From: Robert Haas Date: Mon, 20 Sep 2021 17:25:38 +0000 (-0400) Subject: try to make README a tad more clear X-Git-Url: http://git.postgresql.org/gitweb/?a=commitdiff_plain;h=3a520a8b178844e3202bc6a520969efbaa98aea3;p=users%2Frhaas%2Fpostgres.git try to make README a tad more clear per a question from Mark --- diff --git a/src/backend/access/conveyor/README b/src/backend/access/conveyor/README index 199df3b1aa..ef730dc4ad 100644 --- a/src/backend/access/conveyor/README +++ b/src/backend/access/conveyor/README @@ -12,13 +12,12 @@ data that needs to be stored at any given time. Some other parts of PostgreSQL, for example xlog.c and slru.c, handle similar needs by using a series of files at the operating system level, -adding new ones at the end and removing old ones. Here, we want a system -that can be used within a standard relation fork, and so instead maintain a -logical-to-physical page mapping. Callers allocate new logical page numbers -(which are just consecutive 64-bit integers) and this module takes care of -figuring out where to place them phyically. When the oldest logical pages -are thrown away, the blocks allocated to them can be reused for new -logical pages. +adding new ones at the end and removing old ones. We want to use a standard +relation fork, and so instead maintain a logical-to-physical page mapping. +Callers allocate new logical page numbers (which are just consecutive 64-bit +integers) and this module takes care of figuring out where to place them +phyically. When the oldest logical pages are thrown away, the blocks +allocated to them can be reused for new logical pages. Conceptually, a relation fork organized as a conveyor belt has three parts: