JSON folding in the console was bypassed in any filtered view, including
when only a channel toggle (daemon, RPC or app) was active. A collapsed
block silently expanded the moment a channel was toggled.
- Folding is now kept active unless the view is text-filtered or
errors-only, which are flat so that every match stays reachable.
Channel toggles keep folding, since JSON blocks are atomic per channel.
- The fold triangles and the collapsed " ... }" summary are drawn only
in the fold-aware view, so the glyph and the rendered block can never
disagree.
User-visible: JSON results stay foldable while channel filters are
applied.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A long wrapped console line could extend far past the viewport, yet
every sub-row was still measured and drawn.
- Sub-rows that fall entirely outside the clip rect are now skipped. The
per-row ordinal still advances for a culled row, so scanline parity is
unchanged.
No visible change; only off-screen rows are skipped.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A large RPC result was fully split into lines and copied before being
handed to the console, even though only the last few lines could survive
the model's line cap.
- The formatter now takes a line cap and keeps only the last N lines,
located by a backward newline scan, so a huge result is never fully
split or copied. JSON classification still keys off the first
character of the whole result.
- It works over a string view and moves line text through, and fold
spans are computed over the formatted lines directly instead of a
separate vector of copies.
- getline semantics are preserved: a trailing newline adds no empty
line and an empty result yields no lines.
Tests: getline parity for empty, trailing-newline and blank-line inputs,
the backward tail cap, and JSON classification and fold spans over a
capped result.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The console's pending, not-yet-drained queue was unbounded, so a hidden
Console page that kept receiving verbose daemon-log chunks could grow
without limit.
- Pending lines are now bounded by both a line count and a byte budget.
The oldest pending lines are dropped once either is exceeded, while
the newest line is always kept so one huge line is never lost.
- A formatted RPC result is ingested as one block under a single lock,
so lines from other producers can never interleave inside it.
The front-sequence accounting is also corrected: a pending line dropped
by the cap never entered the visible buffer, so it no longer advances
the front sequence. Only a front eviction or a clear does.
Tests: the line and byte caps keep the newest lines, a single oversized
line survives, a batch appends contiguously, and the front sequence
advances only on a front eviction or clear.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Console text filtering and match highlighting lowercased a full copy of
every line, and of every drawn segment, on each pass.
- A new alloc-free ASCII case-insensitive substring search folds both
sides on the fly, so neither the filter test nor the highlight scan
allocates a lowercased copy per line any more.
- Both call sites use it; the case-insensitive match behaviour is
unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
With any console filter on, each batch of new output rescanned all N
lines to rebuild the visible set. A text filter lowercase-copied the
needle and every line to test it: O(total characters) and 2N
allocations per batch.
When the filter is unchanged and the model only appended or evicted
lines since the last pass, the visible indices drop the evicted lines
and shift down, and only the new lines are tested. The result equals
a full rescan, since line text and channel never change and the
filtered view has no folds. A filter change, Clear, or a model window
that doesn't follow on from the last one still rescans.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Each batch of new output re-wrapped all N visible lines glyph by glyph,
allocating a fresh segment vector per line; a line evicted at the
10,000-line cap shifted every index, so nothing could be kept. And on
every frame, idle or not, seeding the scanline shading summed the wrap
rows of every line above the view.
- The model numbers its lines and the view keeps one wrap per line,
aligned by that number, so an eviction drops a prefix instead of
forcing a rebuild. New output wraps only the new lines; the layout is
then a prefix sum over the N visible lines, with no measuring and no
per-line allocation.
- The wraps are kept while the wrap width, line height, font size, font
(by ImFont::FontId) and rasterizer density are exactly unchanged. A
line-spacing change only re-sums heights; filters and fold toggles
reuse the cached wraps. A resize drag still re-wraps every visible
line per frame, as before, but reuses their storage.
- The scanline seed is read from the prefix sum: O(1) per frame.
Heights, offsets and segments are bit-identical to the old rebuild. A
font reload at an unchanged size (the colour-emoji toggle) now re-wraps
the console at once; before, it waited for the next line of output.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
While theme effects were on (the default) the main loop drew every
frame it could. Where vsync is off (offscreen, some drivers, remote
sessions) an idle wallet spun at ~3,000 fps, 88% of a core; with vsync
it still drew every refresh. Frames are now paced by what needs them:
- full rate while there is input, a held key, a window appearing or
settling, a smooth scroll or the UI sweep;
- ambient (30 fps focused, 10 fps unfocused) for theme effects, the
acrylic backdrop, avatar animations and notifications;
- progress (10 fps focused, 4 fps unfocused) for UI that changes
without input. The updater and bootstrap downloads, console
commands, lite-wallet operations, seed migration, the loading
overlay and the lock-screen countdowns ask for it with
material::RequestProgressFrame();
- otherwise idle, as before: wait for events, with a housekeeping wake
every 2 s (5 s unfocused).
A software frame cap (0.9x the display refresh) bounds the loop where
vsync is off, and hidden, occluded or minimized windows get no ambient
frames. Reduce Motion freezes the theme-effect clock
(ThemeEffects::setMotionFrozen), so it needs no ambient frames either.
The displayed-balance roll goes through one clamped StepDisplayLerp
(balance_lerp.h). An idle-length frame pushed dt*speed past 1, so the
roll overshot, and under Reduce Motion's speed of 999 it diverged to
inf/NaN. The classic layout had its own copy of the old code.
