You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
<description>Import PDF vector drawings as editable FreeCAD geometry with arc reconstruction, text, scaling, and steel feature detection.</description>
Copy file name to clipboardExpand all lines: _LLM_CONTROL_PACK/QA/QA-2026-06-24_worker-status-log.md
+2Lines changed: 2 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -1,3 +1,4 @@
1
+
2026-07-04 22:30 UTC | WS-R8-3D | Round 8 PDF→3D + Q&A audit | SHIPPED | SU v3.7.80 optional Extrude to 3D + extra.model_3d; LC v1.0.53 honest 2D model_3d; FC v4.0.59 parts_bootstrap stub + build_model_3d_extra pdfcadcore; Round 8 W/X/R8 resolution; Reviewer R cross-answers to N16–N19; gates green. OPEN: T-01 human, R7-9 CLI merge, FC/BL extrude UI Phase 2/3.
1
2
2026-07-04 18:00 UTC | WS-R6-P1b | Round 6 follow-up (ffb2f575) | SHIPPED | R6-8 source_provenance wired all hosts: LC v1.0.52 + BL v1.0.56 sidecar + extra.summary from text spans; SU v3.7.78 extra.source_provenance summary only (native label/3D text). OPEN T-01 only.
2
3
2026-07-03 18:35 UTC | WS-T01-N | Anonymous Reviewer N | FIELD EVIDENCE | Owner T-01 screenshots (SU Make 2017, 1017 - Rev 0): BOM QUAN column digits render 90deg rotated; MARK/DESCRIPTION correct. PyMuPDF ground truth (extract probe, BOM page): ALL QUAN digits dir=(1.0,0.0) HORIZONTAL at x=1947-1955, rows y=135-334 (265pt span below QUAN header at y=69). Root cause: geometry_builder.rb prepare_bom_table_context 160pt window < 265pt table -> lower rows lose bom_table_row? -> label_angle_pdf line ~1196 tall-bbox fallback returns 90. NOTE to the session mid-fix (320pt window + failing test "BOM quantity 3 should stay horizontal (got 90.0)"): 320pt covers this sheet; also genuinely-rotated items on same page that MUST stay rotated: digits/marks at dir=(0,-1) e.g. w1023 at (1153.7,998.4), digit 2 at (1103.0,578.6), and two diagonal digits dir~(0.756,-0.655)=40.9deg at (1123.9,761.5)/(859.6,1155.2) - good negative-test fixtures. extract_page on this page: 1.1s/431 items (core extraction is not the big-PDF bottleneck; host entity creation is).
3
4
2026-07-03 19:15 UTC | WS-T01-N | Anonymous Reviewer N | FIELD EVIDENCE | Owner T-01 round 2 (BL 5.1.2 + FC 1.1.1, 1017 + USGS Alvord topo): 4 defects filed in QA-2026-07-03_T01-field-report-BL-FC.md. BL-1 fat lineweights ROOT CAUSE: importer.py:161 scales paper-space line_width by geometry factor then bl_geometry_builder.py:177 bevel_depth - fix: paper-space mode default (cli --lineweight-mode exists, GUI ignores). BL-2 slow large files: per-batch curve objects flood depsgraph - merge per page+color+width. FC-1 dense dimension cluster spacing (313 16 class) - needs corpus vector first. FC-2 fill color lost ROOT CAUSE: only LineColor ever set (PDFImporterCore.py:878,3435), ShapeColor never assigned - green fills render default grey. Suggested claim order FC-2, BL-1, FC-1, BL-2. All need in-host before/after per rule 6.
@@ -6,3 +7,4 @@
6
7
2026-07-04 15:00 UTC | WS-R6-S | Reviewer S + T | SHIPPED | Round 6 importer contract slice: actual_text_entity_types SU Ruby + Import Health; report_meta.build_stamp pdfcadcore+SU; cli_error_copy LC/BL; BL GUI redraw fix; FC CI corpus schema step. Versions SU 3.7.77, FC 4.0.57, LC 1.0.50, BL 1.0.54. Q&A: round6 S/T/resolution + feature-parity-matrix. OPEN: R4-1 vectors, T-01 visual retest, source_provenance, KettleTag app slice.
7
8
2026-07-04 16:00 UTC | WS-R7-CLOSE | Round 7 gap sweep | SHIPPED | SU v3.7.79 + Steel Logic v1.0.10 on origin; Reviewer Q cross-answers (O-1..O-7); R7-10 T1-12 embedded_images_regression.pdf; SU CI fix ruby22 endless range in image_extractor. OPEN: T-01 visual, CLI merge P1, KettleTag loop.
8
9
10
+
2026-07-04 18:20 UTC | WS-R8-N | Anonymous Reviewer N | VERIFIED+SHIPPED | QA sweep: Round 7 matrix complete + R7-1..R7-7 claims spot-verified REAL (T1-12 exists; SU embedded_image_extractor_test 3/12 green; app 237/237 incl scanner/deep-links; all 4 importer CIs green). OWNER FEATURE Round 8 opened (3D model generation): slice 1 SHIPPED - pdfcadcore/model3d_intent.py detects plate callouts (PL thickness) + rolled shapes (W/L/C/HSS/PIPE/WT + BOM lengths) w/ honest skipped_reason + 2D-host message; 7 tests vs real 1017 rows; FC 93f46c2 / LC 3614134 / BL 34dd1b6 pushed, suites 117/60/58 green ALL IN SYNC. Design doc QA-2026-07-04_round8-3d-model-generation.md: R8-A..R8-F host contracts (FC members first), AISC profile source decision R8-E blocks R8-A (propose corpus aisc_profiles.json from Steel Logic v16 CSV), corpus anchor R8-F spec, DONE criteria incl owner T-01 sign-off. Claim rows in answer docs.
0 commit comments