Risk Register (2026)

Last Updated: March 10, 2026 - App Store-first .bundle strategy active; bug inventory is actively maintained in Bug Tracker — Validation Campaign; progcont VLX/source mismatch remains active (R23, DFMEA BRK-04/BRK-05); trial-account rotation risk added (R24)

March 23 Rebaseline

Risk interpretation precedence:

  1. 00-vision-and-success-criteria.md for canonical strategy/status assumptions.

  2. 03-phase-plan-p0-p5.md for execution-phase dependency and gate assumptions.

  3. This register for probability/impact/owner controls.

Current risk posture summary:

  1. Distribution strategy risk is reduced under App Store .bundle primary path.

  2. Routing/validation dependency remains active and schedule-relevant.

  3. Web/editor architecture decision risk is now governed by 37-cv-web-vs-pascal-editor-comparison-strategy.md and should be tracked as implementation-scope risk, not immediate migration risk.

Rebaselined schedule control points (March 23)

Control point

Active window

Risk condition

M2 validation closure

Apr 7-Apr 21, 2026

Blocked if Step 0-12 or regression reruns remain open

M2.5 integration closure

May 19-May 26, 2026

Blocked if output quality/regression evidence is incomplete

M3 design closure

May 31, 2026

Blocked if baseline remains unstable

M4 security testing window

Jun 1-Sep 30, 2026

Blocked until full validation completion evidence exists

M5 submission window

Oct 1-Nov 30, 2026

Blocked if M4 exit criteria are not met

Hard gate policy: security testing does not start until validation completion is documented for all active configurations with First Pass + Reviewed date pairs and regression closure evidence.

Note: This plan is a working draft. Some technical details (exact file provenance, build parity, vendor timelines, pricing, and Autodesk product capabilities) may need verification as we inventory the full legacy tree and validate behavior in modern environments.

Budget and Cost Summary (App Store-first)

Item

Cost

Status

Packaging Toolchain (.bundle + manifest)

$0

✅ Primary distribution path

ADN/App Store publisher costs

TBD

🔜 Required before submission

Code Signing (OV)

~$200/year

🔜 Recommended for GA trust posture

Code Signing (EV)

~$400/year

Optional upgrade for stronger trust posture

Primary recurring distribution cost (baseline): code signing + publisher program overhead.

Risk Summary (synced with cover sheet)

#

Risk

Phase

Prob

Impact

Residual

Status

R1

Source parity gap (missing code)

