Delaunay refinement split a boundary face edge at its midpoint even when that
midpoint lay inside the diametral sphere of a feature edge beside it. Each such
split halves the face vertices' distance to the curve, so they piled up against
it (the heated bracket grew face vertices 0.016 mm from a 4 mm-sized edge), the
improvement schedule could not lift the tetrahedra there to its target, and
inserted vertex after vertex around them: 29257 elements and about two minutes
to open instead of 23641 and 6 s. The SolidCraft pin move (4c91f9b) added one
interior vertex to the bracket's end faces, which changed the CDT enough to
expose it.
Following Shewchuk (1998), an encroaching midpoint now splits the encroached
feature edge instead.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
examples.md and mcp.md tell the one-datum workflow and its numbers; the
crank gallery captures and Start card thumbnails are taken again on this
build; ROADMAP marks the automatic blocks of extruded cross-sections done,
and features.md's Decompose entry says what is left (3D parts).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
The release job signs SHA256SUMS.txt into SHA256SUMS.txt.minisig with rsign2 when the
FIELDCRAFT_SIGNING_KEY secret exists, and checks the signature against the public key the app
is built with; without the secret the draft is unsigned, the run warns and the notes say the
app cannot install that release itself. docs/releasing.md says how the key is set up, kept
and rotated; docs/updates.md describes checking, verifying and installing, the web app's
Reload to update, and FIELDCRAFT_UPDATE_HOST for tests. privacy.md, control-protocol.md,
web.md and the roadmap say what changed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
The same check as the app's Help › Check for Updates and fieldcraft-cli update --check: the
latest release, whether it is newer, its page, whether it is signed and the asset this install
would download. It only reports. Checking reaches the network, so only the desktop app and the
CLI switch it on; tests, scratch sessions and the browser build refuse without asking.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
A region of an extrusion's (or revolution's) cross-section that no
structured scheme fits is now cut by straight lines into blocks that fit,
before interval assignment, so the cuts are curves of the arrangement and
their counts are assigned with the rest; nothing reaches the model
(hex::decompose, after Tam and Armstrong's medial-axis subdivision and Li,
Armstrong and McKeag's primitives).
Per region: several holes are separated by lines across the gaps between
neighbouring holes; a piece with one hole gets the O-grid layout whose
pieces score best (lines from the outer corners towards the hole or along
their bisectors, from the hole's corners, or along four directions from
the hole); pieces without holes are cut by a two-ply search over lines
from concave corners and fillets and from convex corners. Candidates are
scored by meshing the pieces with their schemes (smallest corner scaled
Jacobian). A region's blocks are kept when fair (at least 0.5) or no worse
than its free mesh, and cuts never end inside a seeded curve.
This replaces the cuts from reflex corners along an edge. Tests on an L,
a T, a plate with a hole, with a slot and with three holes, and the half
crank's cross-section: every region structured, the quads conforming and
the smallest corner scaled Jacobian bounded. The half crank arm now
extrudes in 12 structured regions with a hex20 minimum scaled Jacobian of
0.37 (the arm meeting the bosses 20 degrees off tangent bounds it) where
it had 2 free regions and 0.009.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
The first run asks whether to share anonymous usage (only when the host
says this run may), the answer and a random install id are kept in the
preferences, and Help › Settings… changes it. Commands are counted by
where they came from (ui, palette, script; mcp when the MCP server says
so), and cancelled forms and clicks on commands that cannot run yet are
recorded. The preferences go through a PrefsStore (a file, or the
browser's storage). Control channel: ui.settings, ui.prompt answers.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
All six docs/media/examples images at 1400 x 860 in the light theme, the
solved sample in the home view (the crank mirror from below, as its text
says), with the wireframe symbols, no SETUP card and the crank's planes
hidden; the Start card thumbnails cropped to the model from them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
docs/crash-reports.md: what is kept where, what a bug report carries
(and only when ticked), the CLI and commands for agents; README's
privacy note; the control protocol's new prompt answers.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
crash: a panic hook and a locked per-run command journal (adapted from
SolidCraft's crash reports). A panic in a command becomes a report with
the model from before it and the command; one outside commands, or an
unclean exit found at the next start (abort, segfault, a crash in a
graphics driver, a kill: what #248 needs), a report with the model the
journal starts from (autosave, file or a copy) and the commands since,
the last control request and the dead run's log lines. Reports are
bounded, deduplicated by signature and replayable (crash::replay, with
writes confined to a scratch folder). The crash folder follows the
automation temp area in --control and CI runs.
logfile: a small size-rotated file logger behind the log facade.
bug_report: GitHub bug-form drafts. Version, OS and what happened by
default, home folder shown as ~; the command journal only with Share my
session, the model only with Include the model as well; a session too
long for the link becomes a file to attach and the clipboard text.
