Add feedback timing, attempt tracking, and answer review features - #22
Merged
Merged
Conversation
An audit of challenge mode found four ways a student could reach a full
score without reading a notehead, and four defects where the app described
one round and ran another. This closes the ones that are fixes.
Refuse two link shapes that produced evidence for an assignment nobody set:
* A full-measure round never repeats a measure, so a pool of k rhythms can
fill at most k^4 questions. `?scope=measure&cells=quarter,eighths&n=20`
passed validation and then threw while the round was being built — after
the banner, level, scope and pass mark had applied. The student answered
the default five-question beat round under the assignment's stated
conditions, and the card reported its goal. Now refused at the link, and
the round is built before any of it is applied, so the shape of that
failure is gone rather than merely unreachable.
* `pass` was checked against MAX_QUESTIONS rather than the round's real
length, so `?a=Goal&pass=8` printed a goal of 8/5 that no student could
clear.
Stop a posted answer key from transferring. A seed makes every student's
questions identical, which is what makes the scores comparable — and it also
fixed where the correct answer sat, so one student could post "4, 4, 2, 3, 4"
for the class. Questions still come from the seed alone; an optional variant
reshuffles the choices afterwards, from a typed name or a per-browser string
that never leaves the device. A round with no variant is byte-identical to
one generated before this existed.
Make a replayed round say so. Retry replays the same questions in the same
order after showing every answer, and the finish time was stamped once and
never re-stamped, so two attempts could carry one verification code. The
attempt is now counted, re-stamped, and reported on the card, in the copied
summary and in both the code and the attempt reference.
Add `fb=end`, which withholds the correct answer, the running score and the
guide's highlighting until the round is over — instant feedback teaches, and
a graded round cannot hand over an answer key while it is being answered.
Close Practice while an assigned round is unfinished, since it reveals the
counts for the same vocabulary.
Show the diagnosis the app already computed. Every completed round has built
an errorSummary since the result envelope was adopted and none of it reached
a human, so a card said "2 of 5" where it could name the rhythms that went
wrong. The card now lists them, the summary carries them as catalog ids for a
follow-up link, and every round ends with a review of each question beside
the count it wanted — open by default when feedback was held.
Lock only what the link pinned. `?a=Homework 1` froze the level and question
size at their defaults and told the student their teacher had chosen them;
the parser already reported exactly what was pinned and the UI recomputed it.
The attempt number is deliberately NOT added to the result envelope's own
fields: praxis.result.v0_1 is shared with two sibling apps, and adding a
field here would make this app emit a shape the others do not. It belongs in
the schema, which is a decision for the family rather than for this app.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013jGA1JrHcFtpHu711uq5DN
The size note carried 2655 KiB / 995 KiB gzipped. The commit before this release actually measures 1000.88 KiB, so the figure had drifted and nobody would have known until a deploy failed with an unrelated-looking build error — the exact failure mode the deployment section exists to prevent. Records the measured numbers, the delta this release added, and the command that produces them. Also flags that the note's own arithmetic implied a ceiling near 1000 KiB while naming a 1 MiB limit; which is right decides whether there is 20 KiB of headroom or none, and only the dashboard knows. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jGA1JrHcFtpHu711uq5DN
Deploying with
|
| Status | Name | Latest Commit | Updated (UTC) |
|---|---|---|---|
| ✅ Deployment successful! View logs |
count-it | 930f321 | Aug 08 2026, 08:17 PM |
10 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
This release adds three interconnected features to support both teaching and assessment workflows:
Feedback timing control (
fbparameter): Teachers can now choose whether the correct answer appears after each question (each, the default) or only at the end (end). When feedback is held, the running score and guide highlighting are also hidden until the round completes, supporting graded assessments where the answer key cannot be shown mid-round.Per-student answer order variation (device variant + student ID): A stable per-browser identifier varies the position of answer choices so that one student's answers cannot be directly copied by classmates. A typed student name supersedes the device identifier, allowing teachers to reproduce a particular student's answer ordering. The variant is determined before questions are built so it affects choice order only, not the prompts or distractors.
Attempt tracking and review: The app now tracks which run of a round a student is on (incremented by retry, reset by new rounds). Results display the attempt number, and a collapsible review section shows every question with what the student chose versus the correct answer. The review opens automatically for held-feedback rounds (where it is the only place answers are revealed) and stays folded for instant-feedback rounds.
Supporting changes:
assignmentLocks()to report only what the link actually pinned, not what an assignment implies (e.g., naming a round no longer falsely locks level/scope at defaults)Validation
Documentation standard
2026-08-08.5, feedback timing documented in assignment link section, measure-pool bound explained2026-08-08.5)fbparameter and new error codeshttps://claude.ai/code/session_013jGA1JrHcFtpHu711uq5DN