Direction

[hwchen]

Just wanted to lay out a general sense for where this project can go in the near future.

Possible Directions

I think there's a few main areas to think about:

• Friendlier API
  • latest, previous (time)
  • figure out level by member id (i.e. for geography)
  • children, parents, related for members of a level in drilldown
  • substitution for members (automatic crosswalk)
  • deciding which cube to use (filter for cube, and fallbacks)
• Data management
  • ingestion and schema inference features
  • multiple db support
  • reading schema from db
• Calculations
  • Moving away from sql
  • plugin architecture using dll
• Engineering
  • testing (end to end cube mart)
  • docs
  • refactoring/cleanup
• Database support (sql generation for clickhouse, mysql, postgres)

Plans

To shore up foundational issues in the near future:

• full support on clickhouse, mysql, postgres
• end-to-end testing
• Friendlier api basics (TBD, but would include at least time component).

Then determine the path forward from there.

[jspeis]

one thing to just consider as well: machine learning