, .rows = rows } });
}
pane.cols = cols;
diff --git a/src/term_pane.zig b/src/term_pane.zig
index 5e50e15d..9f13832f 100644
--- a/src/term_pane.zig
+++ b/src/term_pane.zig
@@ -19,34 +19,134 @@ const EditSnap = pardes.EditSnap;
const Ovl = pardes.Ovl;
const modal = @import("modal.zig");
+/// The memo behind `shellRows`. ONE entry for the editor, because the motion
+/// surface is built for the pane the cursor is in and a second pane asking
+/// would only double a multi-megabyte buffer for a slot it is about to lose
+/// again. gpa-owned rather than scratch-arena: the whole point is to outlive
+/// the update that built it.
+///
+/// LIFETIME, the part that would rot silently: `rows` is handed out to
+/// callers, so the buffers are freed in exactly two places — `sweep`, at the
+/// TOP of an update before any handler can be holding them, and `reset` when
+/// the editor goes away. Everything that notices the entry has gone bad
+/// (output arrived, the grid reflowed, the pane died, another pane wants the
+/// slot) only marks it `stale`; nothing frees mid-update. That is the same
+/// guarantee the scratch arena gave, spelled out.
+pub const RowsCache = struct {
+ /// whose grid this describes; null = the slot is free
+ pane: ?*const Pane = null,
+ /// the rows joined by '\n' — `flatSurface` hands this back verbatim
+ /// instead of rebuilding the join on every keystroke
+ text: []const u8 = &.{},
+ /// slices INTO `text`, absolute grid rows from 0
+ rows: [][]const u8 = &.{},
+ /// `text` is a prefix of this: blanking a prompt row shortens the join,
+ /// and the slack is not worth a second allocation to reclaim
+ text_alloc: []u8 = &.{},
+ stale: bool = false,
+
+ pub fn reset(c: *RowsCache, gpa: std.mem.Allocator) void {
+ if (c.text_alloc.len > 0) gpa.free(c.text_alloc);
+ if (c.rows.len > 0) gpa.free(c.rows);
+ c.* = .{};
+ }
+
+ /// Free a stale entry. Called at the top of `update`, and nowhere else.
+ pub fn sweep(c: *RowsCache, gpa: std.mem.Allocator) void {
+ if (c.stale) c.reset(gpa);
+ }
+
+ /// `pane`'s grid moved: the entry no longer describes it.
+ pub fn markStale(c: *RowsCache, pane: *const Pane) void {
+ if (c.pane == pane) c.stale = true;
+ }
+
+ /// `pane` is being destroyed. Drop the pointer now — a freed pane's
+ /// address can come back from the allocator as a different pane, and an
+ /// entry still naming it would answer for the wrong grid — but leave the
+ /// buffers to the next sweep, as ever.
+ pub fn dropPane(c: *RowsCache, pane: *const Pane) void {
+ if (c.pane != pane) return;
+ c.pane = null;
+ c.stale = true;
+ }
+};
+
+const Rows = struct {
+ text_alloc: []u8,
+ text: []const u8,
+ rows: [][]const u8,
+};
+
/// A terminal's shell rows as the surface sees them: the WHOLE
/// history+active grid, prompt rows blanked (OSC 133), absolute grid rows
/// from 0. The raw material the motion surface is composed from — the
/// edit buffer is NOT applied here, so it is also what seeding the buffer
-/// reads. Scratch-arena backed.
+/// reads.
/// ghostty's dump trims the grid's trailing blank rows; ONE of them is
/// kept back, the row the cursor sits on below the last line of output.
/// That row is a file's final newline: without it the surface would have
/// one line fewer than the same text in a document, and every motion and
/// linewise edit at the bottom would diverge.
-/// ponytail: O(scrollback) dump+scan per keystroke — window it around the
-/// viewport if a multi-MB scrollback ever makes navigation lag.
+///
+/// Building it is O(scrollback) — a dump of the whole history — and a
+/// keystroke asks for it once or twice, so the result is memoized against the
+/// pane until its grid changes. A pane sitting on 16 MiB of agent transcript
+/// paid that dump per press of `j` before the cache; now it pays it once per
+/// chunk of output.
pub fn shellRows(p: *Pardes, pane: *Pane) ![]const []const u8 {
- const arena = p.scratch.allocator();
- const full = try pane.vt.screens.active.dumpStringAlloc(arena, .{ .screen = .{} });
+ const c = &p.shell_rows;
+ if (!c.stale and c.pane == pane) return c.rows;
+ if (c.pane != null) {
+ // Another pane holds the slot. Take it for the NEXT update (the sweep
+ // frees what is there) and answer this one from scratch: whoever owns
+ // the live entry may still be holding the rows it handed out.
