Restore RepeatedCallsReturnIdenticalPoses for a 1 x 1 square and a 1.5 x 0.75 rectangle.
Both cases fail intermittently because Default Fill can return different, equally scored
lattices for identical calls (Project Memory opennest c98c21bd), so they are skipped with
that reason until the defect is fixed rather than removed.
CutoutLatticeFill fills one closed cutout of a frame part with copies of one part, the
first piece of the cutouts-first pass. It runs Default Fill over the cutout's bounds
plus one part step on every side, shifts the lattice over a (2n+1)^2 grid of offsets
up to half a step each way, and at each offset keeps the copies whose reference point
lies in the part's inner-fit region of the inscribed cutout (part circumscribed and
grown by the spacing). The offset keeping the most copies wins; every returned pose
is then certified with NestLayoutCheck.Clears against the frame and the other copies.
20" ring, 10" round cutout, 1" squares, 0.25" spacing: 37 copies, against 25 for a
block sized to the inscribed rectangle and 32 for the unshifted lattice.
Not wired into any engine or pipeline: parts inside cutouts wait on containment-aware
cutting order. Fill can still return different, equally scored lattices between calls
for some parts, so identical results are not yet claimed.
FillPattern gathered per-angle results in a ConcurrentBag and kept the first best it
enumerated, so an exact tie between angles went to whichever worker finished first:
identical inputs could return different, equally scored layouts. Results now sit in
slots indexed by angle, V before H within an angle (the order the single-angle bag
already produced and the existing tie tests pin), and ties keep the earliest slot.
Single-angle behaviour is unchanged.
Regression: a never-prefer comparer over one 0-degree and fifteen 180-degree angles
must keep the 0-degree layout in 50 of 50 calls; it failed on the old code.
Move BlockCatalog's private-drawing Fill invocation (BestFitCache stabilization plus
the Default FillItem call) into PrivatePlateFill so the cutout fill can reuse it
without duplicating the cache handling. Behavior unchanged; Irregular filter 96/96.
TestAssemblySetup set the unhandled-exception mode and a ThreadException
handler from a module initializer. WinForms stores both per thread
(ThreadContext/NativeWindow thread statics), so they applied only to the
runner thread that loaded the assembly, never to the STA threads the
tests run on. There, an exception escaping a window message or a
BeginInvoke/SynchronizationContext callback still opened a modal
ThreadExceptionDialog in an interactive session, or was silently ignored
otherwise; the handler would also have logged and swallowed it.
Replace the fifteen copied RunSta helpers with StaTestThread.Run, which
selects UnhandledExceptionMode.ThrowException on each STA thread before
it creates a window, so the exception propagates to the test body and
fails the test. Each class keeps its own timeout and message; all
threads are now background threads. Remove TestAssemblySetup.
StaTestThreadTests covers a throwing BeginInvoke callback through
StaTestThread (fails with the original exception) and, as a control,
the default mode on a plain STA thread (the exception goes to the
thread's handler and never reaches the test body). Cross-compiled on
Linux; Windows execution is the windows-desktop CI job.
The first windows-desktop run hung for 25 minutes with no output and
ran until it was cancelled. Pass --blame-hang-timeout 5m so a test that
is silent for five minutes is killed, named in the log and mini-dumped
into TestResults, then cap the job at 20 minutes and upload all of
TestResults, not just the TRX files.
Checked locally with a throwaway xunit project: a Thread.Sleep(Infinite)
test was killed after the configured inactivity, named in the output,
and left a TRX, a test-order (Sequence) file and a mini dump; dotnet test
exited 1.
The new windows-desktop CI job surfaced several tests that had never
actually run on Windows:
- SeparatorPenLifetimeTests expected Control.Disposed to fire only
once; current WinForms re-fires it on every Dispose() call.
- EditNestFocusTraversalTests assumed the Plates tab was selected by
default, which stopped being true once Drawings became the landing
tab; select it explicitly before asserting focus.
- NestReportExportTests asserted an exact plate count that ignored
PlateManager's trailing sentinel plate, and didn't expect the nest
name in the busy-export message.
- AtomicReportFileTests only expected IOException when replacing a
directory target; Windows raises UnauthorizedAccessException instead,
which callers already treat the same way.
Also register a ThreadException handler so an exception inside a
WndProc callback fails the test normally instead of popping a modal
dialog that blocks the desktop.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
DataGridView.EndEdit() hangs/throws instead of returning false when
CellValidating has already cancelled the pending edit, so an unparsable
stock size froze TryGetPlateOptions. Validate the pending edit text
ourselves before ever asking EndEdit() to commit or discard it.
Also include the nest name in the export-busy message, matching how
the invalid-name rejection already does.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Pushes to master only ran the Linux job, which can cross-compile but
not execute OpenNest.WinForms.Tests; Windows test runs happened only on
release branches and tags. Add a windows-2022 job to the cross-platform
workflow that runs OpenNest.WinForms.Tests and OpenNest.FrontEnd.Tests
in Release on every master push and pull request, runs both projects
before failing, and uploads their TRX results.
New only fell back to the legacy DefaultUnit setting when the whole
defaults file was missing or invalid, so a readable file without a
usable unit produced inches. BOM import never assigned units at all,
and the Nest Defaults dialog showed inches for a missing file.
All three now load through MainForm.LoadSavedNestDefaults, which passes
DefaultUnit as the per-field unit fallback. BOM-created nests start from
NestDefaults.ApplyTo, like New, before applying each group's plate size
and spacing.
The ApplyTo test now asserts units; removing the unit assignment from
ApplyTo fails it. WinForms behaviour is compile-verified only on Linux.
Enum.TryParse accepted numeric strings ("7" loaded an undefined unit,
"1" loaded millimeters) and comma-joined names, and a JSON number for
units failed the whole file. Units now load only from a defined unit
name; anything else falls back for that field alone, while the other
valid fields still load.
Add Load(path, fallbackUnits, out status) so callers can keep their
existing unit preference when the file is missing, unusable, or lacks a
usable unit (previously a readable file without units silently became
inches). An undefined fallback is ignored in favour of the built-in
units.
The numeric/comma/JSON-number rows failed against the previous commit;
dropping the fallback assignment or its IsDefined guard fails the new
fallback tests.
NestDefaults.Load caught only JsonException and IOException, so a
defaults.json that File.Exists can see but the process cannot read
threw UnauthorizedAccessException out of New. Report such files as
Invalid and return the fallback values, like a corrupt file.
The regression mode-000s the file on Unix (skipped on Windows and when
running as root); it failed with UnauthorizedAccessException before
this change.
Brings in the independently reviewed exact-contact repair (c7b4ba4,
25755a8, bd5db59): line-only part outlines keep exact analytic bounds,
so an 8x3 part fits rotated on 8.5x3.5 stock at 0.25 spacing, while
curved outlines keep their conservative padding and no validator,
spacing or fit tolerance changes. The held hole-profile and pocket
commits built on bd5db59 stay unmerged.
A request is now marked superseded before the previous one is
cancelled, so a query that completes synchronously on cancellation
returns quietly instead of clearing the page and surfacing a stale
cancellation error. NestBrowseSession.IsLoading follows only the latest
request, and the dialog's Previous/Next and Loading text use it rather
than a count that included superseded requests. Each request disposes
its own cancellation source when it ends.
SQLite LIKE stops matching at an embedded NUL, so search=Alpha%00x
behaved like 'Alpha' followed by a wildcard and returned rows that do
not contain the whole search text. The shared query validation now
rejects NUL, so the server answers 400 and the client throws before
sending.
The Database-mode Open dialog no longer downloads every record. It
browses through NestBrowseSession, which sends one bounded query per
change: the filter box (300 ms debounce) searches on the server and
returns to the first page, column headers sort every match on the
server (a second click reverses), and Previous/Next page by 100 with a
range and total in the status line. A response superseded by a newer
request is cancelled and discarded, refresh steps back when the current
page was emptied, and failures clear the rows and show the error
without a modal box per keystroke.
The query accepts sort=<column>&order=asc|desc over a fixed allowlist
(saved, name, customer, status, material, dates, thickness, plate and
part counts, made by, comments, file size). Text columns sort with
NOCASE, ties break by id in the same direction so pages partition the
matches, and unknown or numeric sort values and other orders return 400.
The client sends the sort as its camelCase name.
GET /api/nests/query filters name, customer, material, made by, comments
and status (stored and display names) with an escaped LIKE parameter,
orders newest saved first with an id tie-break, and returns one page of
at most 500 records plus the total match count, read in one database
hold. Unknown, repeated or out-of-range parameters return 400.
RemoteNestRepository.QueryAsync validates the same bounds before
sending, rejects oversized pages, and reports a 404 from an older server
as a server that needs updating. GET /api/nests stays the full
enumeration used by backup, restore checks and the container smoke.
- Run as the .NET image's non-root app user (UID 1654) with root-owned
application files and an app-owned /app/data that fresh named volumes inherit.
- Add a curl HEALTHCHECK on /healthz (30s/5s/10s/3, 2s start interval), keep the
explicit 8090 URL and clear the base image's 8080 port default.
- Parameterize VERSION/SOURCE_REVISION and base images; add OCI labels with
development defaults.
- Add an image-only Compose example (required image and bind address, named
volume, cap_drop ALL, no-new-privileges) and an env template.
- Document deployment, the upload limit, scoped ownership preparation, and
checked backup/restore/upgrade functions that refuse existing destinations and
verify the metadata list and every archive hash before switching volumes.
- Extend the container smoke: image user/healthcheck/label contract, PID 1 UID,
capabilities and writable paths, Docker-reported health, near-limit and
oversized uploads, stopped-service backup restored into a second volume, and
startup failure on read-only and root-owned data mounts.
Orientation bounds were padded by the arc chord tolerance even when the outline has no arcs, so a part that exactly fills the work area after rotation was rejected before packing. Pad only outlines with flattened arcs; footprints and NFPs still use the chord tolerance.
aj