Idle CPU, Lite build, offscreen (no vsync): 88% -> 5% of a core with
effects on, and 91% -> 3% with effects saved off (with the previous
commit).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Immediate-mode render functions re-run every frame; these rebuilt O(N)
state each time even when nothing changed. From a 6-lens perf audit, each
finding verified on a hot-path basis:
- Console: ConsoleModel gains revision(); the full-model filter scan and the
glyph-by-glyph text-layout pass (BuildConsoleLayout) rebuild only when the
model / filter / wrap-width / zoom change — previously it re-shaped up to
10,000 lines every frame even when idle/scrolled. clear() force-invalidates
the memo mid-render (no OOB on the just-emptied visible set).
- Transactions: the summary-card totals memoize behind the tab's existing
FNV-1a fingerprint (also folds away a now-duplicate O(N) display-key pass).
- Send / Receive recent lists: early-exit the prefix scan (state.transactions
is kept newest-first) instead of filtering the whole tx history every frame.
- network_refresh_service: O(new x total) txid find-and-replace -> hash map.
- Sidebar unconfirmed-tx badge cached on last_tx_update + tx count; the
daemon-memory probe (/proc scan on Linux, popen on macOS) throttled to ~1.5s.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Bundles the session's UI work (the touched files carry several of these
changes together, so they are committed as one coherent UI batch):
- FAQ: new RenderFaqDialog + data-driven faq_content, opened from a
status-bar "?" (and the Windows title bar), styled like the Wallets
modal with search, Wallet/Daemon tabs, and smooth scroll.
- DPI/font-scale audit: multiply hand-drawn absolute geometry by
Layout::dpiScale() across ~30 files so nothing renders native-size at
HiDPI / font_scale 1.5 (verified with a full sweep at 1.5x).
- Mining: chart now fills the horizontal space; thread stepper +/- buttons
match the input-box height; move the stratum-host toggle into
Node & Security (v1.3.0+).
- Settings: fix the auto-shield status text overlapping the grid.
- Sidebar: drop the peer-count badge on the Network button.
- i18n: wrap 193 hardcoded literals with TR() (keys/translations added in
the preceding i18n commit), so the security/PIN/lock flow, seed-backup
wizard, and witness-rebuild dialog localize.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Fixes the cut-off/clipping bugs from the layout audit (all visible at the default
1280/1024 window sizes):
- Receive "Recent Received" rows: the amount collided with the relative-time
("+15.7500 DRGX14 days ago") and the type label touched the address at narrow
widths. Use the shared short time format (formatTimeAgoShort, matching Overview),
chain the amount's right edge off the measured time width, and start the address
after the measured type-label width — so neither pair can collide.
- Daemon & xmrig updater verify-note: drawn unwrapped and clipped at the card's right
edge; wrap it (PushTextWrapPos) within the already-reserved height.
- Request Payment: the three footer buttons shared one fixed width (clipping "Copy
Full Address"); size each to its own label. The Payment URI overflowed a plain
field; render it in a bordered read-only box (bounded, un-chunked).
- Console: the filter input shrank below its own placeholder (gone entirely at 1024);
give it a min width >= the placeholder and drop the "N lines" count when the row
can't fit both.
- Address-label "Choose Icon" grid: had NoScrollbar hiding most of the catalog with
no cue; give it a real scrollbar.
- Overview "Recent Transactions": drop the 4th row at 1024 (it clipped off-screen) by
capping to rows that fully fit the reserved height.
- Sidebar: reserve the unread-badge width in the nav-label centering so History/Chat
labels no longer collide with their badge.
Verified at 1024 and 1280 across full-node + Lite + Windows (ctest green) and an
adversarial diff review (clean). Skipped the legacy settings_window overlay footer
(dead code / removal candidate).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The tail of the DPI/font-scale/responsiveness audit — ~26 remaining findings.
Container / recent-list (Theme-1 leftovers):
- Send: drop NoScrollbar|NoScrollWithMouse on ##SendFormScroll so Recent Sends is
reachable at font_scale 1.5 (parity with receive).
- Receive: cap the QR/form card via std::min(mainCardTargetH, availH - recentReserve)
so RECENT RECEIVED stays on-screen (identity at 1.0x).
- Wallets dialog: size the capped-mode list to whole rows so it no longer clips a
partial row / crowds "Create a new wallet".
Narrow-width (1024px) reflow:
- Console toolbar reserves space for ALL trailing controls (both icon toggles + zoom
buttons) so the +/- zoom no longer runs off-window.
- History sort combo sized to its measured widest localized label ("Newest first").
- Settings Theme/Layout/Language row: scale the wide→stacked breakpoint by dpiScale so
it drops to full-width stacked combos at 1.5x (Consolidated Card no longer clips).
- Recent-tx type label: derive the address column X from the measured label width so it
can't collide at narrow widths.
- Mining Recent Pool Payouts: floor the panel height to fit the empty-state caption.
Cosmetic ×dpiScale() on absolute geometry (no-ops at 1.0x): mining SOLO|POOL toggle &
idle combos, market pair-chips, password/PIN strength bars, receive/send currency
toggles, explorer search bar/rows/rounding, About-card logo, chat empty-state wrap,
recent-list address/time offsets, address-toolbar & two-row action buttons, console
line-gap/status-dot/pane rounding.