+ c.stale = true;
+ return (try buildRows(p.scratch.allocator(), p, pane)).rows;
+ }
+ const built = try buildRows(p.gpa, p, pane);
+ c.* = .{
+ .pane = pane,
+ .text = built.text,
+ .rows = built.rows,
+ .text_alloc = built.text_alloc,
+ };
+ return c.rows;
+}
+
+fn buildRows(alloc: std.mem.Allocator, p: *Pardes, pane: *Pane) !Rows {
+ const full = try pane.vt.screens.active.dumpStringAlloc(p.scratch.allocator(), .{ .screen = .{} });
var pit = pane.vt.screens.active.pages.rowIterator(.right_down, .{ .screen = .{} }, null);
// split yields one more item than delimiters; the extra final slot is the
// cursor row retained below.
- const rows = try arena.alloc([]const u8, std.mem.count(u8, full, "\n") + 2);
+ const n_rows = std.mem.count(u8, full, "\n") + 2;
+ const rows = try alloc.alloc([]const u8, n_rows);
+ errdefer alloc.free(rows);
+ // Blanking a prompt row only ever SHORTENS it and the retained cursor row
+ // adds one separator, so the dump's length plus one bounds the join.
+ const text = try alloc.alloc(u8, full.len + 1);
+ errdefer alloc.free(text);
+ var at: usize = 0;
var n: usize = 0;
var it = std.mem.splitScalar(u8, full, '\n');
while (it.next()) |raw| {
const is_prompt = if (pit.next()) |pin| pin.rowAndCell().row.semantic_prompt != .none else false;
- rows[n] = if (is_prompt) "" else raw;
+ const shown = if (is_prompt) "" else raw;
+ if (n > 0) {
+ text[at] = '\n';
+ at += 1;
+ }
+ @memcpy(text[at..][0..shown.len], shown);
+ rows[n] = text[at..][0..shown.len];
+ at += shown.len;
n += 1;
}
- rows[n] = "";
- return rows;
+ text[at] = '\n';
+ at += 1;
+ rows[n] = text[at..][0..0];
+ n += 1;
+ std.debug.assert(n == n_rows);
+ return .{ .text_alloc = text, .text = text[0..at], .rows = rows };
}
/// The body a terminal renders: the viewport's shell rows (prompt rows blanked
diff --git a/test/perf.zig b/test/perf.zig
index bf5f0200..f440227d 100644
--- a/test/perf.zig
+++ b/test/perf.zig
@@ -102,6 +102,65 @@ const Op = enum {
}
};
+// ---- the terminal scoreboard ----
+//
+// The other axis pardes scales on, and the one a file table cannot see. A
+// terminal pane's state is a ghostty-vt emulator with a 16 MiB scrollback, and
+// three of the costs below walk ALL of it rather than the viewport: a COLUMN
+// change reflows every page in the list (a row change does not — ghostty skips
+// reflow when the width is unchanged, which is why both are here), and the
+// motion surface is built from a dump of the whole history.
+//
+// The case this exists for is a coding agent printing a long transcript into a
+// shell pane: `resize-cols` is what one frame of a window drag costs, and
+// `key-down` is what one press of `j` costs afterwards.
+
+const TermFixture = struct {
+ name: []const u8,
+ /// KiB of pty output fed into the pane before any clock starts
+ kb: usize,
+};
+
+const term_fixtures = [_]TermFixture{
+ // a shell you just opened: the control. Anything here is constant cost.
+ .{ .name = "sb-64k", .kb = 64 },
+ // a build log — also where the 1 MiB raw-byte replay ring fills up
+ .{ .name = "sb-1m", .kb = 1024 },
+ // an agent transcript: half the scrollback ceiling
+ .{ .name = "sb-8m", .kb = 8 * 1024 },
+};
+
+const TermOp = enum {
+ /// one redraw, nothing changed — the floor the rest sit on
+ render,
+ /// one 4 KiB pty read arrives: parse, replay ring, sync, frame
+ output,
+ /// the window got one row shorter: resize WITHOUT reflow
+ resize_rows,
+ /// the window got one column narrower: a full page-list reflow
+ resize_cols,
+ /// `j` in the pane's body — builds the motion surface from the grid
+ key_down,
+ /// one printable typed into the pane's edit buffer
+ edit_char,
+
+ fn label(o: TermOp) []const u8 {
+ return switch (o) {
+ .render => "render",
+ .output => "output",
+ .resize_rows => "resize-rows",
+ .resize_cols => "resize-cols",
+ .key_down => "key-down",
+ .edit_char => "edit-char",
+ };
+ }
+};
+
+/// One pty read's worth of output, built once and replayed by the `output`
+/// row. Blocks are generated, not captured, for the same reason the file
+/// fixtures are.
