fix(symbolic-ai): add Console.WriteLine/prints to silent stubs + re-execute Linq2Z3 and Planners-3 (2 BETA->PRODUCTION)

[jsboige]

Summary

Convention #1946: silent code cells receive informative print output so every cell produces visible output.

Changes

Linq2Z3.ipynb (C# .NET):

• Added Console.WriteLine("Classe CanibalsAndMissionaries definie avec succes") after class definition (was silent)
• Added Console.WriteLine("Exercice a completer : planification de taches avec Z3.Linq") to exercise stub (was // TODO only)
• Re-executed with Papermill .net-csharp kernel: 7 code cells, 0 errors, 0 without output

Planners-3-State-Space.ipynb (Python):

• Added print("Exercice a completer : grille 5x5 avec obstacles, BFS vs A*") to Exercise 1 stub (was comment-only)
• Added print("Exercice a completer : heuristique euclidienne et verification d'admissibilite") to Exercise 2 stub (had pass returning nothing)
• Re-executed with Papermill python3 kernel: 19 code cells, 0 errors, 0 without output

COURSE_CATALOG.generated.json: regenerated, both notebooks now PRODUCTION

Validation

Linq2Z3: 7 code cells, 0 errors, 0 without outputs -> PRODUCTION
Planners-3: 19 code cells, 0 errors, 0 without outputs -> PRODUCTION

Papermill SUCCESS on both notebooks. No NotImplementedError, no raise, no assert False (C.1 compliant).

Related

• Epic Epic: SymbolicAI notebooks -> PRODUCTION (11 BETA residuels) #1947 (SymbolicAI notebooks -> PRODUCTION)
• Convention docs(rules): code cells emit informative output (anti false-positive maturity) #1946 (print for silent cells)

Co-Authored-By: Claude Opus 4.8 noreply@anthropic.com

[github-actions]

Notebook PR Validation: PASS

• Notebooks checked: 2
• Code cells validated: 25
• Result: All passed

Checks: H.1 (no errors), H.3 (execution_count), C.1 (no banned patterns)
Non-Python kernels (.NET/Lean): C.1 + errors only (execution_count advisory)
QuantConnect notebooks: C.1 + errors only (require QC Cloud for execution)

[jsboige]

[NanoClaw]

Reviewed: 3 files (757a/366d)

Summary

This PR re-executes two BETA notebooks via Papermill and promotes them to PRODUCTION. The core work is legitimate: actual Z3 solver outputs replaced the previous versions, silent stubs got print statements, and CS1701 warnings were cleaned up. A few items worth noting.

---

1. COURSE_CATALOG.generated.json (+6/-6)

Two BETA->PRODUCTION promotions with updated cell output counts:

• Linq2Z3: cells_with_outputs 5->6, cells_without_outputs 1->0. Correct: one previously empty exercise cell now has a Console.WriteLine stub.
• Planners-3: cells_with_outputs 17->19, cells_without_outputs 1->0. Correct: two exercise cells now have print stubs.

Consistent with the notebook changes. No issues.

2. Linq2Z3.ipynb (+495/-122) -- MAJOR CHANGE

Re-execution: verified

• Papermill timestamps: 2026-05-31T19:24:16 to 19:25:47 (total ~91s). All 7 code cells have sequential execution_count (1-7), no nulls.
• Papermill metadata with input/output paths uses relative repo paths only. Clean.

Outputs: real Z3 solver results

Verified the Z3 constraint-solving outputs against the known Missionaries & Cannibales problem:

Cell	What	Check
1	NuGet install	Now shows "Installed Z3.Linq, 2.0.1" (was just "Installing")
2	Solve() basic	X1=3,X2=0,X3=1,X4=0,X5=1 -- same as before, output structure unchanged
3	Class definition	Was silent, now has Console.WriteLine("Classe definie avec succes"). The stub fix from the PR title.
4	Solve() 3M3C	Length=22, valid non-optimal solution. Timing updated (0.23s -> 0.71s) -- slower run, consistent with a fresh Papermill execution.
5	Optimize() 3M3C	Length=12 (11 crossings) -- verified step-by-step, matches the known optimal solution for 3M3C/boat=2. Correct.
6	Optimize() 30M30C	Length=28. Timing 30s->80s. Consistent with heavier compute on re-execution.
7	Exercise stub	Was empty output, now prints placeholder text. The second stub fix.

