Found during #155's sweep and recorded there as one line of prose with an
uncommitted proptest seed. This turns it into something that cannot be
lost and that states what is actually known.
Reproduced on main @ e745068 with PROPTEST_CASES=2000 against
rope_matches_crdt_projection_after_arbitrary_edits, then reduced by hand
to five lines: replacing a byte range with IDENTICAL bytes is a textual
no-op but a real CRDT operation, so undoing it advances the CRDT version
while leaving the materialized text unchanged. `undo_crdt_mode` derives
an empty replacement edit and still attaches the op `crdt.undo()`
produced, which trips the proptest's "a no-op edit must have crdt_op =
None" invariant.
The test is `#[ignore]`d rather than asserted-as-correct or left as a
seed, because the resolution is a judgement call I should not make
silently:
- the raw proptest seed is NOT committed, since it would fail the suite
on every run for a case whose correct outcome is undecided;
- the deterministic reduction reproduces without any seed, so nothing is
lost by leaving the seed out.
What the doc comment records, so the next reader need not redo it:
content stays correct (rope and CRDT projection agree either side); both
`crdt_op` consumers read the field unconditionally and do not
short-circuit on an empty range, so replicas still converge; and
`EditorCore::undo` never seeks `edit.range.start`, so no cursor jumps.
The open question is whether the invariant is simply mis-scoped — it was
written for the forward `apply_edit` short-circuit, which returns before
producing an op, and CRDT-mode undo/redo never reach that path. One
artifact is arbitrary either way: `derive_replacement_edit` reports the
empty range at the buffer END rather than at the edit site.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Handoff §1 gains the arc entry: the window-parameter substrate, the two
production `Layout::compute` callers, the recursive minima, hiding as a
durable transition, per-window input gating, the per-frontend jump
origins, and the shared initial-target load seam. §5 gains four durable
lessons, three of them the same class:
- a guard with no production caller passes every direct-call test;
- a geometric readout (`at_bottom`) is not a state predicate;
- a PTY does not translate LF to CRLF, so text equality over clipped
output is vacuous;
- widening an ambient resolver into a scoped one can make a total
function partial — which is what took CI red on all four Test jobs.
The roadmap position, the arc's named deferrals, and DAP's unblocking
are recorded too.
active-work closes the lane, refreshes the canonical base to `e745068`,
and keeps Stage 2's named obligations plus the two gating facts found on
the way (the sweep needs an isolated XDG_CONFIG_HOME; compile_mode
acceptance is load-sensitive, verified pre-existing).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
F1 was a real bug pinned by my own committed test. `end` in
ChunkSource::MathBox is EXCLUSIVE, so source position `end` is the first byte
AFTER the span — but the arm claimed it for the box's left edge, and the test
asserted that wrong value while calling the byte "interior". Consequences it
would have caused once overlays land: a search match starting just after a
span washes the whole box it does not intersect, violating Q#MS11; a peer
caret after the span draws at the box's left edge; caret geometry jumps
backwards. The same class existed in projected_to_source for a line-FINAL box,
where `within` clamps to the run length and the arm returned `start`
unconditionally, so a click past end-of-line landed on the span start. Both
committed hit tests put a chunk after the box, so that edge was never
exercised; there is now a test with the box last.
F2: parse_scripts peeked for the next marker without skipping whitespace, so
`x^2 _i` built a NESTED script — drawing the subscript displaced right by the
superscript's width — and `x^2 ^3` parsed where TeX errors, contradicting the
module's own "whitespace is insignificant" rule.
F3: layout is now fallible. A character the math font cannot draw used to
yield zero metrics and still emit a Glyph item, rendering tofu at zero advance
over its neighbour. Q#MS8's rule is "failure is always show the source", and
the draw pass needs a refusal signal — changed now, before that pass consumes
the API.
F4: the fraction gap was a hardcoded `thickness * 2.0` while the MATH table's
FractionNumeratorGapMin / FractionDenominatorGapMin went unread. Reading them
moved the flagship \frac{a}{b} from 0.732 to 0.867 and the fallback boundary
from depth 3 to depth 5. The round-2 review's hand-arithmetic estimate of
~0.85 was right; my 0.732 was inflated by the guess. The depth-SEARCHING test
absorbed the change without edits, which is the property it was written for.
F5: TeX's \epsilon and \phi are the lunate/symbol forms (U+03F5, U+03D5), not
U+03B5/U+03C6. Their italic mappings had to land with the seed change, since
both sit outside math_italic's U+03B1..03C9 run and would otherwise render
upright beside italic neighbours.
F7: the line-box budget derivation moved out of the test into
`line_box_budget`, so the draw pass and the acceptance test cannot compute
different splits while both stay green.
F8: the live-code clippy items are cleared. The 25 that remain are all
dead-code awaiting the draw pass.
189 pmacs-gpu tests pass.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
PR #155 review round 2, self-review of the round-2 commit.
The round-2 change labelled "minor" — resolving both arms of
pmacs.window.buffer() through the acting frontend for uniformity — made
the NO-ARGUMENT arm fallible. `acting_frontend` follows the interactive
origin, which can name a frontend that has no registered view: a bare
`dispatch_key` from an unattached peer does exactly that. `selected_window`
then raises "acting frontend has no layout" instead of answering.
Nothing surfaced that error, because the runtime callers do not pcall it.
killring, syntax, autosave, pair, indent and comment all read
pmacs.window.buffer() on ordinary edits, so the raise silently dropped
the operation: kill_ring_acceptance went 30/30 to 25/5, with
frontend_detached_drops_per_frontend_state reporting only "B has kill
state". main is 30/30, and reverting this one file restored it.