Verified at font_scale 1.5 and at 1024px across full-node + Lite + Windows (ctest
green) and an adversarial diff review (one over-reserve regression fixed).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Security audit remediation (15 confirmed findings from the codebase audit):
- H-02: scrub+delete the decrypt-flow plaintext key export on ALL exit paths
(RAII guard) and purge stale obsidiandecryptexport* files at startup.
- M-01/L-03/L-04/L-05/L-07: sodium_memzero the Set-PIN and encrypt-PIN worker
passphrase/PIN copies, the RPC Basic-auth string (auth_), the exported/imported
key buffers (App::wipeSecrets, called from ~App and before main's _Exit), and
the first-run wizard "Skip" buffers.
- M-03/M-04/M-05/L-06: return locked COPIES from XmrigManager/EmbeddedDaemon
getters (dedicated error_mutex_; DaemonController::lastError now by value),
route xmrig last_error_ writes through a locked setter, and wrap
shutdown_status_/wizard_stop_status_ in a locking GuardedStatus
(wizard_stopping_external_ -> std::atomic).
- M-02: persist after a console send/shield/import in the lite backend.
- L-01: require the confirm click for z_shieldcoinbase/z_mergetoaddress.
- L-02: quote/escape each Windows daemon argv per the MSDN CommandLineToArgvW rules.
- L-08: pin json/tomlplusplus/libwebp FetchContent to immutable commit SHAs.
- I-01: extract updater archives from the already-verified in-memory buffer
(no disk re-read TOCTOU).
Feature: warn once (full-node) when the active wallet loads empty while a sibling
wallet file in the datadir holds keys. A funded salvage wallet.<ts>.bak routes to
the recovery/Restore flow; a funded sibling .dat routes to the wallet manager.
Per-wallet-file dismissal; gated on synced + address-list-loaded to avoid false
positives on warm reconnect / spent-down wallets.
UX fixes:
- send: show the TOTAL balance (with a spendable "available" note) in the source
dropdown and keep pending-change addresses visible.
- chat: insert emoji at the cursor position; restrict new-chat recipients to
shielded (z) addresses.
- console: optional auto-focus of the command input on tab open (off by default).
- shutdown: when "stop external daemon" is on, keep the shutdown screen up until
the external node actually exits, showing live status.
Adversarially reviewed; verified across full-node, lite, and Windows builds; tests pass.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The RPC console echoed and stored typed commands verbatim, so `walletpassphrase
<secret>`, `z_importkey <key>`, `encryptwallet <pass>` etc. left the secret in the
visible log AND the 100-entry recall history (copyable). Adds a pure, unit-testable
RedactConsoleCommand()/ConsoleCommandCarriesSecret() (allowlist of 13 secret-bearing
first-tokens) in console_tab_helpers; submitConsoleCommand() now echoes and stores
`> walletpassphrase ****` while still executing the real command unredacted. Bare
secret commands and non-secret commands pass through unchanged.
Output-secret commands (dumpprivkey / z_exportkey / z_exportmnemonic) — whose secret is
in the RESULT — are a separate redaction concern, tracked as a follow-up.
First fix in the wallet-hardening P0-A cluster (see docs/wallet-hardening.md). New
testConsoleSecretRedaction (11 assertions); ctest 1/1.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Give the lite console the parity analog of the full-node RPC command
reference: the same searchable two-pane modal (browse/search, detail
pane, examples, Insert / Insert & run, destructive confirm), driven by
the lite backend's own command set instead of daemon RPC.
- ConsoleCommandExecutor::commandReference() returns the category table
(full node = consoleCommandCategories(); lite = new
liteConsoleCommandCategories() -- 25 backend verbs in 5 categories).
The shared renderCommandsPopup reads the table from the executor.
- The Commands button now shows for both variants (gated on
commandReference()!=nullptr); title/tooltip/arg-quoting branch on
hasRpcReference() (clarified to mean "speaks JSON-RPC / full node").
- Lite args are bare tokens (the backend takes one unsplit arg string
and does not strip quotes), so the param-builder's JSON string
auto-quoting is disabled for lite -- Insert & run emits runnable bare
commands. send uses the JSON-array form (its positional form is
unreachable via the single-arg transport); new uses zs/R; import
takes just the key (the backend hardcodes birthday=0).
Adds 7 i18n keys across all 8 languages (additive). Full-node behavior
is unchanged. Adversarially reviewed (data vs the backend registry,
plumbing/regression, i18n) with all findings fixed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The lite console tab lagged the full-node one on i18n: its toolbar status,
status lines, help, and error/warning strings were hard-coded English or
worded for a full node (a "daemon" the lite wallet doesn't have).