Commands: diagnostics.crash_reports, diagnostics.crash_report,
diagnostics.bug_report (Tools › Help, with its icon) and
diagnostics.crash_test (FIELDCRAFT_CRASH_TEST=1 only).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
Right-click on any browser item (body, surface, datum, partition, set,
material, section, step, boundary condition, load, reference point,
coupling, contact, fluid domain) opens its menu: Edit…, Hide/Show,
Isolate, Show All, Active in step (boundary conditions and loads) and
Delete…. A right click on a datum or a partition's cut face in the
viewport opens the same menu. Each item the viewport draws on its own
has an eye; hidden items are neither drawn nor picked. Datums and
partitions are selectable with properties. Delete… and the Delete key
ask first (the prompt lists what uses the item: a datum's partitions
are deleted with it, other definitions in use are refused); ui.prompt
answers it. Every entry runs the same command as the ribbon and forms.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
view.hide, view.show, view.isolate and view.show_all change which
bodies, surfaces, datums, boundary conditions, loads, reference points
and couplings the viewport draws. The hidden items are view state: the
manifest's `view` keeps them, a change marks the model unsaved, but it
is neither journalled nor undone and leaves results current. Items that
go away are forgotten. Headless pictures leave hidden items out; the
glyph builders take a filter. Tools > View > Show All claims the
commands.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
A datum point by coordinates, at a mesh vertex, a fraction along a
design edge, at a face's centre, or where a datum axis crosses a datum
plane. Datum planes take three datum points or a normal to a datum axis;
axes take two datum points. geometry.datum.edit, .axis.edit and
.point.edit redefine a datum of their kind (partitions cut again);
geometry.datum.delete with cascade also deletes the partitions that cut
by it. The ribbon shows Datum Plane, Datum Axis, Datum Point and
Partition, each with its own icon; edit and delete leave the ribbon
(still commands, claimed by their entries).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
crank-arm (and crank-mirror, kept as the same model under the name of
its mirror workflow) now cut the crank on its symmetry plane and the
half into seven cells: across the arm at x = 30, 60, 110 and 148, and on
two diagonals radial to the pocket's corner fillets. Every cell's
cross-section maps (four-sided blocks, the half O-grids round the bores,
triangle primitives at the pocket's corners); exact edge seeds give each
cut the same divisions on both sides, so mesh.cells merges all six cuts
and mesh.mirror reflects the half: 7 710 hex20, scaled Jacobian min 0.37
(was 0.17, and 0.009 for the swept mirrored half), edge ratio at most
7.0. Root σxx ±62.43 MPa against M/Z 62.6 MPa, the spindle carries the
1500 N. The test asserts the quality floor. Gallery images regenerated.
Fixes#247
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
A cross-section region with three corners (a fillet's quarter disk, a
pocket corner cut off by a partition) was left to the free all-quad
mesher. It now gets midpoint subdivision's triangle primitive (Tam and
Armstrong 1991; Li, Armstrong, McKeag 1997): lines from a centre to a
node on each side split it into three mapped blocks, the interior then
smoothed and every quadrilateral checked (a fold falls back to free).
Interval assignment gains dominance constraints (a side shorter than
the other two together) so the split always exists.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
Boundary condition, load and interaction symbols are wireframes (thin lines,
no shaded cones), so they never hide the mesh or a contour plot:
- a crossed wire triangle per constrained translation, tip on the node,
coming from outside the part where the set's faces say which side that is;
a held rotation adds a smaller triangle behind it (the double mark); fixed
shows all three axes, a symmetry plane only its normal;
- forces as open-headed arrows ending on the node, as long as four support
units at one place and 2.5 over a set, so loads and supports stay in
proportion (#78); moments as an axis line and a right-handed curved arrow
(#79); contact and tie surfaces as small squares on their faces;
- a constant size on screen at any zoom (#80): the symbol unit is the size
setting (View > Symbol size, `ui.set {symbol_size}`, 14 pt by default)
times the model length a pixel spans, rounded to a quarter of a doubling;
- spread evenly on a grid a few symbol units across, so their density follows
the screen, never more than 400 per definition;
- a layer of their own on the desktop (GPU and CPU): zooming rebuilds and
uploads the symbols only, never the mesh's scene; framing leaves them out;
- colour per kind from the theme's tokens (a new contact colour), and the
definition selected in the browser drawn in the selection colour, wider;
- depth-tested against the model with the line bias that keeps them on their
own surface; a load at a reference point or a single node and a reference
point's cross are drawn over the part (#78).
Headless pictures (MCP screenshot, `ui.render`) frame the camera first and
draw the symbols at the default size for it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
An orbit turns about the model point under the press (the pick ray's
first hit on the plot, the mesh or the design), or the centre of what is
drawn when it starts on empty space, keeps that point still on screen and
marks it while orbiting. The wheel and a pinch zoom toward the point
under the cursor; the wheel's delta is read as it arrives, so a notch
zooms in full in orthographic and perspective views alike (#273).
