[QTI] Older Kolibri can't render exercises of only legacy items published from unstable

Maintainers usually reply within 1 day

Assessment

Research direction

Description

❌ This issue is not open for contribution. Visit Contributing guidelines to learn about the contributing process and how to find suitable issues.

Overview

On unstable, publishing an exercise whose questions are all legacy items produces a QTI package and no Perseus archive. Older Kolibri renders only Perseus, so it cannot render these exercises. Release v2026.09.24 publishes the same exercises as a Perseus archive.

Complexity: Low
Target branch: unstable

Context

• Publish sends every exercise without a perseus_question item to QTIExerciseGenerator, and adds PerseusExerciseGenerator only when _node_is_perseus_derivable holds: https://github.com/learningequality/studio/blob/f4360ff91b7a2254763c6a58d50f3245559e15a0/contentcuration/contentcuration/utils/publish.py#L404-L415
• _node_is_perseus_derivable returns False for an exercise with no type=QTI item: https://github.com/learningequality/studio/blob/f4360ff91b7a2254763c6a58d50f3245559e15a0/contentcuration/contentcuration/utils/publish.py#L264-L274
• v2026.09.24 sends only exercises with a free_response item to QTI: https://github.com/learningequality/studio/blob/fc30d8311037a2c1629a50a6dd58f5e8b7f7df34/contentcuration/contentcuration/utils/publish.py#L369-L377

The Change

• Legacy-only exercises should get the Perseus archive #6001 gives exercises of native QTI items, under the same rule: QTI only when any item has no Perseus equivalent.

How to Get There

• Publish a channel with an exercise holding only multiple_selection and input_question items.
• Check the exercise node's files: there is a qti-preset file and no exercise-preset file.

Acceptance Criteria

• Publishing an exercise whose items are all single_selection, multiple_selection, true_false or input_question produces both a qti and an exercise file.
• Publishing an exercise with a free_response item produces a qti file only.
• Tests cover both cases.

AI usage

Claude Code found this while reviewing the #6211 branch, reproduced it with a scratch publish test on unstable at f4360ff91, and drafted this issue.

Getting set up

First steps

1. Read the whole issue, then the project's contributing guide.
2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
3. Fork the repository and make your change on a branch.
4. Open a pull request that references the issue number.

More from learningequality/studio

• Set unpublishable: true on PublishedChange by construction to simplify handleMaxRevs predicate

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  DEV: frontend P3 - low

  Difficulty 2/5 1-3 hours Newbie friendliness 74/100

  learningequality/studio#5868 · Opened Apr 29, 2026

  Maintainers usually reply within 1 day

• Change ricecooker version checks to hard warning only

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  TAG: tech update / debt

  Difficulty 2/5 1-3 hours Newbie friendliness 62/100

  learningequality/studio#2245 · Opened Sep 10, 2020

  Maintainers usually reply within 1 day

• [RTE] Toolbar Copy should copy the selection's serialized content, not its DOM

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  DEV: frontend

  Difficulty 3/5 1-2 days Newbie friendliness 45/100

  learningequality/studio#6296 · Opened Oct 6, 2026

  Maintainers usually reply within 1 day

• Rich text editor format button hides the current format from screen readers

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  DEV: frontend P1 - important TAG: a11y

  Difficulty 2/5 1-3 hours Newbie friendliness 28/100

  learningequality/studio#6292 · Opened Oct 6, 2026

  Maintainers usually reply within 1 day

• [QTI] Editing a question keeps the item-level content the editor doesn't write

  Possibly taken @rtibblesbot claimed this 1 day ago. Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  bug

  Difficulty 3/5 1-2 days Newbie friendliness 55/100

  learningequality/studio#6290 · 1 reaction · 1 assignee · Opened Oct 5, 2026

  Maintainers usually reply within 1 day

All issues in learningequality/studio

Similar issues

• Bootstrap: the last two steps run as the user who owns the install

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel
  area/install reliability

  Difficulty 2/5 1-3 hours Newbie friendliness 75/100

  inferstep/ATLAS#412 · Opened Oct 7, 2026

  Maintainers usually reply within 1 day

• test_record_bounds errors when TMPDIR is long, since its socket directory honours TMPDIR

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 2/5 1-3 hours Newbie friendliness 83/100

  FluidNumerics/fluid-walk-blocker#191 · Opened Oct 7, 2026

  Maintainers usually reply within 1 day

• [Bug Report] Qwen3.5 and Qwen3Next adapters don't set `rmsnorm_uses_offset`, so edited or backward-hooked norms drop the `1 +` in `(1 + weight)`

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 2/5 1-3 hours Newbie friendliness 62/100

  TransformerLensOrg/TransformerLens#1868 · Opened Oct 7, 2026

  Maintainers usually reply within 1 day

• claude-task-notifier-submit: test_never_staged_returns_fast_instead_of_waiting_the_full_timeout flakes at its 4 s bound under CI load

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 2/5 1-3 hours Newbie friendliness 76/100

  sonichi/sutando#5219 · Opened Oct 7, 2026

  Maintainers usually reply within 1 day

• Description of soilw_am in forcing.py says it is dimensional, but it should be dimensionless according to speedy_surface_flux.py

  Open

  ⋯Report a problem with this issue

  • Is something wrong?

  • Already taken
  • Not beginner-friendly
  • Stale or invalid
  • Something else…
  • Send Cancel

  Difficulty 1/5 Under an hour Newbie friendliness 85/100

  climate-analytics-lab/jax-gcm#1057 · Opened Oct 7, 2026

  Maintainers usually reply within 1 day

More Python issues