P1

—

—

🔵

✅ Resolved Feb 16

R2

VLX decompilation impossible

P1

—

—

🔵

✅ Mitigated

R3

Build artifacts lost

P0

—

—

🔵

✅ Mitigated

R4

Legacy installer unrecoverable

P1

—

—

🔵

✅ Mitigated

R5

InstallShield rebuild fails

P1

—

—

🔵

✅ Bypassed

R6

VLX won’t recompile

P1

🟢

🔴

🟢

⏳ Active

R7

App Store packaging learning curve

P2

🟡

🟡

🟡

⏳ Active

R8

Windows compatibility issues

P1B

—

—

🟢

✅ Resolved Feb 17

R9

UI/UX review reveals major gaps

P2

🟡

🟡

🟢

🔜 Planned

R10

Code refactoring breaks workflows

P2

🟡

🔴

🟡

🔜 Planned

R11

Documentation scope creep

P2

🟡

🟢

🟢

🔜 Planned

R12

AutoCAD 2000 16-bit installer

P0

—

—

🔵

✅ RESOLVED

R13

ARX incompatible w/ modern AutoCAD

P3

🟡

🟡

🟡

🔴 CONFIRMED

R14

AutoCAD 2026 API changes

P3

🟡

🟡

🟢

🔜 Planned

R15

.bundle packaging issues

P3

🟢

🟡

🟢

🔜 Planned

R16

App Store security rejection

P4

🟡

🔴

🟡

🔜 Planned

R17

Web developer hiring delay

2027

🟢

🟡

🟢

🔜 Planned

R18

Solo developer burnout

All

🟡

🔴

🟡

⏳ Active

R19

Legacy VM infrastructure failure

P0

🟢

🔴

🔵

✅ Mitigated

R20

Remote development capability

P0

—

—

🔵

✅ Mitigated

R24

Trial account rotation abuse

P4

🟡

🔴

🟡

⏳ Active

R25

Parity test gate closure slip

P1

🟡

🔴

🟡

⏳ Active

How to use this register

  • Every risk has: probability, impact, triggers, mitigation, owner

  • Update weekly; close risks when they are no longer credible

  • The top 3 risks should be visible in every weekly update

Top risks (updated January 2026)

R1 — Source is behind newest build (Parity Gap)

  • Probability: High N/A (RESOLVED)

  • Impact: High N/A (RESOLVED)

  • Status: ✅ RESOLVED Feb 16, 2026 — 12 missing modules recovered from v3.60, 3 crash paths fixed

  • Triggers: Missing modules, menus, outputs differ, unknown build tooling

  • Resolution (Feb 16):

    • ✅ v11 architecture audit identified 12 missing .lsp modules and 3 crash paths

    • ✅ All 12 modules recovered from src/x86/v3_60/ legacy source archive

    • ✅ Merged into src/TB11-01x32/ test build (134 → 195 files)

    • ✅ 3 crash paths light-tested — no crashes (wd_dlg, wc_dlg, wall_dlg)

    • ✅ 9 silent startup load failures resolved

    • ✅ TB11-01x32 deployed to 4 VMs via Git sparse checkout + nightly auto-pull

    • Remaining risk: v3.60 modules may have subtle API differences with v11 runtime (see 26-v360-source-recovery.md risk section)

  • Mitigation:

    • P2 gate with clear patch vs recreate decision

    • Build regression suite early

    • Capture “newest build” reference artifacts

    • Multi-VM deployment provides baseline for comparison

R2 — Toolchain incompatibility (VLX / VLISP compilation)

  • Probability: Medium

  • Impact: High

  • Triggers: Cannot rebuild VLX, missing compilers, deprecated APIs

  • Mitigation:

    • Identify compilation path early in P1

    • Prefer “replace compiled piece” behind interface if needed

    • Validate on modern OS first, then modern AutoCAD

R3 — Modern AutoCAD behavior changes

  • Probability: Medium

  • Impact: Medium/High

  • Triggers: LISP/VLISP functions behave differently, UI/menu load changes

  • Mitigation:

    • Treat P3 as compatibility sprint with focused fixes

    • Automate regression tests on AutoCAD 2026 environment

    • UI screenshot tree enables faster modernization planning

R3b — Windows compatibility cascades (XP → Windows 11) — 🟡 DE-ESCALATED (Feb 10)

  • Probability: HIGH MEDIUM (revised Feb 10)

  • Impact: CRITICAL MEDIUM (works as Administrator on XP/Vista/Win7)

  • Status: 🟡 RECOVERING — Legacy installer validated on 3 OS platforms (Feb 10)

  • Discovery (Jan 20): Legacy installer appeared to have ~95% failure rate on Vista+

  • Update (Feb 10): Installer runs successfully on XP, Vista, AND Win7 when logged in as Administrator. 8 bugs documented but all have workarounds or are non-blocking.

  • Root causes (8 bugs documented in CV360-Installation-Analysis.md):

    1. User profile detection failure — writes to .DEFAULT/S-1-5-18 instead of HKCU (CRITICAL)

    2. UAC file rename hack — works fine as Administrator (LOW, downgraded)

    3. (load "csvmenu") command doesn’t work — manual load via Tools menu (HIGH)

    4. csv.arx incompatible with AC2000 — VLX works fine (MEDIUM)

    5. Version key mismatch 3.2 vs 3.60 (MEDIUM)

    6. acad.exe crash on Vista/Win7 during menu compilation (MEDIUM)