Mouse presets and reverse zoom come from SolidCraft (prefs.rs NavPreset,
NAV_PRESETS; MIT OR Apache-2.0): FieldCraft's own (left drag per the
navigation bar) stays the default, Ctrl + left drag from empty space
orbits in every preset. They are chosen in the display menu, settable
with ui.set and kept in prefs.json beside recent.json (the browser build
keeps them for the page's run); an older prefs.json still loads.
Fixes#262, fixes#273.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
Each top-level node that is a setup stage carries a mark at the right
of its row beside its count (done, next, blocked) with the reason and
place in its hover, and a node still lacking what its stage needs shows
a ghost row that starts the providing command. ui.inspect reports each
node's state as browser.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
The form's OK on a mesher whose body already has a mesh opens a Replace
mesh prompt ("CrankArm already has a mesh (7,274 hex20). Replace it?"),
naming sets that list the old labels and results that will be cleared.
Replace runs the form; Cancel leaves it open. The control channel's
ui.dialog submit is the OK button, ui.prompt answers replace or cancel.
Programmatic calls never ask. Fixes#267.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
Every fit frames what is drawn (the deformed shape and the setup's symbols
included) with Camera::fit_in in the part of the viewport the overlays
leave free: the results legend, the selection bar, the gizmo and the
navigation bar (framing::free_part). Standard views fit too, so a long
beam's front view keeps both ends (#147); fit and home leave the legend
clear (#76). The view frames itself in the home view when another model,
sample or design is opened (Session::opened; the same design opened again
after an edit keeps the camera) (#233), and when results come up with part
of the model out of sight (#223). ui.inspect's layout reports the legend.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
The first look at a solved static step draws the von Mises contours on the
true geometry; the deformed shape comes when asked for (Deformed,
results.deformed, results.animate), at the automatic scale. Frequency steps
still open on mode 1 deformed and explicit steps at true scale, as the
owner's notes in #77 ask; a 0.24 mm deflection drawn as 10 mm no longer
reads as a failed part.
Fixes#77
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
file.recovery.list gives each entry's age; File › Recover shows name, age and counts, newest
first, with Discard all. docs/formats/fieldcraft.md states the folder and pruning rules.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
mesh.tet.generate, mesh.hex.*, mesh.hex.blocks and mesh.cells give every
seed on the open design to their meshers, which match them to their own
curves by position, so a seed on an edge two cells share holds in both. The
tet reply gives, per seed, the mesh edges along it. Docs and the roadmap
say what is honoured where.
Fixes#230.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
Seed design edges by their persistent names with a count or a size, a bias
and a constraint; one undo step each, saved with the model. Seeds on edges a
reopened design lacks are reported as stale.
Part of #230.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
A reference field (a material, set, step…) with no items was a free-text box; it is now a greyed
drop-down ("No materials yet") with a button that opens the command making one (the
collection's requirement provider). That form comes back to this one on OK, with the new item
picked, or as it was on Cancel. A required reference with one candidate, or with the browser's
selection of its kind, starts picked when the form opens, so the fields and the JSON view agree.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
The mesh tool runs mesh.cells (method cells) and mesh.mirror (method mirror).
The server instructions and docs/mcp.md list the partition, per-cell, mirror
and symmetry commands and point to the crank-mirror sample.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
file.sample crank-mirror partitions the crank arm on its symmetry plane, sweeps
one half all-hex (hex20) and mirrors it into one conforming mesh, then defines
the crank-arm model. Samples may now run preparing commands (datums,
partitions) before their script sees the bodies. Its test checks the halves
alike and joined, every element valid and the sets on both halves; the heavy
test solves it: sigma_xx at the root -62.43 / +62.43 MPa against M/Z = 62.5.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
A symmetry condition (zero normal displacement, zero rotation about the
in-plane axes) on a datum plane normal to X, Y or Z, with its set given or made
of the open design's faces on the plane (a partition's cut faces), so half a
symmetric part is solved as a half. Half a 40 mm bar cut at its mid-width, with
the condition and half the load, bends as the whole bar to 2e-12 with half the
reaction. Inclined planes are refused as not supported yet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
The bodies' elements are reflected across a datum plane (or a point and
normal) with fieldcraft_mesh::mirror and joined on it. The image is tagged as
the design's body at the reflected place (found by volume and centroid, or
named in `as`), its faces with that body's faces under them, so sets by face
name reach both halves. Half a bar meshed hex8 or hex20 and mirrored has the
whole bar's node and element counts and volume, and its bending deflection to
1e-12; tet4 and tet10 mirror valid and conforming.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
Each cell (a body of the open design) gets its own method: structured hex where
one fits (auto falls back to tet and says why), or tet. All go into one mesh.