- toolbarStatus()/statusLines(): route through TR() (reusing the existing
lite_net_* keys where they already map)
- printHelp(): enumerate the 25 pass-through backend verbs for discoverability,
since the C++ tab intercepts `help` before the backend's own HelpCommand runs
- gate the destructive 'stop' confirmation to the full node (hasRpcReference);
lite has no node, so it lets 'stop' fall through to the backend instead of
showing a phantom "shut down the node" warning behind a dead gate
- word the not-connected error per variant (daemon vs "no wallet open")
- TR() the "(no output)" command-result fallback
Adds 7 i18n keys across all 8 languages (additive); rebuilds the CJK subset
for the one new glyph (U+C5D4 for the Korean "backend"). Build + ctest +
source-hygiene green; changes adversarially verified (no logic/i18n defects).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The status indicator's left inset (+2) and dot-to-text reservation (+6) were raw pixels while the dot radius is DPI-scaled, so the dot shifted a couple native px at HiDPI. Multiply both by Layout::dpiScale() per the project's hand-drawn-geometry rule.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The accent + text-colour toggle buttons pushed/popped ImGuiCol_Text guarded by a flag the TactileButton flips in between, so a click left the colour stack unbalanced for that frame — and ImGui's error recovery drew a red rectangle around the console window (imgui.cpp:11727). Capture the flag into a local before the button and guard both the push and the pop with it.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The inner scroll child (##PeersList / ConsoleOutput) uses NoScrollWithMouse + ApplySmoothScroll; ImGui forwards its wheel up to the outer scroll container, so the wheel scrolled both. Add NoScrollWithMouse to the outer containers (safe: both fill the remaining height and never overflow), matching the contacts/explorer fix.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
render() computes visible_indices_ (indices into model_) once per frame at the
top, before the toolbar. The toolbar's Clear button called clear() → model_.clear(),
emptying model_ mid-frame (the "cleared" marker is only queued, drained next
frame). renderOutput() then indexed model_[visible_indices_[vi]] with the stale
indices → out-of-bounds → crash.
clear() now also drops visible_indices_ and the selection (both hold line indices
into model_), so this frame's renderOutput iterates zero lines and computeVisibleLines
rebuilds them next frame. The right-click "Clear console" menu item now routes
through clear() instead of a bare model_.clear() so it's covered too.
Adversarially verified: no other same-frame path indexes the emptied model
(fold-toggle, ingest, and selectAll are bounds-guarded).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Two issues from the audit:
- Cursor: hovering the modal showed the console's text-select cursor. The
output's mouse-interaction guard only suppressed itself under ImGui popups,
but this modal is a BeginOverlayDialog (not a popup), so the console kept
setting the text cursor underneath. Suppress output interaction while the
reference modal is open (!show_commands_popup_).
- Design: the two panes used hard 1px child borders and floated unpolished.
Restyle to match the portfolio/wallets modals — each pane now sits on a soft
DrawGlassPanel surface (borderless child + inner padding), category headers
are subtle accent labels (no heavy filled bars), and the selected/hovered row
gets a rounded Material fill instead of the sharp Selectable highlight. The
glass panes are theme-adaptive (light on light skins), so the command-name
blue and the mono example green are floored to readable contrast (FloorLight)
and the detail heading uses theme Primary().
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The builder's "omit trailing optionals" loop stopped at the FIRST empty
optional, which discarded any later field the user had filled — e.g. setban
with bantime empty but absolute=true built `setban "ip" "add"`, dropping the
typed `true`, and (worse) left Insert & run enabled so it ran a command that
ignored the input.
Include fields up to the last one that is filled or required; only genuinely
trailing empty optionals are dropped. A gap (an empty field before a later
filled one) keeps its placeholder and marks the command incomplete, so Insert
& run disables and nothing typed is lost. Found by an adversarial pass over all
82 param templates.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The detail pane's Parameters section is now an editable form: each parameter is
a type-hinted input field (string / number / json, optional marked), and a live
"Builds" preview shows the command being assembled. String values are
auto-quoted; a required field left empty keeps its placeholder and disables
"Insert & run"; trailing empty optionals are omitted.
Insert and Insert & run now use the assembled command (not just the template),
so Insert & run works for parameterized commands once the required fields are
filled — destructive commands still confirm first. Param fields reset when the
selected command changes or the modal opens. Keyboard nav (Up/Down/Enter) is
scoped to the search box so typing in a field doesn't move the selection.
A best-effort parseParamSpecs turns the human-readable param templates into
fields; JSON-array params (z_sendmany, createrawtransaction) surface as a single
"json" field the user pastes into. Adds the console_ref_builds label + 8
translations.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace the flat, terse command-reference table with a guided two-pane
explorer built for users who don't know the RPC commands:
- Search by name OR task: keyword synonyms let "balance" find getbalance,
"send money" find sendtoaddress; results are ranked (name-prefix > name >
keyword > desc).
- Two panes (mirrors the portfolio editor): a category-grouped master list
(mono names, a warning dot on consequential commands) + a rich detail pane
with a plain-language explanation, a per-parameter type breakdown (string /
number / json, optional dimmed), and a concrete example.
- Insert into console (fills the input with a template to review + run) and,
for no-parameter commands, Insert & run — with a confirm for destructive
ones. Deferred via pending_submit_ so the modal needs no executor.
- Keyboard: type to filter, Up/Down to move, Enter to insert; auto-focus and
auto-select the top result.
Data model: ConsoleCommandEntry gains details/example/keywords/destructive
(C++17 defaults, so the ~60 un-enriched entries are untouched); ~20 common
commands enriched and 18 consequential ones flagged. Command docs stay English
(technical); the UI chrome + 7 category names are translated into all 8
languages (CJK subset rebuilt, +3 glyphs).
Verified: two-pane renders on dark + light skins and at font_scale 1.5; detail
pane shows explanation/params/example; danger dots + safety badge on
destructive commands.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Clicking a command now actually closes the modal. It called
ImGui::CloseCurrentPopup(), a no-op here (BeginOverlayDialog is Begin/
BeginChild, not an ImGui popup), so the picker stayed open covering the input
and the insert looked like it did nothing. Set show_commands_popup_=false.
- Search filter is now a member cleared when the modal opens, so it isn't stale
on reopen after an outside-click / command-click dismiss (was reset only by
the Close button).
