Assignments

Task. Find the BendSetup class. Tell me, in five lines:

Tools. Grep, file open.

Model answer. Src\Core\Bend\BendSetup.cs in Flux.Core. Dump (TextWriter) writes a per-line text representation of mounts (and clamp positions) used as the test contract. Implements INotifyPropChanged. Constructed via new BendSetup (BendTech owner) inside the bend-tooler — the parameterless constructor is for serialization.

Rubric. | Level | Criterion | |--|--| | 1 | Finds the file and one of: assembly / interfaces / construction | | 2 | All five facts correct | | 3 | Plus: explains why Dump matters for tests | | 4 | Plus: identifies the [PostLoad] migration logic and what it does |

Task. Explain in half a page how a hem bend flows from geometry to the emitted Dump line.

Approach.

Model answer.

Rubric. | Level | Criterion | |--|--| | 1 | Lists hem kinds and points at the phase split | | 2 | Connects the phase split to two BendOps | | 3 | Plus: explains the mKind decision tree | | 4 | Plus: identifies the prebend-locked edge case (BendTech.cs lines 81—​88) |

Task. Find at least three properties on BendOp whose value resolves through the Sheaf cascade. For each: which SH key, and where is the parent dictionary?

Approach. Grep BendOp.cs for Settings. and Sheaf. accesses. Cross-reference with Sheaf.cs for the SH enum.

Model answer. Examples:

Rubric. | Level | Criterion | |--|--| | 1 | Identifies one property and its SH key | | 2 | Three properties; can name the cascade direction | | 3 | Plus: identifies a property that writes into the cascade and where it writes | | 4 | Plus: contrasts with a property that bypasses the cascade — and explains why |

Task. Read BendTooler.cs lines 16—​80 and explain the auto-tooler’s pipeline in exactly 12 sentences. One sentence per phase, plus the inputs and outputs.

Approach. The pipeline is named in the comments next to the calls in DoTooling.

Model answer (skeleton). Inputs: a BendTech, plus EDoFlags and optional auxiliary setups. (1) PrepareForTooling resets state. (2) SetupBasicSequence does dead-reckoning. (3) TurnOnBumpBending switches large radii without tools to bump-bends. (4) DemoteCrashingTools downscores tools that collide in 2D. (5) UpdateToolScores favors broadly-liked tools. (6) DoToolerCore is the heart: final sequence + setups + BG. (7) Outputs: BendTech is now populated with Setups, per-BendOp tools, BG positions, and bending-aid states. (8—​12) Notes on robotic-cell handling, manual-list extraction, validity ordering, and the success/failure return path.

Rubric. | Level | Criterion | |--|--| | 1 | Names 3+ phases correctly | | 2 | Names all six phases in order | | 3 | Plus: names inputs/outputs accurately | | 4 | Plus: identifies one branch (e.g. mRobotic) that changes the pipeline |

Task. Look at commit 4a5119dcb: "Case 45728: Changing the suction cup state no longer deletes adjusted waypoints."

Answer:

Tools. git show 4a5119dcb, Grep.

Rubric. | Level | Criterion | |--|--| | 1 | Identifies the touched files | | 2 | Names the subsystem and the property setter | | 3 | Plus: identifies the test and the assertion | | 4 | Plus: proposes a defensible alternative fix and discusses trade-offs |

Task. Compare BendOpCrash.cs and NBendCrash.cs. Specifically:

Approach. Open both, look at their public methods. Search BendOpCrash and NBendCrash callers.

Model answer (sketch). Both expose check entry points called from BendOp validation. The selection is by a settings flag (look in BendOp.cs and Sheaf.cs for the relevant SH key). Outputs may differ in how they report partial failures.

Rubric. | Level | Criterion | |--|--| | 1 | Names the entry points | | 2 | Identifies the selection mechanism | | 3 | Plus: documents two output differences | | 4 | Plus: writes a recommendation: which one is the migration target, and what blocks the cutover |

Task. Pick the Bend feature bit (from Notes\Tech\2013.FluxFlavors.ftx). Find:

Tools. Grep "EFeature" or whatever the actual enum is — discover its name.

Rubric. | Level | Criterion | |--|--| | 1 | Finds the enum | | 2 | Three call sites | | 3 | Plus: classifies them (UI vs. logic) | | 4 | Plus: explains how the B23 flavor is special re. this bit |

Task. Open X:\TData\Bend\Auto\Bend_04.geo and Bend_04.txt. For each non-empty line in the .txt:

Approach. Match the Dump methods in BendSetup.cs and BendTech.cs to lines in the golden.

Rubric. | Level | Criterion | |--|--| | 1 | Maps three lines | | 2 | Maps every line in the file | | 3 | Plus: identifies one line that is conditional and explains the condition | | 4 | Plus: predicts what would change in the golden if Settings.RamVBend were halved |

Task. Find three places where Flux uses the .It singleton convention. Are any of them stateful?

Approach. Grep "\.It\b" (with caution — this matches plenty of unrelated text; refine the regex).

Rubric. | Level | Criterion | |--|--| | 1 | Finds three .Its | | 2 | Identifies the type and the use case | | 3 | Plus: confirms statelessness | | 4 | Plus: finds one .It whose statelessness is questionable and discusses it |

Task. Pick three classes with [PostLoad] methods. For each:

Approach. Grep "[PostLoad]". Pick three. Read.

Rubric. | Level | Criterion | |--|--| | 1 | Identifies three [PostLoad] methods | | 2 | Reconstructs the old shape for each | | 3 | Plus: traces the responsible commit for one of them | | 4 | Plus: predicts what would happen if the migration were removed (i.e. if old files were loaded as-is) |

Task. Trace what happens when the user clicks "Add Regrip" on a selected bend in the UI. Specifically:

Approach. Start at Src\UI\Bend\BendOpUI.cs line 22 (the AddRegrip method).

Rubric. | Level | Criterion | |--|--| | 1 | Names the glob and the method | | 2 | Traces to a model-side mutation | | 3 | Plus: identifies the render refresh trigger | | 4 | Plus: explains how the same flow behaves in the composite (multi-select) case |

Task. Pick one tour from chapter 07 you haven’t done yet. Walk it. Then write a 200-word brief that another new joiner could read instead of doing the tour themselves.

Goal. This exercise tests teaching, not just understanding.

Rubric. | Level | Criterion | |--|--| | 1 | Brief is correct in fact | | 2 | Brief is teachable: a peer can follow it | | 3 | Plus: brief includes one "watch out for" | | 4 | Plus: brief is short enough to read in 90 seconds |

The original concept proposes a six-month pacing. Here is one way to phase the exercises.

By the end of month 6, you’ve earned the right to a small first PR — typically a test addition or a documentation fix. The reading muscles you’ve trained are the foundation for everything you’ll do thereafter.

After working through the exercises, you should be able to fill in this matrix yourself:

Last updated: 2026-05-04. Exercises are pinned to current code paths; if a file moves, update or retire the exercise.