The no-arg arm is back on ambient active_buffer_id() and now documents
why that is deliberate rather than an oversight: dispatch sets
active_frontend to the acting frontend before running a command, so the
two agree on every real path, while only the ambient resolver has the
fallback that makes it total. The explicit-window arm keeps its Q#BP11
layout validation, which is what the arc actually needed.
acc19c pins it through the real path — a buffer.after-edit subscriber
reading pmacs.window.buffer() during a viewless peer's dispatch_key —
rather than by calling the binding directly. Bite-verified:
scripts/bite bbe4152 src/lua_bindings/mod.rs --test
bottom_panel_stage1_acceptance -- acc19c goes red with the exact
"acting frontend has no layout" traceback.
The ledger also records two gating facts found on the way: the workspace
sweep must run with an isolated XDG_CONFIG_HOME, because the real user
init.lua installs a local package and the losing race leaks a status
message into painted-frame comparisons; and a latent pre-existing main
bug in the buffer CRDT undo path, which is not this branch's and whose
proptest seed is deliberately not committed here.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012j4omtTMn9v1UfmHQb9ap6
The suppression mechanism F2 forced: a RichChunk's only width is its text, so
a suppressed span reserves room with SPACER SPACES the way SourceTab already
does, quantized up to whole advances. Quantizing up keeps the projection
grid-aligned with the surrounding monospace text and keeps hit runs integral,
at the cost of under one advance of slack on the right.
Adding the variant to an exhaustive enum made the compiler enumerate every
seam it must participate in, which is why it is wired through all five rather
than the two I had in mind: projected_to_source, source_to_projected, the tab
expander's source remap, and offset_chunk_source. Hits anywhere inside a box
snap to the span start — the Adornment rule, because Q#MS4 gives the box no
interior byte map — and source positions inside it collapse to the box's left
edge, so text after the span accounts for the whole reserved width.
Two details the tab expander needed: a math chunk's spacer text is generated
rather than source, so it holds no tab byte to expand, and its suppressed
range is already in slice coordinates and never split, so a within-chunk
offset does not move it.
spacer_for_width guards its inputs: a non-finite width, a non-positive
advance, or a pathological ratio reserves nothing or clamps, rather than
panicking or minting an enormous string from a cast.
184 pmacs-gpu tests pass, including the 155 that predate this branch.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Per-line `$…$` scanning with the Pandoc guards F5 required: an opener must be
followed by a non-space, a closer must be preceded by a non-space and not
followed by a digit, and `\$` escapes. Without them "prices are $5 and $6
today" renders "5 and " as math, in exactly the grammar-less prose buffers
this scanner targets. Spans never cross a newline (Q#MS3), so callers scan one
line at a time.
`$$` is opaque, and the test that forced this is worth keeping in mind. My
first version simply refused to OPEN on `$$`, reasoning that display math
would then never match. It still did: in `$$x$$` the first `$` declines to
open, the second one opens, and the third closes it — matching the inner `$x$`,
whose interior parses perfectly well. So display math would have half-rendered
as math with a stray `$` on each side, which is precisely what acceptance 15
forbids. `$$` now neither opens nor closes and abandons any pending opener.
The test that caught it was itself nearly vacuous. It asserted "every span
found must fail to parse", which passes trivially when the matched interior is
`x`. Asserting that NO span is found is the form that actually holds the
contract; the weaker version would have shipped the bug.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
MathBox/MathItem composition for the Q#MS2 subset: characters advance a pen,
scripts shift by the MATH table's superscript/subscript amounts at script
scale, and \frac stacks its operands around a rule at the math axis. Inline
\frac sets its operands one style down, which is TeX's rule and also what the
parent framing's Tier 3 specifies — and it is load-bearing for Q#MS10, since
full-size operands would not fit the line at all.
The height budget is now measured rather than assumed, and the round-2 review
was right to insist on that. Two things were wrong.
First, my own test derived the budget from the MATH font's metrics. Q#MS10
says the budget is the LINE BOX, whose baseline the CODE font places —
JetBrains Mono ascends 16.32 px and descends 4.80 px at 16 px inside the 22 px
line, against Latin Modern Math's 12.90/3.10. Using the wrong font made a
plain \frac{a}{b} score 0.485 and appear to fall below the floor, which would
have meant the flagship case never rendering.
Second, with the budget derived correctly, B6 holds — \frac{a}{b} scales to
0.732 — but rev 3's guessed fallback case does not. A doubly-nested fraction
scores 0.744 and still renders; the floor is not tripped until depth 3, at
0.580. Round 2 predicted precisely this surprise-pass. Worth keeping: depth 2
scores HIGHER than depth 1, because the binding constraint flips from descent
to ascent as nesting grows asymmetrically, so "deeper is always tighter" is
false.
The test therefore SEARCHES for the tripping depth instead of hardcoding it,
and fails if no depth trips the floor at all — which would mean the fallback
arm is unreachable and the floor is dead code. Acceptance 12 records the
measured table.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The round-2 peer press landed in the peer's CONTENT area, so it never
reached `arm_window_drag` — the exact path Finding 5 names — and the
case bit nothing. It now presses the peer's own mode-line row, where a
single global drag slot is overwritten (and, since that lone window owns
no boundary, cleared outright).
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
PR #155 review round 2.
Finding 1 (must fix): Q#BP7 item 1 — "growth reaching the live tail
re-arms follow (top -> None), only when no selection is active" — was
never implemented. `at_bottom` is the instantaneous geometric readout
`scroll_offset == 0`, which a still-anchored view satisfies whenever it
happens to be tall enough to reach the tail, so the round-1 assertion
could not see the gap: the next rows the child printed pushed the
anchored view back into history.
`rearm_follow_on_growth` now clears `top` when a viewport-size
declaration makes the view cover the tail and no selection is frozen,
and every size-declaring path (`snapshot_for_view`, `record_view_size`,
`view_status_for_size`) routes through one `declare_view_size` helper so
grid and semantic declarations cannot disagree. `scroll_view` and
`begin_selection` deliberately stay out: they write `top` themselves,
and `scroll_view` already owns the scroll-driven arm.
New acc32b is the pin the review asked for: scroll into history, grow
past the tail, then release a SECOND burst of child output through a
filesystem gate and assert the view moved with it.
