| Age | Commit message (Collapse) | Author |
|
Firefox on Android may consume two-finger pinch for page zooming
despite touch-action: none on the canvas. Disable page zooming via
the viewport meta (maximum-scale=1, user-scalable=no) so the gesture
reaches the app as pointer events; the app provides its own zoom.
Also join touch pointers seen only via pointermove into the pointer
table, so a swallowed second pointerdown cannot break pinch.
|
|
- two-finger pinch on the viewport canvas: pointer table tracks all
active pointers; pinch emits zoom anchored at the finger midpoint
plus pan by the midpoint delta; a second finger cancels an in-flight
pan/marker drag, the remaining finger resumes panning, and pinch
gestures never fire a click
- 100dvh (with 100vh fallback) for the app shell: on mobile browsers
100vh includes the area behind the dynamic URL bar, which pushed
the compass strip below the visible fold
- file input accepts image/* so Android pickers offer all photos and
transcode exotic formats (HEIC) to decodable ones
|
|
|
|
The ResizeObserver only re-fit the image; with user_moved set (after
any manual pan/zoom) it did nothing, so the canvas backing store kept
its old dimensions and the browser stretched the stale bitmap —
visible distortion on sidebar collapse and window resize. The
observer now always bumps the redraw tick (backing store resyncs in
the draw effect, compass included) and re-fits additionally when the
view has not been moved by hand.
|
|
The flpsed.org site is generated statically; keeping the deployable
bundle in the tree lets the site pipeline pick it up directly.
Regenerate with crates/wipfel/build-release.sh.
|
|
build-release.sh assembles dist/wipfel/ with the loader page, style,
wasm bundle and the gzipped hill database. All runtime fetches are
relative to the page URL, so the directory works under any path on
any static server (verified from a subdirectory).
|
|
|
|
|
|
'Load GPX Track…' and the Track width slider now live in their own
Track section below everything else, held at the bottom of the
sidebar via margin-top: auto.
|
|
- port of the desktop bump() lens-FOV clamp: max_view_angle is the
angle at which the (distorted) projection reaches the image edge;
peaks beyond it cannot be inside the photo even when the fitted
distortion folds their projection back into the image. Recomputed
each render (idempotent, only rebuilds visibility on change) — this
is what let impossible peaks show up at high visibility settings
- 'Load GPX Track…' moved from the top into the Display section next
to the Track width slider (optional, advanced)
- collapsible control panel: collapse button in the sidebar corner,
floating hamburger button over the canvas when hidden; starts
collapsed on viewports narrower than 900px
|
|
- core: track::load_gpx_from_bytes (path wrapper delegates to it;
same hardening: size cap, no entity expansion, direct-child <ele>)
- sidebar: 'Load GPX Track…' button + hidden .gpx file input; the
track swaps in via remove_hills(FLAG_TRACK_POINT) + add_hills, so
marker hill indices never shift (track hills always live at the end
of the hill list)
- rendering: desktop parity — consecutive visible points joined with
round-cap strokes, width = scale*track_width/(dist_m*10) clamped
to [1,1000] image px, red (blue when hidden by terrain); a Track
width slider (50-1000 m, default 200) appears in Display once a
track is loaded
- track points and duplicates are excluded from guide dots, hover and
click hit-testing (they are scenery, not markable peaks)
|
|
The desktop lays out labels in alpha-descending order (the C
pointer-array sort), which on screen means right to left — so the
leftmost peak of a collision cluster ends up on top of the stack.
The web port sorted by screen x ascending, inverting every stack and
making labels look assigned to the wrong peaks. Sort descending; the
break-early scan in layout_label_y stays valid because it only relies
on monotonic x, and labels extend rightward just like in C.
Verified against the desktop rendering of the same panorama photo and
a right-to-left replay of the layout in the label simulator.
|
|
The C++ original draws a cross + vertical + horizontal leader for
every labeled peak, not just marked ones. Without them, collision-
pushed peak labels float unattached and it is impossible to tell
which dot they belong to (verified by replaying layout_label_y on a
synthetic ridge scene). Peak leaders use the muted pale green of the
label text; marker leaders stay yellow with the small cross.
|
|
Display > 'Show peak labels' toggle (off by default): every visible
peak gets its name drawn next to its guide dot, like the desktop
version. Labels join one combined collision-avoidance layout pass with
the marker labels (same layout_label_y), so marker labels keep
priority and nothing overlaps. Unmarked labels are tinted pale green;
hidden peaks and already-marked peaks get no duplicate label.
