Bootstrap

Install dependencies : npm ci

Pre-requisites / Configuration

Create a members.json file in root directory (you have a members.sample.json sample file, but this file could also be generated/exported from an in-house spreadsheet).

Create a community-descriptor.json file in root directory (you have a community-descriptor.sample.json sample file)

Download a sound which will be played whenever a new result beating previous score will be found :

wget https://assets.mixkit.co/active_storage/sfx/2848/2848.wav -O mixkit-gaming-lock-2848.wav

Finding best community groups

Run npm run find-best-community-groups compute <Track name> (example: npm run find-best-community-groups compute "Sessions libres") to look for the best spreading of members across different groups declared in <Track name>.

The algorithm is pretty simple: it will make a shuffling or track subscribers across different track's declared groups, and will give a "score" (based on several criteria) to the configuration (lower scores are better).

Given that this is totally random, the more you spend time on this, and the best results you will get (keep in mind that a score < 0.2 should be nice already).

Whenever a better score is found, a sound will be played, and the Track's group filling will be stored into best-result.json file.

Repeat this for every configured Tracks, so that you get a best-result.json file covering all these Tracks. To speed up things for tracks with single groups (such as "Focus" tracks), you can run npm run find-best-community-groups compute-single-groups which is going to run compute on each single-group tracks to have at least 1 execution (and a corresponding best-result.json entry for these tracks where shuffling is useless)

Showing every Tracks' best results

Run npm run find-best-community-groups show to list every Tracks and spreading of members (+ stats) across tracks' groups.

This should give you an overview of the upcoming cycle configuration to check if everything looks good.

Updating members.json groups

Once the cycle is started and you communicated groups assignments to everyone, you can "enrich" members.json file with the new groups assignments coming from best-results.json file, so that these group will be taken into consideration for the same path constraint during your next cycle.

To do this, simply execute npm run find-best-community-groups record-member-groups and best-results.json file will be updated.

As a side note, if you configured people in the absentsFromThisCycle property, absent people will have their latest group valued to "".