Eve library

Chapter 1 · Tara: practice and continuation · brief · 10 September 2026

Guide

Revision 1

On this page

The first chapter of the Tara · find a practice track, rewritten in the Eve edition on 10 September 2026. It follows the plan in EVE-V1-REDESIGN-TODOS.md and the track brief in tara-find-a-practice-track.md. The receipt is under verification/.

The job#

Tara is the room a member sits down in. What makes it worth a chapter of its own is that almost nothing about a recommendation is left to prose: moods, themes, modalities, context tags, lineages and durations are validated lists the product reads at runtime, each with rules its schema enforces. The persistent failure the chapter teaches against is reading one of those exact values as an answer to a question it was never asked.

Running example. Eleven minutes, and a program that is paused. The member asks for something deeper, and the answer has to be assembled from four separate constraints — fit, time, access and continuity — none of which can vouch for the other three. The one that fails is continuity, and the chapter ends on what the honest answer costs.

Three learning promises (the cover)#

  1. Eligibility is four questions — fit, time, access and continuity, each with its own data and its own way of failing.
  2. The vocabulary is data with rules — moods, modalities, lineages and durations are runtime constants whose consistency is testable.
  3. A saved progression is not a permission — the record can claim readiness the contract never recomputes, so the consumer decides.

Three things to remember (the close)#

  1. A valid value proves its schema. Rate, cue type, download affordance and companion reference each speak only for themselves.
  2. A count is not a position. Completions plus one is a statement about the past; the offer is a statement about now.
  3. A readiness flag is a claim. The blocked branch is checked against the prerequisites; the satisfied branch is not.

Bridge. The catalog those recommendations come from — chapter two.

Sections#

Three sections of four, opened by a section-divider that names all three.

Section Slides
What makes a practice eligible the release map, the four constraints, the journey, the glossary
The vocabulary, and what its values prove the taxonomy rules, the duration bands, the playback policy, the gap after it
Coming back the continuation record, the room's own offer, the worked return, companions

Treatment decisions#

  • The duration table stayed a table, and became quantitative. The plan offered a card grid, a stat panel or a compare panel, and none of them holds five bands across four exact numbers each. Declaring the table quantitative buys the caption, the typed cells and — the reason it matters here — an explicit missing state for the retreat band, whose maximum is null in the contract and is now marked open rather than printed as a number.
  • tara-to-other-rooms became a graph diagram, not a step journey. The inherited layout was a flow, and a flow is what was wrong with it: Arete, Nisaba and Nyx are alternatives after a practice, not stages of one. Drawing them as peers from a single practice also makes room for the two rooms the release defers, which are on the slide with no edge at all.
  • tara-failure became a compare panel rather than a card grid. Its four rows already asked one question twice — what the declared value establishes, and what still has to be observed — so two columns answered for every row says it better than four independent cards, and the verdict line is the slide's actual claim.
  • The glossary defines only words section one has already used. Six terms, each pointing back at the slide that first uses it rather than forward at one that will. Companion lost its card to eligible, the chapter's own word, and is defined in prose where it appears.
  • The capture is the room's own offer. See below.

What the capture had to be#

The chapter needed a real surface, and the one that carries its argument is the top of the Tara room. Rendered over the BFF builder's output for two completed sittings — deliberately not the first two — the card shows four numbers, and two of them answer different questions. The sitting offered is the second on the path, because it is the first one not finished. The session number beside it is completions plus one, which is three, and the progress rail follows the number rather than the sitting. Above both, the kicker reads twelve minutes over a five-minute practice, because the kicker is a literal in the component.

None of that is fabricated and none of it is a bug in the builder taken alone. It is what the chapter means by a constraint having its own data, shown on the one card where all four are printed as if they were one statement.

What the contract test established#

tools/presentations/tests/pinned-tara-practice.test.mjs executes the claims this chapter makes, at the same pin the slides cite:

  • The five duration bands, their cadences, default sessions and minimum gaps.
  • That the bands are contiguous over whole minutes and nothing else: 2.5, 10.5, 25.5 and 60.5 minutes have no bucket, and the resolver raises rather than returning one. durationMinutes is z.number().positive() with no .int(), so such a value is field-valid and reaches the refinement — where the throw escapes safeParse instead of becoming a validation issue. The check meant to catch a mismatched bucket is the one that breaks.
  • That a blocked recommendation must name real unsatisfied prerequisites, and that a satisfied one is checked only for an empty blocked list — so a state may claim readiness while holding an unmet prerequisite, and the schema admits it.
  • That the program arc bounds each count against the total and not against the other, which is why a later session can be finished while an earlier one is the one in progress.
  • The playback policy: nine admitted rates, the snap, the clamp, and the non-finite refusal that clamping does not rescue.