Finding 2: the PTY fixtures emitted LF-only output, which staircases
rightward until every row clips to blanks past the viewport width — so
the round-1 anchor assertions compared "" with "" and could not fail.
Both fixtures now emit CRLF, and each anchor comparison is guarded by
`assert!(!top_before.is_empty())`.
Finding 3: acc33's contrast case asserted nothing, and the behavior it
claimed was false as coded. With the re-arm in place it is true and now
asserted: clearing the selection at the same geometry re-arms follow and
leaves the frozen anchor.
Finding 4: `start_run` gated the panel branch on `display == "panel" or
already_in_panel(..)`, so an explicit `display = "current"` lost to the
inference — and that value is the documented user-facing opt-out from
the Stage 3 default flip. Now gated on OMISSION. acc19b gains the
explicit-"current" case.
Finding 5: `window_drag` is a `HashMap<FrontendId, WindowDragState>`, so
a peer's mode-line press can no longer steal or clear another
frontend's in-flight gesture, and concurrent drags are legal. Cleared on
detach. acc30c gains the mode-line-press case.
Minor: `pmacs.window.buffer()` resolves both arms through the acting
frontend using the shared `lookup_window` / `selected_window` validators
rather than re-implementing them beside an ambient `active_buffer_id()`.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Font, licence, and the metrics half of Tier 3.
The bundled font is Latin Modern Math under the GUST Font License, added as
fonts/GUST-FONT-LICENSE.txt — deliberately a separate file from fonts/OFL.txt,
which covers JetBrains Mono only. GFL is LPPL-derived, not the SIL OFL; the
framing's F6 corrected that error and this is the discharge. At 733,736 bytes
the font is now the largest embedded asset in the repository.
ttf-parser is declared with default-features = false and only
"opentype-layout". Verified differentially: the ttf-parser feature set from
`cargo tree -e features` is byte-identical with and without this dependency
line, so the declaration widens nothing and forces no rebuild of the font
chain.
That check also corrected acceptance 17, which asserted `std` would be absent.
It is not — fontdb already enables it via `std = ["ttf-parser/std"]`, upstream
and independent of us. As written the criterion would have failed a correct
implementation, so it is now stated as the differential property that actually
matters.
MathConstants reads only what the Q#MS2 subset needs — axis height, script
scale percent, the two script shifts, and fraction rule thickness. Reading
more would be speculative: constants for deferred constructs have no consumer
to validate them, which is the Q#LX5 discipline applied to metrics. A font
with no MATH table is a typed error rather than plausible-looking zeros, so a
bundled-font regression cannot be silent (Q#MS7).
math_italic implements TeX's convention as the framing's table states it:
ASCII letters and lowercase Greek italic, uppercase Greek upright, digits and
operators unchanged, with U+210E for `h` because the 1D4xx run has a hole
there and arithmetic would land on a reserved codepoint.
Five tests, all against the real embedded bytes rather than fixtures, since B5
is the bet that would sink Tier 3 if false. One goes beyond the framing: every
italic mapping must resolve to a glyph the bundled font actually has, because
a mapping that produced tofu would be worse than the roman fallback it
replaces.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
pmacs-gpu/src/math_parse.rs — the Q#MS2 subset: characters, groups,
sub/superscripts in either order, and \frac, plus the Greek seed map.
Everything outside the subset is a typed error, which Q#MS8 turns into
"render the raw source".
The AST is semantic, not presentational: \alpha resolves to 'α' here, but the
math-italic mapping stays in layout, where a codepoint becomes a glyph.
Baking italics into the AST would make the tree disagree with the source and
would have to be unpicked by any later non-italic style context.
Two bugs the tests caught before they could reach layout, both from skipping
whitespace in the wrong place. `parse_atom` consumed it, but the ^/_ dispatch
happens in `parse_sequence` BEFORE atoms are read — so `x ^ 2` parsed the
caret as a literal character, and an all-whitespace span produced an empty
group instead of the Empty error. Whitespace is now skipped at the dispatch
point, which fixes both at one seam.
Interior `$` is rejected explicitly so `$$x$$` degrades through the error path
(acceptance 15) rather than half-rendering.
Clippy reports MathNode as dead code, which is correct and expected: the
parser has no consumer until Tier 3 layout lands. That is exactly the
condition Q#LX5 refused to ship, now enforced mechanically. It is not
suppressed; it clears when layout arrives in this same branch.
11 unit tests, no GPU or font required.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Four findings, one of them a real gap.
R2-2 is the substantive one. Rev 2's math-italic fix stopped at ASCII, which
reintroduced the same defect one symbol over: the parser resolves \alpha to
U+03B1, so $\alpha x$ would have drawn an upright alpha beside an italic x —
mixed styles inside a single expression, with the Greek seed map being the
slice's second flagship case after x^2. The mapping now follows TeX's actual
convention and states it as a table: ASCII letters and lowercase Greek in math
italic (U+1D434 and U+1D6FC runs, with the U+210E hole for h), uppercase Greek
deliberately upright, digits and operators unchanged. Acceptance 13 gains
$\alpha$ and $\Gamma$ cases so the uniformity is pinned rather than assumed.
R2-1 was self-inflicted: rev 2's Q#MS3 rewrite added a corrected copy of the
tree-sitter paragraph without deleting the original, leaving the stale rev-1
text immediately below the new one. Removed.
R2-4 sharpens Q#MS11 from "covers" to "intersects". For selections the
distinction is vacuous, but a search match can genuinely half-overlap a span —
searching "2$ af" in "before $x^2$ after" matches from inside the span to
outside it — and the whole-rectangle rule must apply there too, since the box
has no interior byte map to place a partial wash against.
R2-3 moves the follow-up section after prior art; the sections now run 0
through 9 in order.
Acceptance 12 also absorbs a round-2 non-finding: the reviewer's arithmetic
puts \frac{a}{b} near 0.85 and suggests \frac{x^2}{y} also clears the 0.6
floor, so the fallback case has to be computed against the real font rather
than guessed — otherwise the fallback half of the test surprise-passes by
rendering. The criterion now says so and asks for the computed scale to be
pinned.