Candidates now carry name + marked flag (DotView), markers carry the
hill index (MarkerView::hill) for the dedup.
|
|
The fit at load time (and the 120 ms nudge) could still run before the
canvas had a valid layout size, leaving the first image invisible until
'Fit to window' was pressed. A ResizeObserver attached when the first
bitmap arrives fires once with the element's real size immediately
after observing, then on every later resize; it re-fits automatically
unless the user has panned or zoomed by hand since the last load
(user_moved flag, reset on each image load). Window resizes now
re-fit the image as well instead of only redrawing it.
|
|
'gipfel web' is taken by an existing related project. The browser app
is now wipfel — web based gipfel photogrammetry for mountain images:
crate gipfel-web -> wipfel, updated titles, docs, plan and dev script.
The shared engine crate stays gipfel-core.
|
|
The core already ran both visibility heuristics (height/distance
threshold via the Visibility slider, occlusion via mark_hidden with
the cone-visibility slope test) — the web UI just ignored FLAG_HIDDEN.
- guide dots: hidden peaks are filtered out by default; with the new
Display > Show hidden peaks toggle they appear grey instead of green
- hover and click hit-testing skip hidden peaks unless the toggle is on
- Find Peak picker is unchanged (it lists pan.close, which always
includes hidden peaks — same as desktop)
|
|
`if let Some(idx) = nearest_hill(&pan.borrow(), ...)` keeps the
immutable borrow alive through the whole if-let block (temporary
lifetime in the scrutinee). Marking a peak from that path called
do_refit -> borrow_mut() -> panic, which aborted the wasm instance and
froze all further interaction (no marker removal, no drag refit).
Scope the borrow to its own block.
|
|
Distance-ranked candidates are useless while the initial viewpoint is
still rough — the true peak may project far from the click. Mirrors the
desktop dialog instead: click opens a search box over pan.close
(geographically near hills, dedup/track filtered), name-substring
filter, sorted, capped at 300 entries, 'Name (1234 m)' labels,
autofocused search, Cancel button. Selection marks the clicked
position with the chosen hill, whatever its current projection.
|
|
- click away from a dot now opens a Find Peak list (candidates within
60 px, sorted by distance, name/height/pixel-offset labels) — the
desktop dialog equivalent for crowded ridgelines
- sliders update on input, not on release
- fit is graduated like the original: 1 marker fits the center angle,
2 add scale, 4+ the full viewpoint (removed the >=4 gate)
- image load now re-fits once more 120 ms after the first paint
(same nudge as the desktop app) so the picture never stays hidden
|
|
- compass.rs: canvas port of the desktop Aligned compass strip —
adaptive tick step, direction labels (red N), center triangle and
azimuth readout, drag to rotate the view (Newton inverse projection,
same solver as desktop; round-trip tests ported and passing)
- sidebar sections like the desktop: Viewpoint (lat/lon/height number
inputs), Orientation (azimuth/focal/pitch/roll sliders, desktop
ranges), Projection (fit-distortion toggle), Display (visibility
with the 0.11-r inversion, label size 8-28 px)
- every control applies to the Panorama immediately (focal maps via
scale = focal * w / 35, same formula as the metadata path)
- marker labels: ported the desktop layout_label_y collision push-up
with yellow leader lines (cross, vertical, horizontal), label font
size live-adjustable
- main area split: canvas + 56 px compass strip
|
|
- guide dots: dark outline ring + bright core, readable on any photo
- hover over a dot shows 'NAME · height m' in the sidebar (same info
the desktop hover status line gives)
- pointermove doubles as hover report when not dragging
|
|
- fetch gipfel.dat.gz at startup, gunzip (flate2, pure Rust) and load
into a Panorama held in app state
- image loads now also parse the gipfel JPEG comment / EXIF and set the
viewpoint via apply_image_metadata (same path as desktop)
- markers: click a visible-peak guide dot to mark it, drag to adjust,
right-click to remove; refit (comp_params_with, distortion locked by
default) after every change once >= 4 markers
- sidebar: fit-distortion toggle, marker/zoom readout, fitted
az/pitch/roll/scale line
- canvas: guide dots for visible peaks, red crosses + shadowed labels
- serve-here.sh: dev helper (cargo build --target wasm32 + wasm-bindgen
+ gzip the DB into the static dist dir)
Verified: cargo check clean for wasm32; runtime to be confirmed in
browser against a gipfel-saved photo (W2 acceptance: fit must match
the desktop result on the same image).