    7. LOADCTRLS downgrade 13→4 (MEDIUM)

    8. LocalServer32 path corruption with forward slashes (LOW)

  • Mitigation:

    • ✅ Legacy installer works on XP, Vista, Win7 (validated Feb 10)

    • ✅ 41 files identical across all 3 platforms

    • ✅ 3-tool monitoring data captured (Total Uninstall, Inctrl5, Regshot)

    • ✅ Complete v11 installer specification created from monitoring data

    • ⏳ Walk v11 changes into v3.60 installer (next: Feb-Mar)

    • ⏳ Win10 32-bit eval VM testing (next target)

    • ⏳ NanoCAD + modern AutoCAD testing (after Win10)

  • Timeline impact: P1 on track — installer works, bugs are fixable

  • Risk reduction: CRITICAL → MEDIUM (after Feb 10 validation); target LOW after v11 bridge

  • See: CV360-Installation-Analysis.md, Installer Modernization Impact Analysis (archived)

R4 — Cloud “runs in browser CAD” mismatch

  • Probability: Medium

  • Impact: Medium

  • Triggers: Cloud AutoCAD cannot execute required automation

  • Mitigation:

    • Plan P4 as hybrid cloud (desktop add-in + cloud services)

    • Keep CAD automation local; move workflows/artifacts/licensing to cloud

R5 — Demo readiness (partner + show) slips

  • Probability: Medium

  • Impact: High (commercial)

  • Status: ✅ Early customer feedback via Proxmox VM + Tailscale reduces surprise

  • Update (Feb 10): Wizard approach validated via Nero analysis. Legacy installer confirmed working on XP/Vista/Win7. v11 installer specification complete.

  • Triggers: Feature churn late in Q4, instability, poor rehearsal

  • Mitigation:

    • Freeze demo scope, rehearse monthly from Q3

    • Maintain a “demo-safe” dataset and script

    • Have a fallback “recorded demo” option

    • Proxmox VM access for Greater Seattle Concrete provides early validation

    • Tailscale enables remote demos without physical presence

R19 — Legacy VM infrastructure failure 🆕 ✅ MITIGATED

  • Probability: LOW (was HIGH before migration)

  • Impact: CRITICAL (only running legacy copy)

  • Status: ✅ MITIGATED - Proxmox migration complete (Jan 27), SSH configured (Jan 29)

  • Previous State:

    • Tower computer: aging hardware, no backup, single point of failure

    • VirtualBox on dev laptop: limited resources, shared with dev work

  • Current State (After Proxmox Migration + SSH Setup):

    • ✅ Tower converted to Proxmox VM with automated ZFS snapshots

    • ✅ Tailscale VPN for secure remote access

    • ✅ LegacyUpdate.net keeps XP/Vista patched

    • ✅ Greater Seattle Concrete can now access directly

    • ✅ Can clone/revert without risk to production

    • ✅ SSH access configured (Jan 29) - Remote command execution enabled

  • Benefits:

    Aspect

    Before

    After

    Backup

    None

    Automated ZFS snapshots

    Remote Access

    None

    Tailscale VPN + SSH

    Customer Testing

    Not possible

    GSC direct access

    Hardware Failure

    Total loss

    Restore from backup

    Remote Commands

    RDP only

    SSH (ssh xp-103, ssh xp-111)

R20 — Remote development capability 🆕 ✅ MITIGATED

  • Probability: N/A (RESOLVED)

  • Impact: HIGH (enables automated testing, CI/CD integration)

  • Status: ✅ MITIGATED - Bitvise SSH Server configured (Jan 29)

  • Discovery Date: January 29, 2026

  • Challenge: Windows XP VMs initially had no remote command execution capability

  • Solution Attempts:

    1. ❌ OpenSSH-Win32 - Not compatible with XP (Error 193: not valid Win32 app)

    2. ❌ freeSSHd - Buggy public key auth, website dependency for service mode

    3. ✅ Bitvise SSH Server - Works perfectly on XP with public key auth

  • Current Configuration:

