[Patriciaath999]

A report reopened from Past attempts always shows "(final code was not saved)", on the page and in its .md download, even for the attempt just finished. saveHistory never put the editor contents in the history entry, and normalizeProgressEntry would have dropped them before the lobby saw them.

• saveHistory adds code to the entry: the editor contents, or null at a whiteboard, which has no code.
• The short history (codetrial_history, newest 20) keeps it. boundedReviews strips it, so the review store is byte for byte what it was and its 164 KB budget holds as many older attempts as before. Attempts older than the newest 20 keep the placeholder.
• When the quota refuses the short history, or the review store write that follows it, saveLocalReport writes the history again with the older rows' code left out, then without this report's own, retrying the review store each time, so the attempt just finished keeps its code longest and the code never costs an attempt the device would otherwise keep. When no step lets the review store in, the refusal was not the code's, and the history goes back to the step it first landed at.
• saveHistory adds the code after withUnclearLines has bounded the unclear lines, since that bound is the account body's and the account copy never carries the code, so a long solution does not cost the attempt its unclear lines.
• The /api/reports body leaves code out, since the server stores that body as sent. A signed-in lobby fills it in by report id from this device's history (withLocalCode in renderServerHistory), so the code shows wherever the device has it, without changing the server body.
• normalizeProgressEntry carries code as a string or null, and the reopened report and its download use it, falling back to the placeholder when it is null.
• README.md says the browser keeps the final code of up to its 20 newest attempts in local storage, that the account copy never carries it, and that deleting an attempt removes it.

Validation on Windows 11, Node, Playwright Chromium, rebased on main at 6b739c5:

• New tests: a saved entry carries the code, a whiteboard entry does not, and a long code keeps every unclear line (interview-lifecycle.test.js); the short history keeps it while the review store and the account request body are identical to a save without code, a stored __proto__ field survives the copy, a quota-limited storage still saves every attempt, keeps each new report's own code and drops the oldest code first, a review store that would not fit beside a history full of code still keeps every attempt, a review store that refuses every write leaves each attempt its code, and account rows take the code this device kept by report id (account.test.js); normalizeProgressEntry keeps only a string (progress.test.js); device and account attempts reopen with the code this device kept, on the page and in the download, and one without code still shows the placeholder (lobby.test.js). Each was seen to fail with the behavior it covers broken, twenty-eight breaks in all, and pass once restored.
• node --test tests/browser/*.test.js: 112 failures on this branch and the same 112 by name on main at 6b739c5, from vendored assets not fetched on this machine and from the lobby fits on one screen instead of a wall of problems, which measures 802px against 800px here on both. The branch adds 10 tests, all passing.
• ESLint and Prettier pass on the changed files; README.md's only Prettier difference is in its tables, as on main.
• Not run here: scripts/test.sh as a whole, including the Rust lanes, which need a toolchain this machine does not have. No Rust file changed.

Closes #221

[jserv]

Leaving code out of the account body is consistent with what the description says, but it means a signed-in candidate sees the placeholder even on the device that holds the code: renderServerHistory never looks at local storage. A follow-up could fill in code from readLocalHistory() by id before showProgress. That shows the code wherever the device has it, without touching the server body or the open question on #221.

[Patriciaath999]

Included here rather than as a follow-up: renderServerHistory passes the account rows through withLocalCode (web/history.js), which fills code in by report id from this device's history. The server body is unchanged, and the lobby test covers a signed-in candidate.

[jserv]

Avoid ambiguous Refs #221.
The goal should be "Close #221."

[jserv]

Before asking someone to review, you MUST resolve any regressions flagged by GitHub Actions.

[jserv]

Rebase latest main branch and explain how the CI result changes.

The gates (browser) failure here is Strict delayed judge language restoration respects end in tests/browser/strict-selection.test.js. This branch did not break it: the test was flaky on main itself. After End it waited for the ending overlay, but its room has no interviewer, so the offline summary replaced that overlay a moment later and a loaded runner missed the window. #461 (merge b0059b2) fixed the test by pausing a fake clock before End, and main passes the browser lane again at that commit. This branch is based on a commit before b0059b2, so it still runs the old test. After rebasing, reply here with the new CI result, and if anything still fails, say whether it comes from this branch or also fails on main.

The same lane also fails the lobby fits on one screen instead of a wall of problems, which passes on main, so that one comes from this branch and a rebase alone will not clear it.

[jserv]

Rebase latest main branch and resolve conflicts.

[jserv]

Add screenshots to the pull request description to illustrate your work, as @ColtenOuO suggests. Example: #349

[jserv]

Thank @Patriciaath999 for contributing!