+var term_chunk: []const u8 = &.{};
+
const Cell = struct {
min_us: u64 = 0,
med_us: u64 = 0,
@@ -154,8 +213,16 @@ pub fn main(init: std.process.Init) !void {
}
}
- if (json) return reportJson(init.io, &cells, &bytes, reps);
- reportText(&cells, &bytes, reps, base_path);
+ term_chunk = try buildTermText(4 * 1024);
+ var term_cells: [std.enums.values(TermOp).len][term_fixtures.len]Cell = undefined;
+ for (std.enums.values(TermOp), 0..) |op, oi| {
+ for (term_fixtures, 0..) |fx, fi| {
+ term_cells[oi][fi] = try measureTerm(op, fx, reps);
+ }
+ }
+
+ if (json) return reportJson(init.io, &cells, &term_cells, &bytes, reps);
+ reportText(&cells, &term_cells, &bytes, reps, base_path);
}
// ---- the measured gestures ----
@@ -222,6 +289,98 @@ fn measure(op: Op, fx: Fixture, path: []const u8, reps: usize) !Cell {
return summarize(samples);
}
+/// One (op, fixture) cell of the terminal table. Same contract as `measure` —
+/// fresh core per cell, warmups, median of `reps` — with no `seek`: a terminal
+/// has one cursor, and everything measured here scales with the size of the
+/// HISTORY rather than with where in it you are standing.
+fn measureTerm(op: TermOp, fx: TermFixture, reps: usize) !Cell {
+ const warmup = 3;
+ const samples = try gpa.alloc(u64, reps);
+ defer gpa.free(samples);
+
+ const core = try bootTerm(fx);
+ defer core.deinit();
+ const pane = core.panes[0] orelse fatal("terminal boot produced no pane", .{});
+ // tty mode hands every key straight to the shell, so the two gesture rows
+ // would otherwise measure one queued write effect and nothing else
+ if (op == .key_down or op == .edit_char) pane.mode = .normal;
+ if (op == .edit_char) {
+ core.update(.{ .key = .{ .cp = 'i', .text = "i" } });
+ pump(core);
+ }
+ _ = try frame(core);
+
+ var cols = screen_cols;
+ var rows = screen_rows;
+ for (0..warmup + reps) |n| {
+ const t0 = nowNs();
+ switch (op) {
+ .render => {},
+ .output => core.update(.{ .output = .{ .pane = 0, .bytes = term_chunk } }),
+ // one cell of window drag. Alternating rather than sweeping so
+ // every sample does a real resize and the fixture never drifts
+ // away from the size the other rows are measured at.
+ .resize_rows => {
+ rows = if (rows == screen_rows) screen_rows - 1 else screen_rows;
+ core.update(.{ .resize = .{ .cols = cols, .rows = rows } });
+ },
+ .resize_cols => {
+ cols = if (cols == screen_cols) screen_cols - 1 else screen_cols;
+ core.update(.{ .resize = .{ .cols = cols, .rows = rows } });
+ },
+ .key_down => core.update(.{ .key = .{ .cp = 'j', .text = "j" } }),
+ .edit_char => core.update(.{ .key = .{ .cp = 'x', .text = "x" } }),
+ }
+ pump(core);
+ _ = try frame(core);
+ const dt = nowNs() -| t0;
+ if (n >= warmup) samples[n - warmup] = dt / 1000;
+ }
+ return summarize(samples);
+}
+
+/// A shell pane carrying `fx.kb` KiB of history. The bytes go in as 8 KiB
+/// reads, which is both what a pty delivers and what the flood costs: one
+/// core update per chunk.
