search

NKcqx / CLIC

旨在提供一个跨平台计算框架来统一异构软件系统
5 stars 1 forks source link

实现Java的Join Operator #10

Closed NKcqx closed 4 years ago

NKcqx commented 4 years ago

志伟实现的java Operator中还没有Join，简单实现一下

DLUTLiuFengyi commented 4 years ago

https://github.com/NKcqx/IRDemo/commit/213cee4b0fd497bf07e295ef93e56fa34441d067 参考rheem等，步骤是： ①左右表分别更新成join之后自定义要保留的列属性的组合 ②左右表分别构建map<key, <该key对应的行>> ③嵌套循环这两个map，按照key值合并两边的行

DLUTLiuFengyi commented 4 years ago

脚本compile_udf_class.sh用来将udf.java文件自动编译成.class文件

DLUTLiuFengyi commented 4 years ago

https://github.com/NKcqx/IRDemo/commit/7f8b47820bea8134b9426e7a237b36de53125592 将Nested Loop Join改成Hash Join，时间复杂度预计可从 O(m*n) 优化为 O(m+n)

DLUTLiuFengyi commented 4 years ago

https://github.com/NKcqx/IRDemo/commit/f3421b365e593324c24adfef3885b189711b6cf1 完成Hash Join java的join operator已实现Nested Loop Join和Hash Join

DLUTLiuFengyi commented 4 years ago

https://github.com/NKcqx/IRDemo/commit/8a57951b03297686fd7f980377fa28f26d60bb92 ①java join case跑通 ②join operator代码逻辑简化 ③为代码整洁暂把Hash Join算法移除

DLUTLiuFengyi commented 4 years ago

target文件夹已删除