- Esc dismisses the modal (this overlay has no built-in Esc handling), and the
search box auto-focuses on open so the user can type immediately.
- Scale the 2-vs-3 column breakpoint (cmd-min-width) by dpiScale() — it was a
logical-px threshold compared against a physical-px available width.
- Give each category's table a unique ImGui id (##cmdsN) instead of sharing
"##cmds", and translate the 7 category names (Control/Network/... — command
descriptions stay English) into all 8 languages; CJK subset rebuilt (+2).
(Investigated but rejected: the command-name blue was flagged as low-contrast
on light skins, but the modal card is a dark glass panel on every theme, so the
original color is correct — verified by rendering marble/light/color-pop-light.)
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- D2: rpcTraceTimestamp used std::localtime (process-wide static tm) guarded by
a private mutex that can't stop another thread's localtime from clobbering the
shared buffer. Use localtime_r / localtime_s into a local tm (the codebase
pattern) and drop the now-useless mutex. Runs on RPC worker threads.
- C1: the toolbar's "<N> matches" label read filter_match_count_ before
renderOutput recomputed it, so it lagged one frame. Compute the visible/
filtered set once at the top of render() (before the toolbar) and reuse it in
renderOutput, so the count and the output share one consistent computation.
- C4: the "N new lines" indicator counted the raw drain count, inflating it with
lines the active filter hides. Count only newly-added lines that pass the
current filter, so the badge matches what the user sees on jumping to bottom.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
From the console-tab audit:
- HiDPI: the "New output" pill (140x24 + margin/rounding/border) and the JSON
indent-guide thickness were hand-drawn in raw px; scale them by dpiScale().
- Fold triangle: size it from the DPI/density-scaled gutter width (not the
zoomed font) and center it in the gutter band so glyph and clickable cell
stay aligned; only draw/handle it in the unfiltered view (folding_active_) —
in a filtered flat view a click silently flipped the collapsed flag with no
visible effect and the glyph disagreed with the rendered block.
- Selection: ignore left-clicks in the gutter (< output_origin_.x) so toggling
a fold no longer clears the user's active text selection.
- 'stop' confirmation: promote the function-local static to a member and reset
it in clear(), so a toolbar/context-menu clear between the two 'stop's can't
leave a stale arm that skips the shutdown warning.
- RPC-trace callback: dereference the console pointer under the mutex (not
after releasing it) so ~ConsoleTab can't destroy the object mid-call —
closing a latent use-after-free on shutdown (fires on RPC worker threads).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Persist the two console output display preferences and add a second toolbar
toggle for monochrome text.
- Persistence: new Settings fields console_line_accents / console_text_color
(both default true, matching the ConsoleTab statics so an upgrade re-save
can't flip visible behavior). Restored into the statics at startup in the App
constructor, and saved from a diff-and-save after the console renders (only
when a value actually changed — save() is disk I/O). The startup restore also
fixes a pre-existing bug: scanline was only synced from settings when the
Settings page or first-run wizard ran, so an existing wallet ignored a saved
scanline preference on launch.
- Monochrome text: a new toolbar toggle (FORMAT_COLOR_TEXT, dimmed when off)
next to the accent toggle. channelTextColor() early-returns the neutral
COLOR_RESULT for every channel (and JSON syntax role) when off, leaving the
left accent bars independent. COLOR_RESULT is contrast-floored per theme, so
text stays readable on light and dark terminals.
Adds console_toggle_text_color to the English source + all 8 translations.
Verified: both toggles independent, persistence round-trips across restart,
monochrome readable on light + dark skins.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adds a paint-bucket toggle button to the console toolbar (in the appearance
cluster next to zoom) that shows/hides the per-line left color accent bars for
a cleaner monochrome gutter. The icon reflects state (FORMAT_COLOR_FILL when on,
a dimmed FORMAT_COLOR_RESET when off) with a tooltip. Session-only, like the
console zoom factor; default on.
Adds console_toggle_accents to the English source and all eight translations
(no new CJK glyphs — the subset font already covers them).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The console log text washed out on Marble/Light (~1.2:1 contrast): refreshColors
chose text colors from the stored IsDarkTheme() flag (default true, and only
re-checked on a dark/light toggle) while the terminal surface used IsLightTheme(),
so dark-theme pale text landed on the light surface. Tie both to the same live
background-luminance predicate and refresh on any theme (schema generation) change.
Light themes now derive each channel from that theme's own palette (Primary,
OnSurface, Error, Secondary, ...) nudged to a WCAG contrast floor — on-theme AND
high-contrast (measured 4.6-17.8:1 across the five light skins).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Two follow-ups to the modal float work:
- Console RPC command reference (renderCommandsPopup) had the SAME auto-height
feedback bug as Wallets: BlurFloat with no cardHeight, and its fill-height
command list read GetContentRegionAvail() inside the AutoResizeY child ->
self-referential, sliding content off-screen after a monitor move. Give it a
FIXED viewport-relative cardHeight (fills ~84% of the viewport). Also removed
its duplicate body H6 heading (the framework h6 title already shows it) —
missed earlier because it's spec-form, not the positional overload.
- Wallets footer clipping: after the fixed-height conversion the content child
clips overflow (NoScrollbar), and the hardcoded 130*dp reserve below the table
was too small for the create/add-folder rows + separator + footer (~160*dp),
so the buttons clipped — worse at 150%. Compute the reserve from real widget
metrics (GetFrameHeightWithSpacing + caption height) so it holds at any DPI.
Added a modal-console-commands sweep surface (debug-gated ConsoleTab setter) so
the popup is captured. Verified both at 100% and font_scale 1.5 (150%): footers
on-screen, single headings, lists fill, nothing clipped.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
More robustness fixes from the audit (the import-key pattern, lower severity):
- Trim pasted input before use/validation: network add-server URL/label,
lite-wallet key import (+ block empty), validate-address, address-book entry.
- Confirmations for destructive/irreversible actions:
- Address-book Delete now needs a confirming second click (no undo).
- Console `stop` needs a confirming second `stop` (it shuts down the node).
- Wizard "Skip" encryption needs a confirming second click (keys stored
unencrypted).
- Send amount: normalize to 8dp (satoshi precision) on both the DRGX and USD
inputs so digits past 8dp aren't silently dropped between the preview/review
and what's actually sent.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Lower the zebra-stripe contrast (scanline-line-alpha 4 -> 2; the derived dark
band softens with it) so the alternating rows are subtler.
- Give the input's terminal bar a 1px outline along its own edges (theme-aware),
matching the output panel's glass rim, and a near-white surface on light skins.
- Add a blinking terminal caret at the end of the input when it isn't focused.
- Report a daemon started before the wallet ("no wallet-managed EmbeddedDaemon"
but RPC connected) as "Running" instead of the misleading "no daemon".
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Address the high-severity, systemic legibility issues found in the per-skin
screenshot review (verified across all 9 skins):
- Status pills (history): "Confirmed" text was 55%-alpha (dim); make it full
opacity, and give status/shield pills a visible tinted fill (a48) + 1px border
(a90) so they stop vanishing on dark-red / near-white / gradient skins.
- Console light theming: light skins now use a near-white terminal surface, and
the log text/JsonBrace colors switch to the dark defaults in light themes (the
schema colors are dark-terminal-authored — honor them only in dark themes) so
text isn't pale-on-white.
- Dark-skin muted text: bump --on-surface-medium 0.85 / --on-surface-disabled
0.58 on dragonx/dark/color-pop-dark/obsidian (obsidian addresses were ~1.1:1).
- Disabled Send button: the disabled fill was hardcoded white (invisible on
light skins) — use a dark overlay in light themes.
- Light-skin inputs: stronger frame fill + border color and a 1px frame border
for light themes so fields (and buttons) have defined boundaries on white/
pastel surfaces.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
UI-standardization audit: ~66 call sites still used raw ImGui::Button /
ImGui::SmallButton instead of the app's canonical TactileButton, so those
buttons missed the standard glass fill + rim + tactile overlay (and the
light-theme flat treatment). Convert 59 plain action-button sites across the
wizard, security dialogs, settings, market/console/explorer tabs, and the
address dialogs to material::TactileButton / TactileSmallButton (drop-in:
same signature/return, args preserved verbatim). Danger buttons keep their
PushStyleColor wrappers — Tactile renders through them then overlays.
Deliberately left raw (not plain buttons): InvisibleButton hit-targets, the
frameless transparent-bg collapsible-header toggles (RPC/Debug), the icon-only
pagination chevrons, and the receive All/Z/T segmented filter — a glass rim
would clash with those intentionally borderless/segmented looks.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per the user's selection, convert the large multi-section dialogs to the BlurFloat
style (floating content on the blur, plain heading, no glass card), each KEEPING
its authored width so width-coupled content is untouched:
- Export all keys, Console RPC reference, Explorer block detail
- Updater dialogs: daemon update, xmrig update, bootstrap download (multi-state)
Settings and Address book intentionally kept as cards (the modal Settings *window*
is dead code — show_settings_ is never set; the live Settings surface is the nav
page, not a modal; Address book has a nested edit that reads better contained).
All other confirm/input dialogs stay card-on-blur ("card for small"). Updater
dialogs are multi-state — GUI-check each state (confirm/progress/done/failed).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The zebra banding is emitted per text row, so it stopped at the last line and
left the bottom fade-padding (up to 18% of the panel) unbanded — very visible
when scrolled/pinned to the bottom. After the text-row bands, continue the
alternating pattern (same parity, lineHeight pitch) down to the panel bottom so
the zebra fills the whole output area. Also covers the case where the content is
shorter than the panel.
Full-node + lite build clean; source hygiene clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The per-line scanline shade alternated on rowIndex = floor((cumulativeY +
yOffset) / lineHeight + 0.5), but cumulativeY includes the 2px inter-line gap
while the divisor is lineHeight alone. That drifts ~0.14 row per line and, with
the rounding, jumps the index by 2 every ~7 rows — so two adjacent rows get the
same shade and the alternation visibly breaks in patches.
Replace it with a true per-visual-row counter (one increment per drawn sub-row),
seeded with the first on-screen line's sub-row ordinal so the shading stays
stable while scrolling. Parity now alternates exactly every row.
Full-node + lite build clean; source hygiene clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Work through the console items in todo.md:
- Lite filter toggles (#5): replace the single hasLogFilters() bool with a
ConsoleLogFilterCaps struct so each backend advertises which toggles apply.
Full node = daemon/errors/rpc-trace/app; lite = errors-only + app (its
diagnostics ring maps to the App/Error channels; the text filter is always
shown). Lite previously showed no toggles at all.