Framing only; no implementation, no runtime code.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Nine findings, two of them decisions the implementation could not have
proceeded without, one a compliance error. Two rev-1 claims were wrong rather
than imprecise and are withdrawn where they occur.
F1, the fraction height budget, was hidden inside "a box that would exceed the
line" without saying width or height. Lines cannot grow — metrics are uniform
at 16 px over a 22 px line — and a textstyle fraction does not fit. Q#MS10 now
picks fit-to-line scaling with a 0.6x floor and source fallback below it,
records why overdraw and line growth were rejected, and states the honest
consequence: v0 shrinks nested math uniformly where TeX would use style
levels. B6 makes the floor falsifiable against the flagship case.
F3 was a genuine contradiction: acceptance 10 demanded source-width boxes
while Q#MS4 implied layout-chosen width, and that gap IS the caret-toggle
reflow question. Resolved for layout-chosen, because pinning to source width
leaves every fraction sitting in a large permanent gap, whereas reflow is
transient and tied to a deliberate caret move. Acceptance 10 is rewritten to
bound it rather than forbid it.
F2 withdrew the zero-glyph strut: a RichChunk's only width is its text, so the
mechanism is the SourceTab precedent — projected spacer text, width quantized
to whole advances. B1 is restated against the mechanism that exists.
F5 would have bitten real prose immediately: the rev-1 rule renders "prices
are $5 and $6 today" as math. Pandoc currency guards adopted, spans confined
to one line, both in acceptance.
F6 is factual and a licence matter. Latin Modern Math is the GUST Font
License, not OFL, and is 733,736 bytes rather than ~200 KB — verified against
a local TeX Live copy. The bundled licence file must say GFL, and this becomes
the largest embedded asset in the repo. The parent framing carries the same
error; correcting it is recorded as a follow-up rather than done here, since
the parent is already merged.
F4 names the new shaping-invalidation edge as the #120 trap class, including
that suppression must read the effective caret or flap during optimistic
typing. F7 puts math italic in scope, without which the flagship case renders
a roman x. F8 clarifies that glyph IDs are still resolved internally and that
drawing must pin Attrs to the math family. F9 closes the smaller gaps: the
detection hook wording, an overlay rule for selections and washes, degradation
of deferred display syntax, and folding Char/Symbol.
Framing only; no implementation, no runtime code.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
`TerminalViewStatus.scroll_offset` is the retained rows between the
VIEWPORT and the live tail, so it necessarily tracks viewport height: an
assertion that it survives a panel height change unchanged is either
vacuous or wrong, and it went red once under a loaded sweep for exactly
that reason. Q#BP7's invariant is that the ANCHOR is frozen, so acc32
and acc33 now compare the first visible row's text across the change,
and additionally pin the follow behavior that distinguishes them: a
shrink never re-arms follow, growth reaching the tail does, and growth
with a frozen selection does not.
Both also wait for the child's last line before sampling, so neither
races further output.
Also records the round in docs/active-work.md.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The obvious next unit was Tier 2 (parser) plus Tier 3 (layout). This framing
rejects that for the arc's own reason: the substrate lane's Q#LX5 refused to
land the parser ahead of layout because MathNode's shape is only validated by
a consumer, and that argument does not stop at Tier 2. MathBox is only
validated once a renderer consumes it, so Tiers 2+3 with no Tier 4 reproduce
the same objection one layer up.
The unit is therefore thin and vertical: characters, sub/superscript and
fractions, detected, parsed, laid out against a real OpenType MATH table, and
drawn — with breadth deferred to follow-on work against an API that has been
exercised rather than only designed.
The scout also found a load-bearing error in the parent's integration table.
It places math_parse.rs and math_layout.rs in the core crate, which cannot
work: pmacs-gpu depends only on pmacs-protocol and never on pmacs, so a
core-crate parser is unreachable from the frontend that renders it, and
ttf-parser reaches only pmacs-gpu. Both also contradict the parent's own
"instance never learns a pixel" contract and its "pure frontend
responsibility in v0" protocol section. Q#MS1 moves both modules into
pmacs-gpu.
Two decisions buy the slice most of its simplicity. Rendering is suppressed
whenever the caret is inside a span, which makes editing-shows-source and
moving-away-renders mutually exclusive states and removes the
caret-inside-math problem rather than approximating it. And layout positions
characters rather than glyph IDs, which is sufficient for this subset and
explicitly does not generalise: glyph variants are what stretchy fences and
big operators need, and both are deferred.
Acceptance puts every claim that something reaches the screen on real pixels
through the existing headless render seam, because a layout engine wired to
nothing would pass any weaker test.
Framing only; no implementation, no runtime code.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
PR #155 review round 1.
Finding 1 (must fix): `try_split_active` had no production caller —
`pmacs.window.split_horizontal` / `split_vertical`, and therefore
`C-x 2` / `C-x 3`, still went through plain `split_active`. Splitting a
focused panel made the root wrapper's final child a split rather than
`Leaf(side)`, which both `Layout::compute`'s fixed pass and
`document_subtree` key on: the panel band reverted to 1:1 weight
division and an ordinary window ended up living inside it. Both bindings
now route through the guard, and acc26 asserts through the real Lua
path — a direct core call passes with the guard unwired, which is how it
survived the first round.
Finding 2: the armed-drag early return now checks the arming frontend,
so one frontend's in-flight gesture cannot cancel or swallow another's
mouse events. New acc30c.
Finding 3: `paint_mode_line_graphemes`'s doc block was left heading
`paint_divider_segment`; moved back.
Finding 4: a recompile carries no `display`, so it took the raw switch
and duplicated a panel-placed `*compilation*` into the document window.
`start_run` now detects that the buffer already owns the panel slot.
`pmacs.window.buffer` gained an optional window argument so an adopter
can ask without selecting the panel first. New acc19b.