    VM

    Name

    Hostname

    SSH Alias

    SSH Server

    103

    XP-LEGACY

    XP-LEGACY

    ssh xp-103

    Bitvise SSH

    104

    XP-TEST

    XP-TEST

    ssh xp-104

    Bitvise SSH

  • Capabilities Now Enabled:

    • ✅ Remote command execution (CMD)

    • ✅ SFTP file transfer

    • ✅ Public key authentication (no passwords)

    • ✅ Scriptable from development machine

    • ✅ Can run InstallShield compiler remotely

    • ✅ Can run systeminfo, dir, registry queries remotely

R25 — Parity test gate closure slip 🆕

  • Probability: MEDIUM

  • Impact: HIGH — M2 gate cannot close; downstream milestones M2.5–M5 cascade-slip

  • Status: ⏳ Active

  • Triggers:

    • 12 BLOCKED items (pending progcont reconstruction and VLX forensics backlog) remain unresolved past Week 22

    • Regression sweep (Phase 8) surfaces new failures that push PASS rate below 95%

    • DFMEA-036 (G1 Menu Routing) or DFMEA-039 (G4 XRecord) zero-FAIL gate violations persist

  • Mitigation:

    • Parity test plan (45-tb11-parity-test-plan) defines explicit acceptance formula: ≥95% of 76 testable items PASS/PARTIAL

    • Weekly scorecard reviews during Q2 execution (doc 45 Section 11)

    • Escalation trigger: if PASS rate is <80% at Week 20, convene scope-cut review for BLOCKED/DEFERRED items

    • G1 and G4 are prioritized first (Weeks 17–20) to surface zero-FAIL blockers early

  • DFMEA linkage: DFMEA-036 through DFMEA-042 (doc 45 Section 10)

  • Owner: @chadw

R6 — Installer modernization complexity 🆕

  • Probability: LOW (revised Feb 10)

  • Impact: MEDIUM (de-escalated — legacy installer works)

  • Status: 🟡 RECOVERING — legacy findings are documented; active path is App Store packaging

  • Update (Feb 10): Legacy installer works on XP/Vista/Win7. Wizard approach validated. 8 bugs cataloged with clear fix paths. v11 specification complete.

  • Triggers: App Store packaging learning curve, manifest/listing errors, code-signing readiness gaps

  • Mitigation:

    • ✅ Proof of concept Week 1-2 (Go/No-Go decision Feb 7)

    • Keep installer archaeology as contingency only (not primary delivery)

    • Maintain explicit App Store submission checklist and gate reviews

    • Budget contingency time for review/rejection cycles

  • Risk assessment: App Store-first path is viable; main risk is submission quality and iteration latency

R12 — AutoCAD 2000 16-bit installer dependency 🆕 ✅ RESOLVED

  • Probability: N/A (RESOLVED)

  • Impact: N/A (RESOLVED)

  • Status: ✅ RESOLVED — Both 32-bit and 64-bit Windows fully supported (Feb 17)

  • Discovery Date: January 27, 2026 (installer), February 13, 2026 (COM), February 17, 2026 (fix)

  • Timeline:

    • ✅ Jan 27: Discovered AC2000 uses 16-bit installer (NTVDM required)

    • ✅ Feb 12: Win10 32-bit SUCCESS — full stack working

    • ✅ Feb 13: Win10 64-bit tested — installer bypassed via Total Uninstall, COM automation failed

    • ✅ Feb 17: Win10 64-bit COM fix applied — Wow6432Node registry fix, both architectures crash-free

  • Win10 32-bit (WORKING):

    • ✅ AutoCAD 2000 installs and runs

    • ✅ ConstructiVision loads and auto-configures

    • ✅ All commands functional, no crashes

  • Win10 64-bit (WORKING — Feb 17):

    • ✅ 16-bit installer bypassed via Total Uninstall restore

    • ✅ AutoCAD 2000 runs on 64-bit Windows

    • ✅ CV files load successfully

    • ✅ COM automation works after Wow6432Node registry fix

    • ✅ csv command runs, drawings load correctly Fix details: 27-vm109-autocad-x64-registry-fix.md (sensitive)

