Issues

• Velocity inlet with slip side faces amplifies a small transverse disturbance until the guard trips (TRT, τ ≤ 0.507, empty box)

  needs-triage
  tangency-labs/fieldcraft#820 · Filter by author bluecmd opened on Nov 3, 2026

• Uniform pressure outlet destroys an exact uniform-flow state and trips the stability guard (TRT and MRT)

  area:cfdneeds-triage
  tangency-labs/fieldcraft#819 · Filter by author bluecmd opened on Nov 3, 2026

• Bit-identical CPU speedups from a profiling pass (#804–#813)

  needs-infoneeds-triage
  tangency-labs/fieldcraft#818 · Filter by author semmelrinde opened on Nov 3, 2026
  1

• FDM parts: take the layer direction from SolidCraft's print face, and report the stress across the layers

  needs-triage
  tangency-labs/fieldcraft#802 · Filter by author Tyler-Chiriboga opened on Nov 2, 2026

• Translate the desktop app's interface, starting with French

  enhancementneeds-triage
  tangency-labs/fieldcraft#801 · Filter by author HiroKenEz opened on Nov 2, 2026

• lbm-gpu: the throughput benchmarks fail in parallel on NVIDIA with "the readback did not complete"

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#800 · Filter by author thomschlie opened on Nov 2, 2026
  1

• fieldcraft-cli panics when its stdout is closed (| head) or full

  bugneeds-triage
  tangency-labs/fieldcraft#799 · Filter by author thomschlie opened on Nov 2, 2026

• Staircase geometry: drag depends on how many cells span the body and on where it sits on the grid (periodic Stokes sphere +11.7 % at R = 3 cells; Schäfer–Turek C_d,max +6.7 % at D/Δx 20 and −8.5 % on a half-cell shift)

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#798 · Filter by author bluecmd opened on Nov 2, 2026

• At the default lattice velocity 0.05 the Re 3900 cylinder develops ±5 % density excursions and a local Mach of 0.15; halving it cleans the field but leaves the drag at +38.6 %

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#797 · Filter by author bluecmd opened on Nov 2, 2026

• Short-window block error bars understate LES drag error: a later 400–600 D/U window raises the Re 3900 cylinder's C_d by 22.6 % while the force noise grows, and ±0.0003 intervals sit on a 6.7 % bias

  area:cfdarea:uibugneeds-triage
  tangency-labs/fieldcraft#796 · Filter by author bluecmd opened on Nov 2, 2026

• Channel studies' stopping rule tests the signed lag-1 block correlation, so anticorrelated (alternating) blocks pass as "decorrelated"

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#795 · Filter by author bluecmd opened on Nov 2, 2026

• spalding_bounce_halfway amplifies an odd–even mode of the wall rows with every collision operator (+5.6e-5 to +1.3e-1 per step) and breaks the run

  area:cfdarea:iobugneeds-triage
  tangency-labs/fieldcraft#794 · Filter by author bluecmd opened on Nov 2, 2026

• The runtime guard certifies execution, not validity: completed LES runs with ±25–37 % density excursions, local Ma 0.35 and force spikes 10–20× the mean are reported like any other

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#793 · Filter by author bluecmd opened on Nov 2, 2026

• Sphere at Re 100 with TRT at τ = 0.518: the run goes unstable at the velocity inlet (GPU at 11.4 D/U, CPU at 19.7 D/U); the regularized collision survives but gives C_d +12.5 % and spurious unsteadiness

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#792 · Filter by author bluecmd opened on Nov 2, 2026

• Coarse TRT LES runs to completion with a negative mean drag (C_d −3.8 ± 4.5) and the guard never trips

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#791 · Filter by author bluecmd opened on Nov 2, 2026

• MRT fails within the first D/U at every LES bluff-body setting tried (τ ≈ 0.5001–0.5005)

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#790 · Filter by author bluecmd opened on Nov 2, 2026

• Square cylinder at Re 22 000: C_d 4.11 ± 0.14 against about 2.1, with force spikes to 24.8 and an unusable Strouhal number

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#789 · Filter by author bluecmd opened on Nov 2, 2026

• Cylinder Re 3900: mean drag moves by −25 % (3D vs one-cell span), +80 % (Musker wall model on) and +22 % (dynamic HRR σ, which also suppresses spanwise motion)

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#788 · Filter by author bluecmd opened on Nov 2, 2026

• Circular cylinder at Re 3900: C_d 1.354 ± 0.040, 36.8 % above the experimental 0.99, and not converged across N = 12 / 16 / 24

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#787 · Filter by author bluecmd opened on Nov 2, 2026

• Musker wall model: the carried wall stress is 16.75 % below Dean while the model's inferred stress is within 2 % (channel Re_τ 950)

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#786 · Filter by author bluecmd opened on Nov 2, 2026

• LES accuracy and stability: drag and wall-stress errors (tracking)

  area:cfdbugneeds-triage
  tangency-labs/fieldcraft#785 · Filter by author bluecmd opened on Nov 2, 2026

• app: the macOS save chooser panics when AppKit returns no NSSavePanel

  needs-triage
  tangency-labs/fieldcraft#780 · Filter by author jyviko opened on Nov 1, 2026

• mcp: screenshot_takes_a_camera_orientation is flaky on lavapipe in CI (two pictures of one camera differ)

  area:infrabugneeds-triage
  tangency-labs/fieldcraft#778 · Filter by author jyviko opened on Oct 30, 2026

• Tet mesher: the exact-geometry snap lowers the worst dihedral angle by 10° and more on curved parts

  area:meshbugneeds-triage
  tangency-labs/fieldcraft#777 · Filter by author jyviko opened on Oct 30, 2026
  3

• lbm-gpu: Smagorinsky Couette GPU-vs-CPU density bound fails on Apple Metal

  area:cfdneeds-triage
  tangency-labs/fieldcraft#776 · Filter by author jyviko opened on Oct 30, 2026
  3