+fn bootTerm(fx: TermFixture) !*pardes.Pardes {
+ const core = try pardes.Pardes.init(gpa, .{ .tty_only = true });
+ core.update(.{ .resize = .{ .cols = screen_cols, .rows = screen_rows } });
+ pump(core);
+ const text = try buildTermText(fx.kb * 1024);
+ defer gpa.free(text);
+ var off: usize = 0;
+ while (off < text.len) {
+ const end = @min(off + 8192, text.len);
+ core.update(.{ .output = .{ .pane = 0, .bytes = text[off..end] } });
+ pump(core);
+ off = end;
+ }
+ return core;
+}
+
+/// Plausible pty output, at least `want` bytes of it: an OSC 133-marked
+/// prompt, the command, and a run of coloured result lines. The prompt markers
+/// are the point — the motion surface blanks prompt rows, so a history without
+/// them measures a branch no real shell ever takes.
+fn buildTermText(want: usize) ![]const u8 {
+ var out: std.Io.Writer.Allocating = .init(gpa);
+ errdefer out.deinit();
+ var n: usize = 0;
+ while (out.written().len < want) : (n += 1) {
+ try out.writer.print(
+ "\x1b]133;A\x07\x1b[32muser\x1b[0m@host \x1b[34m~/work\x1b[0m $ \x1b]133;B\x07zig build -Dstep={d}\r\n\x1b]133;C\x07",
+ .{n},
+ );
+ for (0..12) |k| {
+ try out.writer.print(
+ "\x1b[90m[{d:0>5}]\x1b[0m compiling module_{d}_{d} ... \x1b[32mok\x1b[0m ({d} ms)\r\n",
+ .{ n, n, k, (n *% 7919 +% k) % 900 },
+ );
+ }
+ }
+ return out.toOwnedSlice();
+}
+
/// Park the cursor and the view at sample `n`'s position, walked across the
/// file by a prime stride so consecutive samples land nowhere near each other
/// and a whole run covers the document rather than one neighbourhood of it.
@@ -345,9 +504,15 @@ fn writeFile(path: []const u8, text: []const u8) !void {
// ---- reporting ----
-fn reportText(cells: *const [std.enums.values(Op).len][fixtures.len]Cell, bytes: *const [fixtures.len]usize, reps: usize, base_path: ?[]const u8) void {
+fn reportText(
+ cells: *const [std.enums.values(Op).len][fixtures.len]Cell,
+ term_cells: *const [std.enums.values(TermOp).len][term_fixtures.len]Cell,
+ bytes: *const [fixtures.len]usize,
+ reps: usize,
+ base_path: ?[]const u8,
+) void {
const o = std.debug.print;
- const base = if (base_path) |bp| readBase(bp) else null;
+ const base = if (base_path) |bp| readBase(bp, reps) else null;
o("pardes perf — {d}x{d} viewport, {d} samples/cell, median us\n\n", .{ screen_cols, screen_rows, reps });
o("{s:<12}", .{"fixture"});
@@ -390,6 +555,33 @@ fn reportText(cells: *const [std.enums.values(Op).len][fixtures.len]Cell, bytes:
}
o("\n", .{});
}
+
+ // second table, same shape: the terminal costs, against SCROLLBACK
+ o("\n{s:<12}", .{"scrollback"});
+ for (term_fixtures) |fx| {
+ if (base != null) o(" {s:>19}", .{fx.name}) else o(" {s:>12}", .{fx.name});
+ }
+ o("\n{s}\n", .{if (base != null) "-" ** (12 + term_fixtures.len * 20) else "-" ** (12 + term_fixtures.len * 13)});
+ for (std.enums.values(TermOp), 0..) |op, oi| {
+ o("{s:<12}", .{op.label()});
+ for (0..term_fixtures.len) |fi| {
+ const c = term_cells[oi][fi];
+ if (c.med_us > 0) {
+ const j = (@as(f64, @floatFromInt(c.p90_us)) - @as(f64, @floatFromInt(c.med_us))) / @as(f64, @floatFromInt(c.med_us));
+ if (j > worst_jitter) worst_jitter = j;
+ }
+ if (base) |b| {
+ const prev = b.term_med[oi][fi];
+ if (prev == 0) {
+ o(" {d:>12}{s:>7}", .{ c.med_us, "-" });
+ } else {
+ const ratio = @as(f64, @floatFromInt(c.med_us)) / @as(f64, @floatFromInt(prev));
+ o(" {d:>12} {d:>6.2}x", .{ c.med_us, ratio });
+ }
+ } else o(" {d:>12}", .{c.med_us});
+ }
+ o("\n", .{});
+ }
// The spread is not only scheduler noise: samples are taken at DIFFERENT
// places in the file on purpose (see seek), so a cost that still depends on
// where the cursor is shows up here as well. Both are reasons not to
@@ -406,7 +598,13 @@ fn reportText(cells: *const [std.enums.values(Op).len][fixtures.len]Cell, bytes:
/// would make the next comparison silently print no ratios at all.