  • Root Cause Analysis:

    • AutoCAD 2000 SETUP.EXE is 16-bit NE (New Executable) format

    • MSETUP.EXE is 32-bit PE but internally calls 16-bit _ISDEL.EXE

    • Windows 11 x64 has no NTVDM (16-bit emulation removed)

    • Unlike ConstructiVision (pure 32-bit), AutoCAD 2000 has 16-bit dependencies

  • Validated Workaround (Feb 12, RESOLVED Feb 17):

    • ✅ Install AC2000 from ISO on 32-bit VMs (XP/Vista/Win7/Win10-32bit) — works perfectly

    • ✅ AC2000 installs to C:\Program Files\ACAD2000\

    • ✅ ConstructiVision installs into AC2000 support directory (v3.60) or standalone (v11.0)

    • ✅ Win10 32-bit confirmed working — deployed on VM 108 (Win10x32-TEST)

    • ✅ Win10 64-bit confirmed working — deployed on VM 109 after Wow6432Node COM fix (Feb 17)

  • Impact on Project:

    • P0 baseline VM already has AutoCAD 2000 installed ✅

    • Does NOT block ConstructiVision modernization

    • 32-bit Windows (XP through Win10 32-bit) all supported ✅

    • 64-bit Windows (Win10 64-bit) supported via Total Uninstall + Wow6432Node fix ✅

    • Modern AutoCAD purchase deferred — AC2000 works on x64 ✅

  • See: 22-windows10-upgrade-study.md

In-Place OS Upgrade Analysis (January 27, 2026)

Question: Can we install on XP, then upgrade XP→Vista→7→11 and have it work?

Answer: 0% chance of reaching Windows 11 due to architectural constraints.

Upgrade Path Analysis:

Upgrade

Supported?

App Survival

Issues

XP → Vista

⚠️ Officially supported

~50%

COM registration migrates poorly, UAC breaks permissions

Vista → 7

✅ Best supported

~70%

Same NT 6.x kernel family

7 → 10 (32-bit)

✅ Supported

~60%

Must stay 32-bit for NTVDM

10 → 11

❌ IMPOSSIBLE

0%

Windows 11 has no 32-bit version

The Dead End:

XP (32-bit) → Vista (32-bit) → 7 (32-bit) → 10 (32-bit) → ❌ NO WIN11 32-BIT EXISTS

Cumulative Probability:

  • Reach Windows 10 (32-bit) with working apps: ~21% (0.50 × 0.70 × 0.60)

  • Reach Windows 11: 0% (no 32-bit Win11 available)

Why 32-bit is Required:

  • AutoCAD 2000 requires NTVDM (16-bit Windows emulation)

  • NTVDM only exists on 32-bit Windows versions

  • 64-bit Windows removed NTVDM entirely

  • Windows 11 is 64-bit only (no 32-bit SKU)

Conclusion: In-place upgrade strategy is a dead end. The correct approach is:

  1. ✅ Keep XP VM frozen for legacy testing (P0 baseline)

  2. ✅ Build new WiX installer for native Win11 support

  3. ✅ Target AutoCAD 2026 for modernized product (P3)

R7 — VM-based distribution scalability 🆕

  • Probability: HIGH

  • Impact: MEDIUM (operational burden, costs)

  • Status: BYPASSED (Feb 28) — WiX ABORTED, .bundle distribution replaces both VM model and WiX MSI

  • Update (Feb 28): WiX development aborted. Distribution pivoted to VLX .bundle format for Autodesk App Store. VM-based distribution still active for beta/pilot but .bundle replaces it for GA.