- Color-coded toggles (#6): each filter checkbox is tinted with its channel's
accent color (daemon=blue, errors=red, rpc=secondary, app=teal) via a new
channelAccentColor() that also drives the output's left accent bar — one source
of truth for channel color.
- Darker terminal look (#7): drop the light glass rectangle on the input; both
output and input now get a terminal-dark overlay (tabs.console.bg-darken-alpha,
default 110) and the input field blends into it (transparent frame bg).
- Monospace font (#8): bundle Ubuntu Mono (res/fonts/UbuntuMono-R.ttf, Ubuntu
Font License — same family as the existing Ubuntu fonts) via INCBIN, load it at
caption size as Type().mono(), and render console output + input in it so
pretty-printed JSON columns and terminal text align. Falls back to the
proportional caption font if unavailable.
Full-node + lite build clean; ctest green; source hygiene clean; sandboxed
startup smoke confirms the mono font loads + atlas builds without crashing.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add fold/unfold of JSON objects and arrays in console command results, built on
the refactored channel + model foundation.
- ComputeConsoleFoldSpans() (console_input_model): pure brace/bracket matcher
over a block of pretty-printed JSON lines; each opener line gets the offset to
its matching closer at the same indentation (empty/one-line blocks are not
foldable). Unit tested (nested objects, arrays, empty blocks, non-JSON, and a
string value that merely contains a brace).
- ConsoleModelLine gains foldSpan (relative, so it survives front-eviction by the
line cap) + a collapsed flag; ingest() carries the span and toggleCollapsed()
flips openers (main thread). addFormattedResult computes the block's spans and
ingests them atomically.
- computeVisibleLines() skips a collapsed block's interior (opener stays, its
foldSpan lines through the closer are hidden); folding is bypassed while a
filter is active so every match stays reachable.
- drawVisibleLines() draws a fold triangle in the free left gutter of opener
lines (result/JSON channels carry no accent bar there), toggled by a click in
the gutter cell, and appends a dim " ... }" / " ... ]" summary on collapsed
openers.
Full-node + lite build clean; ctest green (incl. fold-span + model-fold tests);
source hygiene clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Finish the thin-view pass on the remaining large render methods (pure code
motion, no behavior change):
- renderToolbar (225 -> 99 lines): extracted drawToolbarStatus(), the
drawLogFilterToggles() checkbox group, drawFilterInput(), and drawZoomControls().
- renderInput: factored the command echo + built-in interception + submit into
submitConsoleCommand(exec, cmd) -> bool, leaving renderInput to own just the
glass panel + InputText + completion/history callback.
- renderCommandsPopup (194 -> 135 lines): removed the duplicated lowercase-and-
find filter logic (match-count loop and per-row filter now share
consoleCommandMatchesFilter()) and extracted the [optional]-param dimming loop
into drawConsoleCommandParams().
Full-node + lite build clean; ctest green; source hygiene clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
renderOutput() was a ~410-line method doing filter-building, mouse/keyboard
interaction, the per-line draw loop, the new-output pill, and the right-click
context menu all inline. Split it into a thin orchestrator that calls five
single-purpose private methods (pure code motion, no behavior change):
- computeVisibleLines() -> the filter -> visible_indices_ + match count
- handleOutputInteraction() -> wheel-up, selection drag lifecycle, Ctrl+C/A
- drawVisibleLines() -> accent bars, JSON indent guides, selection +
filter highlight, scanline capture, text
- drawOutputContextMenu() -> the right-click menu
- drawNewOutputIndicator() -> the jump-to-bottom pill
Ordering and every operation are preserved exactly (interaction still runs
after layout build and before the draw; selection bounds are recomputed post-
interaction; scanline_rows_ cleared before the draw pushes to it). Full-node +
lite build clean; ctest green; source hygiene clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Pull the auto-scroll concern out of ConsoleTab into a focused, ImGui-free
ConsoleScrollController (console_scroll_controller.{h,cpp}). It owns the three
coupled fields that were loose on the god-class — auto_scroll_, the wheel-up
cooldown, and the new-lines-while-scrolled-up backlog count — and the state
machine tying them together: wheel-up detaches + starts a cooldown, the cooldown
gates the at-bottom re-enable check, backlog counting happens only while
detached, and re-pinning (checkbox / jump pill / reaching the bottom) clears it.
The view keeps only what needs ImGui: it measures scroll position, hands the
controller a plain `atBottom` bool + frame delta, and issues the actual
SetScrollHereY. ConsoleTab loses three more members and all the scattered
auto-scroll bookkeeping collapses to scroll_.* calls.
Adds testConsoleScrollController: pinned start ignores backlog, wheel-up detach +
cooldown gating, re-enable only when at bottom, checkbox toggle semantics, and
the jump-to-bottom pill. Full-node + lite build clean; ctest green; source
hygiene clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Pull the text-selection concern out of ConsoleTab into a focused, ImGui-free
ConsoleSelectionController (console_selection_controller.{h,cpp}). It owns the
anchor/caret state and all the fiddly logic around it: the mouse-drag lifecycle
(beginDrag/updateDrag/endDrag), select-all, range ordering, text extraction over
the visible model, and the index shift applied when the buffer cap evicts lines
from the top.
ConsoleTab now delegates: the four scattered selection fields (is_selecting_,
has_selection_, sel_anchor_, sel_end_) and five helper methods (selectionStart/
End, isPosBeforeOrEqual, getSelectedText, clearSelection) collapse to a single
selection_ member plus a thin selectedText() wrapper. screenToTextPos (still
view-side — it needs the frame's layout) now returns the shared ConsoleTextPos,
so ConsoleTab::TextPos is retired.
