Siddhi Pattern Matching - sacjaya/siddhi-3 GitHub Wiki

Pattern matching should support the following scenarios

Queries of the type from a1 = A -> B select a1.price
Logical and,or should be supported
- A -> B or C -> D
  - Will match A, B, D
  - Will match A, C, D
  - Will it match A, B, C, D ?? No, it will only match A, B, D for events received in the order A, B, C, D
- A -> B and C -> D
  - Will match A,B,C,D
  - Will match A,C,B,D
Counting patterns
- A -> B <3:5> -> C
  - Will match A, B, B, B, C
  - Will match A, B^5, C
  - Will match A, B^6, C
  - Does using upper limit make sense in counting patterns?
- A -> B <3:> -> C
- A -> B <:5> -> C
Logical not
- A -> not B and C -> D
  - Will match A, C, D
- A -> not B within 5 min -> C
  - Will match A, {no B events for 5 min, and then} C
Support for 'within'
- Within can be used inside any pattern matching construct. e.g.
  
  A -> (B -> C) within 5 min -> D
  - Will match A, B, C, {B,C within 5 min}, D
Support for 'prev' keyword within a filter of pattern
- a1 = A[(not (a1[prev] is null) and a1[prev].price > price)]<2:10> ???
  - The first A event will be matched anyway. From second A event onwards, it will match only if the previous A event price is greater. With this we can detect monotonically increasing/decreasing patterns.
Support for aggregate functions on counting patterns. e.g.
- For the pattern a1 = A [prev.price > price]<2:10>, the selector avg(a1.price) would give the average price of all the events matched inside the counting pattern for that single total match.

Following will not be supported