This repository is a combination of applications that interact to
- conduct surveys,
- collect them in a database
- and visualize the results.
flowchart TB
web["web app (wasm)"] -- REST --> backend
native["native app"] -- REST --> backend
subgraph backend [prono-backend]
rest["REST endpoints"] --> core["prono: surveys, submit / results / scores, Repo"]
end
cli["prono-cli"] --> core
core -- sqlx --> db[(MariaDB)]
api["prono_api: survey model and definition, answers, rules, REST contract"]
web & native & backend & cli -.-> api
| Crate | Role |
|---|---|
api (prono_api) |
The model every binary shares: Survey and how it is defined (Survey::from_json), Answer, the validation rules and the REST contract. Compiles to wasm. api/surveys/*.json are example definitions, not part of any binary. |
prono |
Server-side core: the use cases (upload_survey, submit, results, scores), the Repo trait with its MySQL and in-memory implementations, and the database config. |
backend |
actix-web REST server around prono. |
app |
egui app for desktop and web. Both talk to the backend over REST; the UI only draws a small state machine (app/src/state.rs). |
cli |
Admin tool on top of prono: a menu to navigate, or commands for scripts. |
There is one definition of each survey type (in prono_api), and the rules
live there too, so the app checks a submission with exactly the code the
backend uses to accept it.
The diagram in
docs/architecture.drawio.pngpredates this layout.
Surveys live in the database, not in the binaries: the backend serves them to
the apps, so a new one needs no redeploy. Write a definition and upload it
(it replaces the survey with the same survey_id):
cargo run --bin prono-cli -- surveys upload my_survey.json{
"survey_id": 3,
"title": "My survey",
"description": "What the survey is about",
"link": "https://example.com",
"questions": [
{ "question": "When will it happen?", "answer_type": "prediction_date" },
{ "question": "Why?", "answer_type": "text" },
{ "question": "Which?", "answer_type": "poll", "possible_answers": ["A", "B"] },
{ "question": "Any allergies?", "answer_type": "poll", "possible_answers": ["No"], "other_label": "Yes, namely:" }
]
}-
Pass phrase: every survey needs one, so that nobody sees or fills in a survey unasked. It is not part of the definition (there is no
accesskey): set it withprono-cli surveys pass-phrase <survey id> <pass phrase>. A participant enters one pass phrase in the app and gets every survey that has it, so surveys for the same group can share one. Until a survey has a pass phrase it stays closed. -
name_listis optional (defaultfalse). With"name_list": trueonly the participants on the survey's name list may answer it, on top of the pass phrase, which is always needed: put names on it withprono-cli surveys names allow <survey id> <first name> <last name>(names are matched ignoring case). A name that is not on the list is refused as soon as it clicks the survey, so it never sees the questions (and again when answers are submitted). -
private_resultsis optional (defaultfalse). With"private_results": truethe results screen shows a participant only their own answers, not everybody's. The backend enforces it: it recognises the participant by the device id of their submission (so only on the device that answered), and anyone else gets an empty result. Not for scored surveys, whose ranking shows everybody. -
linkis optional;possible_answersis forpollquestions only. So isother_label, the name of the free answer next to the options (default:Other), for questions where that does not fit. -
answer_optional(defaultfalse), fortextandpollquestions only: with"answer_optional": truethe participant may leave the answer blank and it is not stored; the app shows "(optional)" next to the question. -
multiline(defaultfalse), fortextquestions only: with"multiline": truethe participant gets a bigger input field (several rows high) in which long answers wrap. -
Scores: give every question a
correct_answer(text or option fortextandpoll,{"month": 7, "year": 2028}forprediction_date) and the survey is scored: participants see a ranking of the correct answers after submitting. All questions need one, or none. The correct answers never leave the backend, so a prediction survey can be uploaded again with them once the events happened; the answers already given are kept.prono-cli surveys scores <id>shows the ranking. -
Answers:
prono-cli surveys answers <id>prints a table of everybody's answers (a row per participant, a column per question); add--user "First Last"for the answers of one participant. The interactive menu (prono-cliwithout a command) offers both under Surveys.
A question's id is derived from its text and answer type, and stored answers reference that id: editing a question's text starts it from scratch.
This repository contains the following applications: