EQL: Sequence/Join parsing and model #54227
Conversation
|
Pinging @elastic/es-ql (:Query Languages/EQL) |
|
Considering the grammar around timeunits and fork is still discussed, I've let the parser and is and did not spend additional time on the tests; once the that part of the grammar gets locked again, this needs to be revisited. |
| case "days": | ||
| timeUnit = TimeUnit.DAYS; | ||
| break; | ||
| default: |
There was a problem hiding this comment.
I added #54760 to track an issue for removing unknown time units and only supporting ones that are recognized by range queries.
|
@elasticmachine run elasticsearch-ci/2 |
|
@elasticmachine update branch |
|
@elasticmachine update branch |
| Number number = (Number) visitIntegerLiteral((IntegerLiteralContext) numberCtx).fold(); | ||
| long value = number.longValue(); | ||
|
|
||
| if (value <= 0) { |
There was a problem hiding this comment.
does the sequence of 0 timespan make sense at all?
There was a problem hiding this comment.
I think not - if no maxspan is needed I opted for not declaring it.
If we agree to have 0 indicate no limit across EQL, the parser can be improved.
Add parsing and (logical) domain model for sequence and join (cherry picked from commit 9e9632d)
PR that parses the grammar for EQL join and sequence and creates their respective domain model.
The physical plan and execution will follow-up up in a separate PR.