Cells on either side of a partition are joined: where their element faces
match node for node the nodes are merged; otherwise (hex next to tet, or
different divisions) the two partition faces get a set each and a dual mortar
tie, the reported hex-tet transition. A 40 mm bar cut in half, one half hex,
the other tet, stretches F L / (E A) to 1e-6.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
Datum planes by principal plane and offset, three points, or point and
normal; datum axes by principal axis, two points, or point and direction.
geometry.partition cuts the open design by a datum plane and keeps the
partition in the model, so geometry.open, file.open and undo cut the design
again. Requirements, forms and the Prepare entries are wired; features.md
lists them as ready.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
Datums are named planes (point, unit normal) and axes (point, unit direction);
a partition names the datum plane that cuts and, optionally, the bodies. A
datum a partition uses cannot be deleted or turned into an axis. Both are
stored in model.json, partitions in the order they cut.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
The ribbon buttons, drop-down rows, command search and the title bar's Solve read
Session::availability: greyed, with what is missing (and where to get it) in the tooltip.
Starting such an entry opens no form; the status bar names what is missing. ui.toolbar
reports each entry's `enabled` and `unavailable`, ui.inspect the Solve button's.
Fixes#220
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
CommandSpec gains `requires: &[Requirement]` (crates/engine/src/requirement.rs): an open
design, geometry, nodes, a mesh, something to analyse, materials and sections for the
elements, sets, the definition an edit or delete acts on, a step, a result. Session::execute
checks them before running, so a command missing one fails at once with no partial work,
naming each missing piece and the toolbar entry that provides it ("Needs a material —
Properties › Material"). Requirements are real prerequisites (the model's own reference
checks and model.check's rules), never a workflow order. engine.commands / list_commands
list them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
Opening (File > Open, a drop, the recent list) a design that has none of the
bodies the model's mesh and sets are on now asks, like the unsaved-changes
prompt: Start a new model with this design (default, Enter), Keep the
current model (the status names what no longer resolves), or Cancel. The
opened session waits for the answer; the same design opens without asking.
ui.inspect shows the prompt as {"design": path}; ui.prompt answers it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
A Markdown table sizes its columns by the captions, so the pictures came out
at different sizes; an HTML table with a third of the width per cell keeps
them even. The crank-arm picture is scaled to 1400x860 like the others, and
flow-past (no picture yet) moves to the end.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
cargo-deb and cargo-generate-rpm (pinned) package the same stripped binaries and files as the
tar.gz under /usr. The release job compares the file lists, runs lintian, and installs both
packages in Ubuntu 22.04 and Fedora containers.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
Plane strain, plane stress and axisymmetric models take contact pairs and
general (self) contact on the linear edges of tri3 and quad4 elements:
edges oriented out of their elements, the edge mortar of fem with the
thickness or 2 pi r in the weights, the element's area over the edge's
length as the thickness under it, and the in-plane rows of the tangent
blocks. Verified by the contact patch test in plane strain and
axisymmetric (to 1e-9 small sliding, 1e-4 finite), Hertz's line contact
on quad4 edges (Johnson 1985, 4.3; equal to the hex8 model's to 1e-5),
and a hybrid rubber C-ring closing on itself that moves node for node as
the one-layer solid ring.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
One constant pressure per element (mean dilatation, Nagtegaal, Parks &
Rice 1974; the hoop strain in the divergence) for quad4, quad8 and tri6:
the linear stiffness, stresses and thermal load, and the finite- and
small-strain forms of nonlinear steps with their consistent tangent. The
linear triangle and plane stress are refused by name.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
file.sample crank-arm: a 170 mm 6061-T6 crank built in SolidCraft
(crates/engine/samples/crank-arm.json), swept along z by mesh.hex.generate
as 7274 hex20 elements of 3 mm (scaled Jacobian min 0.17). The hex variant
drops the pocket floor fillet and boss chamfers, and its root arcs meet the
bosses 20 degrees off tangent so the arm's top face has no zero-angle cusp.
The square bore is fixed and 1500 N acts on the pedal hole; the faces are
picked by their surfaces, not their names. Its heavy test matches the root
bending stress to M/Z = 62.6 MPa within 3 % (0.3 % in practice).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
file.sample bracket-hex: the corpus bracket split into 30 SolidCraft cells,
meshed by mesh.hex.blocks as one conforming hex20 mesh of 3 mm, clamped at
the upright and loaded at the foot. The sample's script finds the clamp and
load faces in every cell that has them. Its test matches the tip deflection
of a fine tet10 solution within 1 % and checks the clamp carries the load.
Closes#166
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UhhZMBTviDQxZvwnyP89fo
