Github user bhlx3lyx7 commented on a diff in the pull request:
https://github.com/apache/incubator-griffin/pull/442#discussion_r228215711
--- Diff:
service/src/main/java/org/apache/griffin/core/measure/entity/Rule.java ---
@@ -81,6 +81,9 @@ Licensed to the Apache Software Foundation (ASF) under one
@Column(name = "\"out\"")
private String out;
+ @JsonInclude(JsonInclude.Include.NON_NULL)
--- End diff --
This parameter `cache` in rule configuration indicates spark will cache the
output data frame, to avoid redundant calculation of the same data frame if it
is used twice or more.
In "griffin-dsl" mode, the rule will be translated internally, and the
`cache` parameter would be filled by the translator. That's why it only works
in "spark-sql" and "df-ops" mode.
In the current versions, UI doesn't support user defined "spark-sql" rules
or "df-ops" rules, the `cache` parameter seems confusing for users either. But
if we want to support user defined "spark-sql" rules, no matter through UI or
service, it would be better to make the parameters configurable. It's ok to set
it optional, I prefer to set it `false` by default.
---
