vingkan / epimetheus Goto Github PK

• Consider cost in time, bandwidth, and size of request.
• Check if memoization is helping.

Provide the Firebase credentials for the Prometheus demo site and example calculations to go along with it so people can invite the sample bot to their own Slack team and try it out.

Better fix that. Suggestions:

• Use slash commands
• Respond only to messages where bot is tagged
• Respond only to messages in the designated channel

I favor the second idea, but slash commands might give more flexibility in the future.

Questions must be worded exactly to trigger the calculation: can we improve this by adding NLP?

Fetching the user ids and lastVisit timestamps for ~2000 users took around 5 seconds in tests earlier this weekend, but the time has inflated to ~60 seconds in the HackIllinois demo room.

By saving the object map once after the first calculation, we can avoid fetching it again in future calculations.