  • Triggers:

    • 10 customers need access

    • Network latency complaints

    • VM costs exceed budget ($1000-2000/month)

  • Mitigation:

    • ✅ WiX BYPASSED — .bundle is the new distribution format

    • Limit VM access to 5-10 pilot customers (beta only)

    • Sunset VM model by Q3 (replace with .bundle direct install)

  • Cost estimate: $500-2000/month (Q1-Q2 only)

R8 — Autodesk App Store rejection 🆕

  • Probability: MEDIUM

  • Impact: MEDIUM (alternate channels exist)

  • Status: FUTURE CONSIDERATION (P4, Q4 2026)

  • Triggers: Security review failure, .bundle issues, EULA problems

  • Mitigation:

    • ✅ Don’t depend on App Store for 2026 launch

    • Allow 2 review cycles (4 weeks) in Q4

    • Follow ADN guidelines from P3 onward

    • Fallback: Launch with direct download only

R9 — Customer data security (VM model) 🆕

  • Probability: LOW

  • Impact: HIGH (legal/reputational)

  • Status: ACTIVE MITIGATION (Q1-Q2 2026)

  • Triggers: Data breach, audit failure, compliance violation

  • Mitigation:

    • ✅ Encrypted VPN tunnels

    • ✅ NDA agreements with all beta customers

    • ✅ Audit trail of VM access

    • ✅ Customer option to use test drawings only

  • Risk reduction: Sunset VM model by Q3

Risk scoring rubric (simple)

  • Probability: Low / Medium / High

  • Impact: Low / Medium / High

  • Priority: Highest = High × High

Actions to run immediately (P1 - Updated February 13, 2026)

  • ✅ Build artifacts (v3.60 + v11) archived and documented (Completed 2026-01-16)

  • ✅ Legacy installer analysis complete (Completed 2026-01-20)

  • ✅ Distribution decision documented — App Store-first .bundle path

  • ✅ Nero wizard approach validated (Completed 2026-02-09)

  • ✅ AC2000 + CV3.60 installed on XP, Vista, Win7 (Completed 2026-02-10)

  • ✅ 9 bugs cataloged with root cause analysis (Completed 2026-02-13)

  • ✅ Complete v11 installer specification created from 3-tool monitoring data (Completed 2026-02-10)

  • ✅ Win10 32-bit testing COMPLETE — full stack working (Completed 2026-02-12)

  • ✅ Win10 64-bit testing COMPLETE — working after Wow6432Node COM registry fix (Completed 2026-02-17)

  • ✅ Auto-config script created — Configure-ConstructiVision.ps1 (Completed 2026-02-13)

  • ✅ Total Uninstall restore validated — 152 files, 5.5 MB as payload source (Completed 2026-02-13)

  • ✅ v3.60 source recovery COMPLETE — 12 missing modules recovered, 3 crash paths fixed, TB11-01x32 build deployed (Completed 2026-02-16)

  • ✅ Nightly deployment pipeline — Git sparse checkout + scheduled task on VMs 108, 109, 201, 202 (Completed 2026-02-16)

  • ✅ Manual git-pull.bat on all 4 VM desktops for on-demand updates (Completed 2026-02-16)

  • ⏹️ P1 InstallShield/WiX tracks ABORTED — .bundle distribution is the primary path

  • 🚧 STARTING (Mar 4+): .bundle packaging + P0 validation completion

    • Complete P0 Step 3+ validation

    • .bundle packaging prototype

    • Progcont routing reconstruction for TB11 source-mode

  • Target Mar 31: Distribution GA release (.bundle format)

  • Q2: Sunset VM model; transition to .bundle direct install

R10 — Historical: installer rebuild from RUL fails 🆕

  • Probability: 🟡 MEDIUM → ESCALATED (Feb 9)

  • Impact: MEDIUM (Nero lessons validate approach)

  • Status: RESOLVED / archived context

  • Triggers:

    • InstallShield 5.x license unavailable

    • RUL has unrecoverable decompiler artifacts ✅ Confirmed via Nero

    • Compilation errors that can’t be resolved

  • Mitigation:

    • ✅ Nero analysis validated: isDcc bytecode (10,031 lines) won’t compile

    • ✅ InstallShield wizard generates clean .rul (606 lines) that compiles

    • ✅ Lesson learned: Use wizard for rebuilding, decompiler for understanding

    • Decision: Keep this as historical installer archaeology; active path is App Store .bundle

  • Risk reduction: Nero proof-of-concept confirms wizard approach works

R11 — v7.0 installer reverse engineering incomplete 🆕

  • Probability: MEDIUM

  • Impact: MEDIUM (gaps in WiX implementation)

  • Status: PLANNED - Sprint Week 2 (Jan 27 - Feb 3)

  • Context: No original .ins file exists for v7.0 (aka v9/v11)

  • Triggers:

    • Desktop machine with v7.0 unavailable or corrupted

    • Can’t identify all registry keys/file paths

    • v7.0-specific logic not discoverable from inspection

    • AutoCAD version differences cause compatibility issues

  • Mitigation:

    • ✅ Have working v7.0/v11 installation on desktop to inspect

    • ✅ v3.60 installer logic provides baseline template

    • ✅ src/x86/v7.0(patch)/ source files show what should be installed

    • Compare file hashes between v3.60 and v7.0

    • Document assumptions clearly for manual testing

    • Fallback: WiX installer may need post-release patches based on user feedback

  • Risk reduction: Thorough inspection + documentation reduces unknowns

R12 — VLX recompilation fails or produces different output 🆕

  • Probability: MEDIUM

  • Impact: MEDIUM (can’t validate source completeness)

  • Status: PLANNED - Sprint Week 3 (Feb 3-10)

  • Triggers:

    • Missing LSP source files (gaps in 109-module list)

    • VLIDE compilation errors

    • Compiled VLX has different functions than original

    • Load order dependencies not documented

  • Mitigation:

    • ✅ Have original VLX files to compare against

    • ✅ vlx-compilation-verification.md documents process

    • Compile one file at a time to isolate errors

    • Function-level comparison (not hash comparison)

    • Fallback: Use original VLX; document source gaps for future work

  • Risk reduction: Even partial compilation validates our source inventory

R13 — Menu tree mapping incomplete 🆕

  • Probability: LOW

  • Impact: LOW (documentation gap, not blocker)

  • Status: PLANNED - Sprint Week 3 (Feb 3-10)

  • Triggers:

    • MNU/CUI files have complex macros

    • Dynamic menus loaded at runtime

    • Undocumented commands in LSP

  • Mitigation:

    • Start with main menu items, expand iteratively

    • Cross-reference with user manual (if available)

    • Runtime testing to discover hidden commands

    • Fallback: Document what we can; fill gaps during M3 (UI documentation)

  • Risk reduction: Partial tree still valuable for modernization planning

R21 — Win10 VLX WinHelp incompatibility 🆕

  • Probability: HIGH (confirmed)

  • Impact: MEDIUM (blocks PB11/VLX on Win10, does not affect source-mode)

  • Status: ✅ ACCEPTED (Mar 4, 2026) — PB11 testing abandoned on Win10

  • Discovery Date: March 2, 2026 (Bug 37)

  • Root Cause: VLX calls WinHelp() internally. Win10 removed WinHlp32.exe. When VLX executes WinHelp(), Windows opens a browser to download a non-functional WinHelp installer, creating an OS-level modal dialog that blocks AutoIT automation.

  • Impact:

    • PB11 (VLX mode) testing ABANDONED on Win10 VMs (108, 109)

    • TB11 (source-mode) is unaffected — source code doesn’t call WinHelp()

    • No fix possible without VLX recompilation (requires missing source code)

  • Mitigation:

    • ✅ Win10 VMs (108, 109) now test TB11 source-mode only

    • ✅ PB11 VLX testing continues on XP VMs (102, 103, 104) where WinHelp works

    • Future VLX recompilation from reconstructed source will remove WinHelp dependency

R22 — x64 stale acad2000 subdirectory 🆕

  • Probability: LOW (fixed)

  • Impact: MEDIUM (caused Step 2 failure on VM 109)

  • Status: ✅ FIXED (Mar 4, 2026) — Commit cecc9ef

  • Discovery Date: March 4, 2026 (Bug 38)

  • Root Cause: The src/x64/TB11-01x64/ build contained a stale acad2000/ subdirectory with 648 tracked files. ACAD registry path on VM 109 pointed to ...\ConstructiVision\acad2000 instead of root ...\ConstructiVision. Stale acad2000/csvmenu.lsp still had #| block comments (Bug 29 variant). progopts.lsp missing from acad2000 subdir.

