\n What if to define table structure along with aggregation functions in the same information schema?\n That could look like some kind of transaction definition that is represent in json schema.\n
\n\n The idea is to have definitions of cross table relations and calculation rules in one place. \n Such table scheme could easy listen to the events stream and do changes\n in the related tables accordingly.\n
\n\n The functionality aimed to provide rapid access to continuously changed\n data to the modern web application.\n
\n\n The changes in tables could be transformed and\n propagated as same events to the another aggregation tables \n where it could be mixed with another sources.\n Aggregate engine provides data change events for the subscribed clients.\n In the same way standard web queue could subscribe to such the aggregated\n event sources.\n
\n\n In general the idea looks like materialized view that is gathered on event stream.\n The information schema provides ability to validate incoming messages against the schema.\n
\n\n```javascript\n \"invoice\": {\n \"type\": \"object\",\n \"properties\": {\n \"date\": { \"type\": \"string\" },\n \"customer\": { \"type\": \"string\" },\n \"amount\": { \"type\": \"string\" }\n },\n \"primary_key\":[ \"date\", \"customer\" ],\n \"propagate\":{\n \"detail\":[{\n \"key\": [ \"customer\" ],\n \"fields\": { \"total\": \"amount\" },\n \"filter\": {\"type\":\"sale\"}\n }]\n }\n },\n\n \"detail\": {\n \"type\": \"aggregate\",\n \"properties\": {\n \"customer\": { \"type\": \"string\" },\n \"total\": { \"type\": \"string\" }\n },\n \"primary_key\": [ \"customer\" ],\n \"sum\": [ \"total\" ]\n }\n```","readmeFilename":"README.md"}