|
|
trunk resolves the crate manifest next to the entry HTML; with the
entry in www/ it looked for www/Cargo.toml and failed. Move the entry
to the crate root (standard trunk layout), keep style.css in www/.
|
|
First milestone of wasm-plan.md:
- new crate gipfel-web: Yew 0.21 (csr), canvas 2D viewport
- file input + drag-and-drop; browser-native decode via createImageBitmap
(faster than image-crate decode in wasm, GPU-backed blitting)
- pan (pointer drag with capture, touch-ready) and cursor-anchored
wheel zoom, DPR-aware canvas sizing, fit-to-window
- dark sidebar shell: open/fit buttons, status line
- Trunk.toml + www/ (index.html, style.css); README with dev setup
Verified: cargo check clean (native and wasm32-unknown-unknown),
release wasm builds (5.3 MB pre-wasm-opt), all 38 native tests pass.
Runtime behavior to be confirmed with trunk serve on a desktop browser.
|
|
Scope (feature-parity map with binding v1 non-goals), architecture
decisions (Yew, canvas 2D hybrid rendering, localStorage/PWA storage,
byte-oriented core APIs), port map from the desktop app modules,
milestones W1-W6 (~3 weeks), risks, and open questions to settle
before W1.
|
|
|
|
Proves the gipfel core can run in the browser via WASM. Enabled by the
earlier GSL removal: the core is now pure Rust.
gipfel-core gains byte-oriented entry points (no filesystem) alongside
the existing path-based ones:
- hill::load_from_bytes, Panorama::load_data_from_bytes
- metadata: load_image_from_bytes, read_jpeg_comment_from_bytes,
write_jpeg_comment_to_bytes, save_image_to_bytes (EXIF via Cursor)
New crate gipfel-wasm (experimental, not shipped with the app):
- wasm-bindgen surface: version() + fit_demo(db_bytes, jpg_bytes)
which runs the whole pipeline in memory: parse hill DB, read gipfel
JPEG comment / EXIF, synthesize 5 observations from visible hills,
perturb, refit (pure-Rust LM), report JSON with reprojection error
- demo/index.html: file pickers + result pane (wasm-pack --target web)
- build.sh: adds the wasm target, builds with wasm-pack, prints serve cmd
- native test runs the same pipeline against gipfel.dat + test.jpg and
asserts sub-pixel reprojection error (passes here)
To try on a machine with rustup:
cd crates/gipfel-wasm && ./build.sh
|
|
|
|
|
|
The .gitignore pattern 'harness/build' was anchored to the repo root and
did not match tests/harness/build, so the compiled C++ harness binaries
leaked into the tree and the source tarball. Fix the pattern and untrack
them (they are regenerated by tests/gen_golden.sh).
|
|
|
|
|
|
Version now lives only in [workspace.package] in the root Cargo.toml;
all three crates inherit it via version.workspace = true (this also fixes
gipfel-core, which had drifted to 0.1.0 while the others were 0.5.4).
edition/license are inherited the same way.
The About dialog now reads env!("CARGO_PKG_VERSION") instead of a
hardcoded string. A new test (changelog_top_matches_version) asserts the
top gipfel-X.Y.Z entry in ChangeLog equals the crate version, so the
manifest, the About dialog and the ChangeLog cannot drift apart.
AGENTS.md gains a "Releasing a new version" section documenting the flow
(bump workspace version, add ChangeLog entry, tag, git archive source
tarball).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The Rust port (formerly gipfel-rs) is now the official successor to the
FLTK/GSL C++ code base. This squashes the port into a single commit and
removes the C++ sources (they remain in the git history below).
Layout:
crates/gipfel-core headless core: hills, panorama geometry, projections
(maxima-generated derivatives), pure-Rust LM fit,
gipfel: JPEG comment + EXIF I/O, profiles, stitching
crates/gipfel-cli command-line tools (export, info, stitch, profiles)
crates/gipfel-app Cosmic DE (libcosmic) GUI, binary name `gipfel`
codegen/ maxima scripts + regen.sh for the projection code
tests/ golden files + C++ reference harnesses (historical)
The original README is kept for now; it will be updated to describe the
Rust build. gipfel.dat, AUTHORS, ChangeLog and NEWS are unchanged.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|