Stage-2 hazard pins the review asked for, both in `src/daemon.rs`:
a fresh attach while LOCAL is focused in a panel inherits LOCAL's
document buffer, and an initial-target bootstrap whose `after-load`
hook creates and selects a panel still reasserts into a document window.
Minor: dropped listview's dead `p.side`; documented `focus_window`'s
caller-validates contract; `jump_back` restores through `focus_window`
so the "every focus change" contract holds; `params` / `resize` default
to the acting frontend's selected window rather than the ambient one;
widened the flexible-division math to u64 intermediates.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Review finding, verified: rev 2's "no new build cost" was true only under
an unstated condition. ttf-parser re-exports `math` behind
`#[cfg(feature = "opentype-layout")]`, and it is compiled today only
because fontdb requests that feature — with `default-features = false` and
a set that is NOT ttf-parser's own default (fontdb's adds no-std-float and
omits std). A plain `ttf-parser = "0.25"` therefore unions std in and
forces a one-time rebuild of ttf-parser, fontdb, cosmic-text and glyphon.
Record the zero-rebuild spelling, `default-features = false, features =
["opentype-layout"]`, in Tier 3 §A and in the component table, so the
Tier 3 implementer declares it deliberately rather than tripping over it.
The C1 row points at the detail rather than repeating it.
Also note in the header that every anchor was re-checked at f07b75b. The
scout pin stays at ddaa80d because that is when the scouting happened;
#153 landed between the two and is test-only, moving no anchor cited here.
Framing only; no implementation, no runtime code.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Both earlier attempts in this branch were wrong, and the diagnostic they
added is what proved it: the macOS failure reported a stable
`rendered prefix: 6/15 ("VTERM_")` BEFORE and AFTER the CRLF change, with an
identical cursor, across two completely different child layouts. Identical
truncation under different layouts cannot be a layout problem.
The real mechanism is pinned by the repository's own unit test,
`cell::tests::diff_split_by_unchanged_cell_is_two_spans`: `cell::diff` splits
a run at any cell where `prev == next` and never transmits that cell. So when
a character of the marker already happens to sit at its destination, the host
receives the marker with that byte MISSING, not merely escaped around. The
constant 6 is the distance to the first such hole.
That makes escape-stripped matching unsound in kind rather than merely
insufficient: no matching strategy recovers a byte that was never sent. It is
removed, and `wait_for_output` is strict again. What remains asserted through
host bytes are protocol escapes pmacs writes directly — the OSC 52 clipboard
reply, the alternate-screen and bracketed-paste resets — which are not painted
cells and which the differ never touches.
Readiness now gates on a file the child publishes, the pattern the reliable
sibling test in this file already uses. That the child's output reaches the
SCREEN stays asserted in-process over `snapshot_text`, at the layer that can
actually see it; this test keeps what it uniquely owns, the host lifecycle.
`strip_ansi` and `longest_rendered_prefix` are kept as failure diagnostics
only, and now carry a case pinning the dropped-cell shape so the wrong remedy
is not reached for again. The new readiness wait reports startup breadcrumbs
on timeout; the plain helper reports only the missing path, which is the least
useful thing to know at exactly that moment.
Both the readiness gate and its timeout diagnostic were falsified by pointing
the child at a path the test does not watch.
Test-only; no runtime code.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Adds the lane to docs/active-work.md: branch, base, what Stage 1
implemented, the verification run, and the two known local-only test
caveats (the parallel-load GPU flake and compile_mode_acceptance's
single-thread requirement).
The durable handoff snapshot stays untouched until the PR merges, per
its own update protocol.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Revision 1 was written against protocol v18, before LaTeX Stage 1 (#144),
web grammars (#146), folding Stages 1-2 (#142/#149) and the GPU initial
target (#148) landed. Revision 2 changes no design decision; it corrects
the ground truth those merges invalidated and records the staging decision
the sibling substrate framing already took. A new section 0 lists every
correction so a reader who knows revision 1 can read it alone.
Two corrections change implementation choices rather than line numbers:
- The MATH-table dependency story was wrong in both directions. Revision 1
said a crate must be added and that "neither is in the tree today";
ttf-parser 0.25.1 already reaches pmacs-gpu non-optionally through
fontdb -> cosmic-text -> glyphon, the same fontdb the frontend already
calls. And the choice is not "one of ttf-parser or read-fonts": only
ttf-parser exposes the MATH table, supplying exactly the constants Tier
3 names. read-fonts 0.37.0 is present but has none, so selecting it
would be a dead end.
- Tier 2's staging was already decided elsewhere and this note did not say
so. The sibling framing's Q#LX5 puts the parser beside its Tier 3
consumer, never ahead of it, because MathNode's shape is only validated
by a layout consumer. That makes Tier 2 not independently shippable,
which is worth stating explicitly: it is pure and conflict-free, so
landing it alone while other lanes hold the render path is exactly the
tempting move Q#LX5 refused.
Tier 1 is materially de-risked: the LaTeX grammar already exposes
math_environment and math_delimiter, and the in-repo query overlay this
tier proposed already exists and captures both, so the mechanism is proven
rather than speculative. The guessed node name (math_expression) is
corrected to the grammar's own. Markdown still needs the overlay
treatment.
Tier 4 gains a contention note. Revision 1 described the GPU render path
as though math were its only claimant; folding Stage 3 and the
bottom-panel arc's Stage 2 now converge on it, so whichever lands second
re-scouts against the first.
Framing only; no implementation, no runtime code.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
- `listview.open`, `compile.run`, and `pmacs.terminal.open` all take the
same strict `display = "current" | "panel"`, validated before any
buffer, session, process, or wrapper exists. Omission keeps today's
behavior; Stage 3 flips the default.
- `listview.quit` / `compile.quit` delegate to `window.quit` only when
the buffer really is in a side window, so the presentation is deleted
or restored instead of leaving a source buffer stranded in the slot.
