@autoschedule/queries-scheduler

Queries scheduler

Stats

stars 🌟issues ⚠️updated 🛠created 🐣size 🏋️‍♀️
015Jun 1, 2020Oct 23, 2017Minified + gzip package size for @autoschedule/queries-scheduler in KB

Readme

Queries Scheduler

Place queries on a timeline.

Build Status Maintainability Test Coverage code style: prettier semantic-release Commitizen friendly Greenkeeper badge npm version

workflow: queries -> schedule

needs:

  • once material is placed, don't replace it. Causes loops. Temporary query instead. (Current behavior is different: potentials and materials streams are re-compute after each mutation and we check for stability. Cause issue: updatePP isn't accurate. Stick to first strategy)

operations:

  • place potentials with most pressure.
  • When all possible queries are placed and some potentials are left, throw exception. New provider query is pushed (send needer id so provider query can have a direct dependency). Can now be placed thanks to user state potential.
  • Temp Queries: fixes from agent feedback, placeholder, content with transforms.
  • two ways to provide: provider depends on needed task; needed task depends on provider.
  • tasks with zero potential have pressure of -1 instead of +Infinity.
  • task potential = intersection of query potential (internal constraints) + query dependencies + user state potential + materials.
  • impossible to place (material time) => throw exception. (conflict)
  • query that occurs x time before/after state: Link to material that update state -> link field, queries-scheduler module.
  • How to allow link to a material from another agent ? Could be after the second phase, when agent receive queries's context. (Link to an ID)
  • Only one exception is permitted: conflict error that needs user. Others are caught (from userstate) with conflict-resolver
  • providers handle is complex. Use priority list. Can cause side-effect conflict. How to go back to the source of conflict ? How to switch from 1st priority to another ? Simulation with satisfaction threshold for all materials ? heavy cost (recursive and exponential, due to provider sub-dependance to other providers).
  • Agents -> queries ; conflict-resolver (error): query -> agent ; manage agents as a stream

when the stream complete without error, it should be the final timeline.

schedule (tasks (materials)):

  • tasks (potentials)

tasks (potentials):

  • queries
  • tasks (potentials)
  • tasks (materials)
  • user state (potentials)
  • user state (materials)

tasks with needs:

  • tasks (materials)
  • agent service

Stream:

  1. agent queries
  2. user interaction
  3. [1, 2] queries
  4. [3, 5] queries with temp
  5. [4, 5, 7] tasks (potential) - use user-state function on tasks (potential & material)
  6. [5] tasks (material)
  7. [7] needs and fixes, placeholder fill or validate

a. Catch errors from [7] => generate new queries at [1].

Module: [1] queries-fn [2, 3] main app [4, 5, 7] queries-scheduler userstate-manager [a] conflic-resolver [b] agent-relay

Module a: if provider is impossible to place, either there is no need for it, or there is a conflict. Use user-state to determine. userstate-manager: query + (potential/material) with needs + config + base needs => ranges of possibilities

Queries restrictions:

  • provider (insert/update with 'wait: true' can't specify start/end) -> will prevent correct material placement order.

If you find any bugs or have a feature request, please open an issue on github!

The npm package download data comes from npm's download counts api and package details come from npms.io.