  • Fix:

    • ✅ Registry on VM 109: Changed ACAD search path to root directory

    • ✅ acad2000/csvmenu.lsp: Replaced #| with ;;; block comments

    • ✅ acad2000/acad.lsp: Neutralized (load) calls to prevent double-loading

    • ✅ Configure-ConstructiVision.ps1: Auto-detects x64, cleans up legacy acad2000 path entries

  • Prevention: Configure-ConstructiVision.ps1 now strips acad2000 paths automatically

R23 — VLX/Source progcont routing mismatch 🆕 CRITICAL

  • Probability: HIGH (confirmed — the source code for VLX does not exist in the repo)

  • Impact: HIGH (TB11 cannot fully replicate VLX menu behavior without reconstruction)

  • Status: 🚧 ACTIVE (discovered Mar 1, 2026)

  • DFMEA Link: BRK-04 / BRK-05 in 31 — Comprehensive Workflow & Human Factors Analysis §9 (RPN 240)

  • Discovery Date: March 1, 2026

  • Root Cause: The compiled CSV.VLX was built from source code not present in this repository. In VLX mode, csv.mnu sets progcont to numeric bitmask values (1-20) and calls csv;. The VLX reads progcont and routes to distinct dialogs via a numeric-key md_dlg.dcl. In source mode, progcont is set but never read — all menu items fall through to the same dialog.

  • Impact:

    • TB11 source-mode cannot pass full menu validation until progcont routing is reconstructed

    • P0 Step 2 validation currently tests only basic CSV initialization, not full menu routing

    • Full menu routing validation (P0 Step 5+) will fail until reconstruction is done

  • Mitigation:

    • Reverse-engineer the VLX’s progcont routing from VLX behavior observations

    • Reconstruct md_dlg.dcl and md_dlg.lsp for source-mode

    • Add progcont reading to csv.lsp startup flow

    • Validate each progcont value (1-20) against golden VLX behavior

R24 — Trial account rotation abuse (App Store licensing) 🆕

  • Probability: MEDIUM

  • Impact: HIGH (revenue leakage + unfair free usage)

  • Status: ⏳ ACTIVE (added Mar 10, 2026)

  • Discovery Date: March 10, 2026

  • Root Cause: Free/trial access can be abused if users rotate Autodesk identities and repeatedly claim trial periods. Current project docs do not yet define hard entitlement enforcement behavior for trial expiration.

  • Current build scope: No licensing implementation ships in this build; this risk remains tracked as a design/next-build implementation risk.

  • Unknowns to validate:

    • Whether Autodesk App Store blocks trial reuse across related accounts/payment identities

    • Whether download/install eligibility requires an active paid AutoCAD subscription for the target product

    • How entitlement status is exposed to the app at runtime for hard-disable behavior

  • Mitigation:

    • Document and freeze the target entitlement/metering architecture in this build

    • Carry implementation work to next build where runtime licensing is in scope

    • Implement server-side entitlement checks (or Autodesk entitlement checks where supported) as source of truth, not local-only trial counters

    • Bind trial/activation to stronger identity signals where policy allows (Autodesk account + device fingerprint + timestamped entitlement records)

    • Add anti-abuse policy: one trial per organization/domain where possible

    • Design explicit degraded mode on trial expiry (read-only/locked commands) and enforce at startup + command entry points

    • Add telemetry and alerts for repeated trial creation patterns

Last Updated: 2026-03-10 (App Store-first .bundle path active; bug inventory and status live in Bug Tracker — Validation Campaign)