- LSP `visit_location`, LSP go-to-definition, and compile `visit_error`
route through `display_file`, so a visit from a panel lands in the
document target and fires its hook with that window active.
- `window.quit`'s Delete arm focuses the revalidated remembered origin.
- Capability fallback discards an accompanying `height` rather than
rejecting the call.
- `window.min-height` clamps a below-floor value on read instead of
refusing the write.
- `tests/bottom_panel_stage1_acceptance.rs`: 42 tests over the framing's
Stage 1 criteria, including the two production `Layout::compute`
callers, the recursive minima, hide/reappear, the final-focus matrix,
quit chains at the depth cap, per-frontend jump origins, the divider,
and a real-PTY pin of Bet B1.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The diagnostic added in the previous commit answered the question on its
first macOS run:
rendered prefix: 6/15 bytes ("VTERM_") — child text rendered only partially
So the child wrote and the host received part of the marker, but stripping
escapes did not rejoin the rest: other repainted cells sit between the two
pieces, not just cursor moves.
Six characters is exactly what fits before the right margin of this
session's 40-column child. The probe writes bare `\n`, and the supervisor's
PTY trampoline runs `stty raw`, which clears OPOST — so a lone `\n` moves
down without returning to column 1 and every line staircases five columns
right. After twenty lines the marker starts in the right margin, wraps
mid-word, and reaches the host as two pieces that no contiguous match can
join.
That also explains the intermittency: the wrap column depends on whether
pmacs has already resized the PTY from the requested 40 columns to the
window width, which races the child's first writes.
Write explicit carriage returns so every line returns to column 1 and the
markers start there. The fixture was wrong about its own line discipline;
the emulator was behaving correctly throughout.
The escape-stripped matching and the rendered-prefix diagnostic from the
previous commit are kept: they are what produced this answer, and they keep
the next such failure legible.
Test-only; no runtime code.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Stage 1 substrate for the bottom-panel arc (docs/bottom-panel-framing.md).
- `WindowParams` (side / fixed_rows / dedicated + implementation-owned
quit action and remembered document origin), `Side`, `QuitAction` with
a bounded replacement history, and the `MIN_WINDOW_OUTER_ROWS` floor.
- `Layout::compute(area, fixed)` allocates fixed rows before dividing the
remainder by weight; both production callers feed the same shared map,
including the peer-presence overlay pass that derives its own rect.
- `subtree_min_rows` / `interactive_min_rows`: the recursive minima, and
`boundary_below` for the shared drag / keyboard resize boundary rule.
- `FrontendView` gains `panel_capable`, `frame_geometry`, and the derived
`panel_hidden`, each spelled explicitly at every construction site.
- `EditorCore`: `primary_document_window`, the non-side target rule,
`display_buffer` + placement policy, `quit_window`, side-window removal
on `kill_buffer`, per-frontend jump entries with origin windows, and the
shared resolve/load-without-switch seam the initial-target bootstrap now
uses too.
- `EditorState`: the panel reconciliation transaction, geometry
declaration, the side-window `dispatch_idle_for` gate, divider paint,
and divider drag.
- `pmacs.window.display / display_file / quit / panel / params /
set_params / resize / display_target`, plus `builtin/runtime/window.lua`
with `window.panel-height`, `window.min-height`, and the resize commands.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The macOS `VTERM_ALT_READY` flake finally reported itself through #151's
breadcrumbs, identically in two runs: pmacs still running, `init.lua`
reached, `terminal.open` ok, and a settled screen whose tail is nothing but
`\x1b[22;42H` repeated 133 times.
That cursor is the evidence. A blank terminal parks at 1;1. Column 42 is
where the cursor lands after writing a 15-byte marker that ends at column
41 — so the child DID write and the emulator DID receive it. What failed
was the assertion: `wait_for_output` required the needle to appear as
contiguous bytes, but the TUI differ paints only changed cells and skips
ones that already match, so a run held contiguously on one screen row can
still reach the host as `PREF<cursor-move>IX`.
Match over escape-stripped bytes when the needle is plain text. This cannot
mask the failure that matters: text the child never wrote is absent from
the stripped stream too, so a genuinely silent child still fails. Needles
carrying their own escape (the OSC 52 clipboard reply) keep the strict path,
since stripping would consume the bytes under test.
The failure arm now also reports how much of the needle rendered, so the
next occurrence distinguishes "nothing reached the host" — a PTY/spawn
fault — from a partial render, instead of leaving a tail of pure escapes
that cannot tell them apart.
Both helpers are pinned directly, including that stripping rejoins a split
run without inventing absent text. `strip_ansi`'s first draft mishandled
`ESC ( B`, whose intermediate byte makes it three bytes rather than two;
its test caught that.
Test-only; no runtime code.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Adds the approved framing for the bottom-panel arc: a buffer displayed
in a fixed-height window pinned to the bottom of the frame, targeted by
policy rather than by stealing the selected window, plus the missing
display-buffer/window-parameter concept underneath it.
Revision 4 follows three review rounds, an integration review, and a
landed-state audit against GPU initial target (#148, protocol v20) and
folding Stage 2 (#149).
Amended before branching by the pre-implementation dependency
verification recorded in section 0.6:
- the folding dependency is cleared and re-verified against canonical
main at ddaa80d (nothing in flight, folding Stage 2 acceptance 48/48
green, every borrowed anchor reproducing, and folding's only window.rs
edit confined to one 22-line hunk that leaves the layout functions
pre-folding code);
- R5-B1: Layout::compute has TWO production callers, not one. The
second, the peer-presence overlay pass in src/overlay_paint.rs, builds
its own text-area rect from active_layout() and never routes through
window_placements, so the planned compute(area, fixed) signature change
would otherwise leave every peer cursor painted at its no-panel row.
Corrected in section 1.1 and Q#BP2, pinned by acceptance 1, and the
fixed map is now specified to come from one shared helper rather than
being assembled per call site.
Stage 1 (window placement + TUI side windows) changes no wire shape.