The byte-offset / eviction math was previously only reachable through the live
ImGui renderer; it now has direct unit coverage (testConsoleSelectionController:
ordering, drag lifecycle, upward drag, select-all, partial/full/no-op eviction
shift, and filtered extraction). Full-node + lite build clean; ctest green;
source hygiene clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Introduce ConsoleModel (console_model.{h,cpp}): an ImGui-free, thread-safe line
store. Producers on any thread call ingest(), which only briefly locks a small
pending queue; the main thread calls drain() once per frame to move pending
lines into the visible deque and enforce the line cap, reporting how many were
added and evicted from the front.
This fixes the real data race. Previously a single lines_mutex_ guarded both the
line container AND the UI-only interaction state (selection, scroll, auto-scroll
counters), and it was held across the whole render pass — yet render() touched
auto_scroll_ / new_lines_since_scroll_ outside that lock, racing the background
RPC-trace and app-logger producers that call addLine() on worker threads.
Now:
- addLine() only ingests (lock-free w.r.t. the model); safe from any thread.
- render() drains once per frame, then does the cap/selection/scroll bookkeeping
on the main thread (new drainModel()).
- The visible deque + all selection/scroll fields are main-thread-only, so every
lines_mutex_ lock is gone — including the one wrapping the entire renderOutput.
Adds testConsoleModel() covering ingest/drain ordering, the line-cap eviction
counts, view-only clear (queued lines survive), and an 8-thread concurrent
ingest/drain stress with no lost lines. Full-node + lite build clean; ctest
green; source hygiene clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Make ConsoleChannel the canonical semantic classification of every console
line. Producers (command executor, app log forwarders, result formatter) tag
each line with a channel; the UI derives both the text color and the left
accent-bar color from it at draw time, and the output filter keys off it.
This replaces the prior scheme of storing an ImU32 color per line and then
reverse-engineering the source from color-equality plus a "[daemon] "/"[xmrig] "
/"[app] "/"[rpc] " text prefix. Consequences:
- ConsoleLine now stores {text, ConsoleChannel} — no per-line color.
- addLine()/addRpcTraceLine() take a channel; the redundant text prefixes are
gone (the accent bar carries the origin).
- The executor's ConsoleAddLineFn hands a channel, not an ImU32, so the backend
no longer depends on ConsoleTab::COLOR_* — FullNode/Lite executors emit clean
text + channel.
- The theme-remap loop that rewrote stored colors on light/dark flip is deleted;
colors are resolved per-channel each frame (refreshColors on flip only).
- ConsoleOutputFilter drops its color fields; consoleLinePassesFilter() keys off
the channel. Test updated to the channel-based API.
- Added a Warning channel (amber severity peer of Error/Success) so the
notification + logger warning forwarders keep their color.
Full-node + lite build clean; ctest green; source hygiene clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Lift the console's word-wrap layout, screen->text hit-testing, and selection extraction
out of ConsoleTab into a pure, ImGui-free module (console_text_layout.{h,cpp}). Text
measurement is injected via a ConsoleTextMeasure interface — production uses an
ImFont-backed impl (identical CalcWordWrapPositionA/CalcTextSizeA calls as before), while
tests drive a fixed-width stub.
- BuildConsoleLayout: per-line wrap segments + heights + cumulative Y (was inline in
renderOutput).
- HitTestConsoleLayout: point -> (visibleRow, byte col) (was screenToTextPos).
- ExtractConsoleSelection: selected text across visible lines (was getSelectedText).
- ConsoleTab now holds a single `ConsoleLayout layout_` (replacing 4 mutable members;
drops the dead cached_wrap_width_) and calls the pure functions.
- New unit test testConsoleTextLayout covers wrapping, multi-row hit-testing, bottom
clamp, and filtered/unfiltered selection — the biggest previously-untested surface.
No behavior change (algorithm preserved verbatim). Full-node + lite build clean, ctest
green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Zero-behavior-change cleanup ahead of the console refactor:
- Delete dead API/members: addCommandResult (unused), handleSelection (declared, no
def), isAutoScrollEnabled (no callers), and scroll_to_bottom_ (written 8× but never
read) with all its writes.
- Drop the unused screenToTextPos `line_height` parameter (+ 4 call sites) and the unused
`segEnd` local — clears the two -Wunused warnings.
- Remove the toolbar filter checkboxes' dead `static bool s_prev_*` change-detectors,
whose only effect was setting the dead scroll_to_bottom_.
- Fix the console_new_lines format/arg mismatch (format takes one %d; the call passed a
spurious plural-suffix arg).
- Fix the channel/color incoherence: a "[daemon] error:" line kept red text but got a
blue daemon bar — the prefix no longer downgrades an error channel.
- Delete 11 orphaned lite_console_* i18n keys left over from the console merge (keep
lite_console_help_passthrough).
Full-node + lite build clean, ctest green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Draw faint vertical guides per 2-space nesting level on result (non-channel) lines, so
nested RPC JSON is easier to read. Draw-only in the per-line loop (like the channel bar),
so it doesn't touch the text layout or selection.
Note: collapsible JSON nodes were intentionally deferred — folding entangles with the
wrap cache / filter / selection line indices and warrants a runtime-verified pass.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
DanSandClaude Opus 4.8