Changelog • brainwritR

brainwritR 0.6.0

Configure any positive number of questions separately for each topic. Answer fields, drafts, submissions, YAML, prompt exclusion and exports follow that count; existing two-question settings and databases remain supported.
Show the authored question above each answer in results and plenum views, and list the questions in the analysis tab.
Apply the pseudonymization checkbox immediately to the result view as well as downloads, without changing stored participant identities.
Fix weighted plots and PDF generation on ggplot2 3.5.x, where facet_wrap() does not accept space. Keep download handlers active and send explicit MIME types for PDF, XLSX, RDS, CSV, Markdown and YAML files.
Add server and real-browser regressions for variable questions, immediate pseudonyms, weighted plots and downloaded binary file contents.

Add Analyse & Plenum to finished sessions: participants see all contributions anonymously and distribute 100 % per topic with sliders, their own contributions included. On a shared hot-seat device the weighting passes from person to person.
The moderator starts, ends and reopens the weighting, follows progress per topic and then sees a ranked overview with a chart; results stay hidden while voting is open.
Weights appear in a new CSV download, an extra XLSX sheet, the RDS snapshot, the Markdown protocol and a third PDF page. For sessions without weights, CSV, XLSX, Markdown and PDF are unchanged; the RDS snapshot always carries the votes table and the new plenum and plenum_turn session columns.
Ask for the planned number of participants in setup and describe the resulting format live, for example “Format 16-2-5 … 5–6 people per group”; the lobby counts joins against it. Names in setup are marked optional; pseudonyms are welcome.
Replace the footer line with a CTTIR link and an inline GitHub icon.
Add a light/dark toggle next to the language control. It switches in place without touching drafts, is remembered per browser and follows the device setting by default.
Migrate session files created by 0.4.0 at startup.

Follow-up to an external audit of this release:

Ending a round only ends the round on screen: repeated clicks or a second moderator tab no longer skip a round in individual and group-device mode.
Weighting sliders are named by their question and contribution and announce the remaining budget; their values are spoken as percentages.
Charts adapt to phones: rank-numbered weighting chart sized to its bars, single-column terms, smaller network and wordcloud, horizontal continuity bars, larger text.
One rule decides what counts as a contribution: whitespace-only text (tabs, line breaks, no-break, ideographic and zero-width characters included) is ignored by analytics, weighting, protocols and the previous-contribution view.
Open voting pages of the same person follow weights stored from another tab or device.
Moderator logins also expire in open tabs after 12 hours.
Optional spreadsheet-safe CSV downloads that defuse formula-like cells, also after embedded separators and line breaks.
Publish the website through GitHub Pages actions, matching the repository setting.
Open SQLite files with synchronous = NORMAL (SQLite’s durable setting for WAL) after the busy timeout, instead of RSQLite’s default OFF, so recently saved contributions survive an operating-system crash or power loss.

Run several sessions side by side, each with its own address (?s=<code>) and QR code. The bare address keeps serving the standard session; existing databases need no migration.
Add a moderator session overview to create, open, restart, archive, restore and delete sessions, with large downloadable QR codes.
Restarting a session offers its settings for review in a new session and keeps the original results. Archived sessions refuse participants and stay available for review and export.
Keep one participant identity per session on each device.
Keep moderators signed in per browser tab and pause new logins after repeated wrong PINs.
Rewrite the example question sets after the 6-3-5 principle: one open problem per topic, up to three new ideas, then develop an idea already on the sheet.
Draw participant IDs from the operating system’s secure generator. Drawing a wordcloud no longer resets R’s random seed, which had made later IDs and group shuffles predictable and could make a join fail on a duplicate ID.
Never recreate a deleted session file from a lingering page; open pages reload to a notice.

Add an in-place German/English/French language slider without interrupting drafts.
Offer reversible, editable example questions in the selected interface language.
Provide an English screenshot gallery and an entirely English classroom vignette.
Add a multilingual descriptive analytics view with five shared ggplot figures.
Count Unicode German terms with transparent stopword and prompt-vocabulary filters.
Summarize adjacent-round lexical overlap without treating it as an idea-quality score.
Compose deterministic two-page A4 reports directly with Cairo and patchwork.
Add lossless RDS and styled four-sheet XLSX exports alongside CSV and Markdown.
Apply optional stable author pseudonyms consistently across all five downloads.

Preserve the default individual-device workflow and add optional roster buttons.
Add canonical group-device identities and sequential hot-seat handovers.
Prepare sessions using the setup form or validated, safely parsed YAML files.
Migrate existing databases in place, retaining sessions and contributions.
Persist settings in SQLite and support sparse hot-seat groups with an empty sheet.
Extend lifecycle, browser, migration and configuration regression coverage.

Initial classroom brainwriting application with German participant and moderator views.
SQLite persistence, frozen sheet counts, balanced groups and late joining.
Server-authoritative rounds, reconnect/resume and debounced draft saving.
Server-side moderator authorization and transactional lifecycle operations.
CSV and Markdown export, container deployment and classroom guide.