Warnings cleanup

All CS1701 assembly version mismatch warnings (System.Linq.Expressions 8.0.0 vs 10.0.0) were removed from cells 2-6. This is a positive cleanup -- these warnings cluttered the outputs without adding pedagogical value. Likely the .NET runtime or Z3.Linq version was updated between runs.

Comment accent stripping

Comments had French accents stripped: "determinee" instead of "determinee", "recupere" instead of "recuperer", etc. This looks like an encoding issue in the re-execution environment (likely missing UTF-8 locale). Low severity -- the markdown cells (visible to students) appear untouched, only code comments are affected.

Leaked local IP addresses -- WARNING

Cell 1 output contains a dotnet-interactive HTTP probe script with internal IPs:

• http://192.168.0.20:2048/
• http://172.27.208.1:2048/
• http://172.28.176.1:2048/

This is auto-generated by Polyglot Notebooks / dotnet-interactive when rendering in Jupyter. It reveals the internal network topology of the execution machine. Not a credential leak, but it does expose private network ranges. Worth considering whether to sanitize these from notebook outputs in the Papermill pipeline.

3. Planners-3-State-Space.ipynb (+256/-238) -- OUTPUT REFRESH

Nature of changes

• 356 of 494 changed lines are Papermill timestamp updates (2026-05-19 -> 2026-05-31). This is the dominant change.
• 0 source code lines changed. The notebook logic is identical.
• 1 image/png output re-generated -- the matplotlib graph was re-rendered on the new execution environment. Different base64 (different random state or rendering backend).
• 1 networkx version change: 3.6.1 -> 3.5 (downgrade -- different Papermill environment has an older networkx).
• 2 print stubs added to exercise cells (same pattern as Linq2Z3):
  • Cell 38: print("Exercice a completer : grille 5x5 avec obstacles, BFS vs A*")
  • Cell 40: print("Exercice a completer : heuristique euclidienne et verification d'admissibilite")

No fake outputs

All outputs are either timestamp updates, a re-rendered graph, a version change, or the new stub prints. No N/A, placeholder, or fabricated data.

---

Verdict

LGTM with two minor notes:

1. Accent stripping in Linq2Z3 code comments -- cosmetic but worth fixing if the locale can be set to UTF-8 in Papermill.
2. Leaked local IPs in the dotnet-interactive probe script -- consider adding a post-processing step to strip probeAddresses blocks from outputs before committing.

[clusterManager-Myia]

[Hermes] — COMMENT_WITH_CONCERNS

Security scan: CLEAN (PublicKeyToken refs sont des warnings compilateur .NET, pas des secrets).

Concern complementaire au review NanoClaw :

1. Pattern recurrent de leak IPs locales — Linq2Z3.ipynb cell 1 contient 192.168.0.20:2048, 172.27.208.1:2048, 172.28.176.1:2048 (probes dotnet-interactive). Meme pattern identifie dans #1948, #1950 et maintenant #1953. Ce n'est pas un credential leak mais cela expose la topologie reseau a chaque execution. Recommande ajout d'un post-processing Papermill pour striper les blocs probeAddresses des outputs.

[jsboige]

Doublon de collision SymbolicAI (1 editeur/serie). Linq2Z3 couvert par #1948 (APPROVED, merge ce cycle), Planners-3 couvert par le master #1952. Aucune perte : le travail est preserve dans #1948 + #1952. Ferme pour deconfliction. Merci po-2025:CoursIA-2 — partition rafraichie sur le dashboard workspace.