Move the lane from active-work.md into its Closed section, retire the
protocol v20 / main-hash references to LANDED form in agent-handoff.md, and
record both review-round lessons (failure-socket containment, upgrade-gated
replica publication) in the ops-lessons ledger.
18 of the last 60 CI runs failed (30%), including repeatedly on `main`.
Every failure is macOS-only and hits BOTH Lua flavors, so it is the
runner, not LuaJIT. Sampling 7 showed only two tests.
**m8_9 outline budget** — `outline_5_level_100_entry_renders_within_100ms`
is a wall-clock budget observed at 147ms and 149ms against 100ms on
GitHub's shared macOS runners, while Linux lands comfortably under. The
measurement and its printout now always run; only the ASSERTION is gated
on `!cfg!(target_os = "macos")`, exactly as
`composition_overhead_under_ten_percent` already is in `src/editor.rs`
for the same reason. A real regression still surfaces on Linux, on the
perf gates, and in the number printed to the log.
**vterm PTY smoke** — deliberately NOT a timeout bump. Instrumenting
locally showed the failing wait completes in 40ms against a 10s budget
(250x headroom), while the genuinely tight wait in the same test (2.86s
against 5s) never fails. Four hypotheses were eliminated with evidence:
- python3 cold start: the sibling test at :494 spawns the same
/usr/bin/python3 with a TIGHTER 5s budget and passes in the very
runs where the smoke fails ("3 passed; 1 failed");
- config resolution: XDG_CONFIG_HOME is read straight from the env
(src/config.rs:49), no platform branch;
- a stalled idle loop: the run loop polls on a 60Hz frame timeout and
ticks the supervisor every frame (src/editor.rs:2522), so child
output drains without input;
- a too-small budget: see the 250x headroom above.
The real defect this commit fixes is that NONE of those could be
distinguished from the CI log, which carried only "host output never
contained VTERM_ALT_READY" plus a tail of escape bytes. init.lua now
writes breadcrumbs (reached / terminal.open ok-or-error) and the timeout
reports whether pmacs is still alive plus each breadcrumb, so the next
occurrence names its own cause:
startup: pmacs still running; init.lua reached="1"; terminal.open="ok"
Verified non-vacuous: forcing the needle to never appear produces the
line above, which also proves io.open works in pmacs's Lua — otherwise
the breadcrumbs would silently read MISSING and mislead.
Test-only; no runtime code.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q5BkezMppbpCgGAYk2ftxV
Integrate folding Stage 2 and its landed-state documentation with the
protocol-v20 GPU initial-target branch. Preserve per-session fold projection
selection in the target bootstrap transaction and retain v19 compatibility
coverage after the later protocol bump.
Shut down bootstrap sockets on every dispatcher-side failure and reject
frontend events whose session state was never installed. This prevents a
lingering failed client from reaching absent render/size state.
Track target-side CRDT upgrades independently from load/create status so a
deduplicated hidden buffer is published to every existing grid replica. Add
real-daemon regressions for both failure containment and replica publication.
Post-merge housekeeping owed from #149, kept as its own docs-only PR per
the #138-#140 / #147 convention. No runtime code.
- active-work.md: base snapshot and the recovery check bump 47581f4 ->
6ed4fe9. The Stage 2 lane is retired and replaced by a folding lane that
records both stages as merged with nothing in flight, and states Stage 3
(GPU) has no branch and no framing yet — carrying its named obligations
(GPU collapse at TUI parity, caret/hit-test fold-awareness, the
BufferSnapshot fold-mirror clear, CRDT-origin unfold, and flipping
FrontendView.fold_projection true for semantic frontends) as that
framing's starting point. "Closed since the last snapshot" gains #149
and #147.
- agent-handoff.md §1: main @ 6ed4fe9, the "Last updated" line and section
date, the Stage 2 bullet flipped from IMPLEMENTED/PR-OPEN to LANDED with
the Stage 3 obligations attached, and the roadmap entry (remaining arcs
now read "6 folding Stage 3").
Both stages' design points are recorded as traps Stage 3 inherits rather
than as history: the merged-hidden-component unit, per-window/per-target
map instances, per-frontend projection, position-not-row normalization,
and the post-intercept edit site.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q5BkezMppbpCgGAYk2ftxV
PR #149 review round 5 flagged this PR as stale: it still claimed `main`
@ `c49a8c7`, Stage 2 framing "rev 2, under review; no implementation, no
PR", while `main` is `47581f4` and Stage 2 is implemented and open.
- Base snapshot and the recovery check bump `c49a8c7` -> `47581f4`.
- The folding Stage 2 lane becomes IMPLEMENTED / PR #149 OPEN: framing
rev 4 approved, the `VisibleLineMap` spine, the base-moved merge (and
why it was merged rather than rebased), and the five review rounds'
design-changing findings — each of which is a trap Stage 3 inherits.
- `main`'s ledger had gone unrefreshed through four merges, not one, so
"Closed since the last snapshot" now also records web grammars HTML +
CSS (#146) and LaTeX Stage 1 (#144) with its inline-math framing
(#145), including their durable lessons.
- agent-handoff §1: `main` @ `47581f4`, the "Last updated" line, the
Stage 2 substrate bullet, and the roadmap entry.
Rebased onto `47581f4` so it stays one documentation-only commit
directly off canonical main.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q5BkezMppbpCgGAYk2ftxV
Post-merge housekeeping owed from #142, kept as its own docs PR (no
runtime code).
- agent-handoff.md §1: bump main to c49a8c7, add the folding Stage 1
substrate bullet (store/View, structural source, C-c @ surface,
command-path unfold, FoldState production; no protocol bump), refresh
the "Last updated" line and the roadmap Arc 6 entry, and note Stage 2
is in framing on folding-tui (the visible-line-map reframe).
