PG/GP排序中的abbreviated key

PG/GP排序中的abbreviated key

PG/GP排序私有状态结构Tuplesortstate中有一个成员SortSupport sortKeys;而sortKeys中有这么几个成员abbreviate、abbrev***：

"abbreviate" concerns whether or not the abbreviated key optimization is applicable in principle。也就是说可以使用此进行一个缩写key优化。那么这个优化是什么呢？

这个是9.6引入的新特性，看下它的patch介绍：

Add sort support routine for the UUID data type. This introduces a simple encoding scheme to produce abbreviated keys: pack as many bytes of each UUID as will fit into a Datum.  On little-endian machines, a byteswap is also performed; the abbreviated comparator can therefore just consist of a simple 3-way unsigned integer comparison.The purpose of this change is to speed up sorting data on a column of type UUID.

Make abbreviated key comparisons for text a bit cheaper.If we do some byte-swapping while abbreviating, we can do comparisons using integer arithmetic rather than memcmp.

也就是对uuid,bytea,char(n)的排序性能增强，使用abb keys, 整型比较算法取代memcmp。

Patch地址：

/?p=postgresql.git;a=commit;h=a76ef15d9fc9207a0758e8d6f6700dc8c931a934

/?p=postgresql.git;a=commit;h=bfb54ff15a447fb22e9deae096e0d45b3e4bd56f