const json_report_max_bytes = 32 * 1024;
-fn reportJson(io: std.Io, cells: *const [std.enums.values(Op).len][fixtures.len]Cell, bytes: *const [fixtures.len]usize, reps: usize) void {
+fn reportJson(
+ io: std.Io,
+ cells: *const [std.enums.values(Op).len][fixtures.len]Cell,
+ term_cells: *const [std.enums.values(TermOp).len][term_fixtures.len]Cell,
+ bytes: *const [fixtures.len]usize,
+ reps: usize,
+) void {
var storage: [json_report_max_bytes]u8 = undefined;
var out: std.Io.Writer = .fixed(&storage);
out.print("{{\"cols\":{d},\"rows\":{d},\"reps\":{d},\"fixtures\":[", .{ screen_cols, screen_rows, reps }) catch return;
@@ -426,22 +624,40 @@ fn reportJson(io: std.Io, cells: *const [std.enums.values(Op).len][fixtures.len]
first = false;
}
}
+ for (std.enums.values(TermOp), 0..) |op, oi| {
+ for (term_fixtures, 0..) |fx, fi| {
+ const c = term_cells[oi][fi];
+ out.print("{s}{{\"op\":\"{s}\",\"fixture\":\"{s}\",\"min_us\":{d},\"med_us\":{d},\"p90_us\":{d},\"max_us\":{d}}}", .{
+ if (first) "" else ",", op.label(), fx.name, c.min_us, c.med_us, c.p90_us, c.max_us,
+ }) catch return;
+ first = false;
+ }
+ }
out.writeAll("]}\n") catch return;
std.Io.File.stdout().writeStreamingAll(io, out.buffered()) catch {};
}
-const Base = struct { med: [std.enums.values(Op).len][fixtures.len]u64 };
+const Base = struct {
+ med: [std.enums.values(Op).len][fixtures.len]u64,
+ term_med: [std.enums.values(TermOp).len][term_fixtures.len]u64,
+};
/// A previous --json run, reduced to the medians this table compares against.
/// Cells the old run did not have stay 0 and print as "-": the op list may
/// have grown since, and a missing number is not a regression. An unreadable
/// file is fatal rather than silently ratio-less — a comparison you asked for
/// and did not get is worse than no comparison.
-fn readBase(path: []const u8) ?Base {
+///
+/// So is one that is quietly WRONG, which is why `reps` has to match. `seek`
+/// walks sample n to `n *% 7919 % span`, so two runs with different counts
+/// measure different PLACES in the file, and every file row comes out 20-50%
+/// apart with nothing whatever having changed.
+fn readBase(path: []const u8, reps: usize) ?Base {
const src = readFileAlloc(path) catch fatal("--base: cannot read {s}", .{path});
defer gpa.free(src);
- var b: Base = .{ .med = @splat(@splat(0)) };
+ var b: Base = .{ .med = @splat(@splat(0)), .term_med = @splat(@splat(0)) };
const parsed = std.json.parseFromSlice(struct {
+ reps: usize = 0,
cells: []const struct {
op: []const u8,
fixture: []const u8,
@@ -449,6 +665,10 @@ fn readBase(path: []const u8) ?Base {
},
}, gpa, src, .{ .ignore_unknown_fields = true }) catch return null;
defer parsed.deinit();
+ if (parsed.value.reps != reps) fatal(
+ "--base: {s} was taken with --reps {d}, this run is --reps {d}. Same count or no comparison.",
+ .{ path, parsed.value.reps, reps },
+ );
for (parsed.value.cells) |c| {
for (std.enums.values(Op), 0..) |op, oi| {
if (!std.mem.eql(u8, op.label(), c.op)) continue;
@@ -456,6 +676,14 @@ fn readBase(path: []const u8) ?Base {
if (std.mem.eql(u8, fx.name, c.fixture)) b.med[oi][fi] = c.med_us;
}
}
+ // op labels repeat across the two tables ("render", "key-down"); the
+ // fixture names never do, so the pair still names exactly one cell
+ for (std.enums.values(TermOp), 0..) |op, oi| {
+ if (!std.mem.eql(u8, op.label(), c.op)) continue;
+ for (term_fixtures, 0..) |fx, fi| {
+ if (std.mem.eql(u8, fx.name, c.fixture)) b.term_med[oi][fi] = c.med_us;
+ }
+ }
}
return b;
}