- active-work.md: retire the Stage 1 folding lane (PR #142 was OPEN),
add a "Closed since the last snapshot" entry for #142, open the
Stage 2 (grid/daemon collapse) framing lane on folding-tui, and
refresh the canonical base snapshot to c49a8c7.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q5BkezMppbpCgGAYk2ftxV
PR #149 review round 5, finding 1 — correct, and both round-4 bugs did
survive through the intercept path.
Round 4 moved the widening off the point and onto `edit_start_of(&op)`,
but ran it in `run_buffer_edit` BEFORE `run_managed_edit`. A managed
buffer intercept may legally rewrite `pos` / `start` / `end`
(`LuaInterceptView::intercept_edit`), so the requested op is not where
the edit lands:
- requested outside -> intercept relocates inside: the edit stayed
hidden;
- requested inside -> intercept relocates outside: an unrelated fold
opened.
The seam still covers BOTH paths — hooking only `run_managed_edit` would
let an interactive `bypass_intercept` edit escape, which is why the
framing put it on the common entry — but each path now keys on its own
effective site:
- `run_bypass_edit` applies its op verbatim, so `run_buffer_edit`
hooks it there;
- `run_managed_edit` hooks after the intercept chain settles and
before the apply, on the op the chain returned. A chain that raises
applies nothing, so it unfolds nothing.
The registry borrow is released at that point, and the helper reads only
`SharedCore` + the fold registry, so no borrow conflicts with phase 3.
Three new tests, all through real `M-x` with a real
`pmacs.buffer.add_intercept`: relocate-into-a-fold must unfold,
relocate-out-of-a-fold must not, and a rejected chain must not. Each
asserts the buffer text first, so the test fails loudly if the intercept
stops relocating rather than silently passing.
Suite 45 -> 48.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q5BkezMppbpCgGAYk2ftxV
Keep foreign BufferSnapshot publications out of existing semantic GPU
sessions while retaining grid-replica coherence. Treat dead peer writes as
peer-local failures, restore active-frontend cleanup, deterministic probe
readiness, GPU logging, shared tilde expansion, and accurate docs.
Add focused publication and cleanup coverage and record the two-window
Wayland/Vulkan smoke plus the complete post-review gate results.
All four findings were correct.
F1 (major) — the Lua widening keyed on the POINT, not the edit site.
Q#FD19 condition (iii) is `edit.range.start`; `run_buffer_edit` discarded
`op` and opened whatever fold contained `window.cursor`. Two wrong
behaviors followed: a command inserting elsewhere opened an unrelated
fold at the point, and — worse — a command editing INTO a fold from an
outside point left its edit hidden. `run_buffer_edit` now reads
`edit_start_of(&op)` before `op` is consumed and passes it down. The
`apply_active_edit` funnel stays point-keyed, which is correct and
deliberate: the six primitives, yank, and query-replace all place point
at the edit site first, and only the Lua path can diverge — now said so
in the doc comment.
The acceptance test encoded the wrong behavior (inserted at byte 0,
expected the cursor's fold to open). Replaced by two tests that pin both
directions, plus the framing's named **comment-toggle** case driven
through the real `M-;` on a file-backed Rust buffer rather than a
synthetic mutator.
F2 (major) — hidden-cursor motion normalized only the row. `move_up` /
`move_down` clamped `coord.row` but derived `goal` from the hidden
line's raw column, and returned early at a buffer boundary without
normalizing at all, so `Up` inside a fold headed on line 0 left the
logical cursor hidden. Added `normalize_cursor_to_visible`, which
projects the whole POSITION through `visible_position` as a real
mutation before any step is computed (and drops the sticky goal column,
since the jump is discontinuous). Paging shares it — the same latent
bug. The old test used equal-width lines and asserted only the line, so
it could not tell the two apart; the new fixture is deliberately ragged
and asserts the resulting BYTE in both directions, plus the boundary
case.
F3 (moderate) — `set_view_top` bypassed the fold clamp. Q#FD12/Q#FD18
say `view_top` is always set through `clamp_view_top`, naming the
`set_view_top` contract specifically. Rendering repaired it at the next
frame, but until then `view_top()` handed out a hidden line and
command/event reckoning could start from a non-visible origin. The
clamp now lives in the setter — the contract's home, what `saveplace`
and `pmacs.editor.set_view_top` call. New test reads `view_top()`
directly after the setter, before any paint.
F4 (moderate) — the suite claimed items 1–14 but omitted approved
clauses. Added: peer SELECTION endpoint projection with hidden interiors
dropped; the crossing-fold repeat for the peer cursor AND a peer
selection endpoint; a style/syntax span straddling a fold (alignment on
the shifted row); the completion popup anchored below a fold; and peer
presence in the split-buffer case using the RECIPIENT window's map.
Suite: 35 -> 45 tests. Eight new bite cases, each falsifying exactly the
line that implements its claim.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q5BkezMppbpCgGAYk2ftxV
`main` moved from c49a8c7 (folding Stage 1, #142) to 47581f4 (web
grammars, #146) while Stage 2 was in framing and implementation. The
text merge is clean, but it is NOT semantically clean: #146 added three
new `Viewport { .. }` literals to `src/highlight.rs`'s unit tests, and
Stage 2 gives `Viewport` a `folds` field. Merged alone, `cargo test
--lib` fails to compile — so the carry-over is resolved here rather
than left for CI to discover.
Merged (not rebased) so the four framing revisions the review rounds
cite by SHA (59410c0, e221f13, 8160d66, 4222ffa) stay reachable.
Full gate suite re-run on the merged tree.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q5BkezMppbpCgGAYk2ftxV
`cargo test --lib --features crdt` compiles two `#[cfg(feature =
"crdt")]` daemon tests that call `build_fresh_frontend_view`. Both model
a semantic session, so they pass `false` — the Q#FD21 value a
`semantic_render` attach gets until Stage 3. Bet B8 in the framing is
exactly this: the projection is chosen explicitly at every construction
site, never inferred, so a new call site cannot silently default.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q5BkezMppbpCgGAYk2ftxV
Levi Neuwirth