Reference

This reference describes a library for playing with csvs datasets. It lists data types, signatures and implementations for each function. Implementations are written in Program Design Language, or pseudocode.

It currently provides four methods - select, update, insert and delete, each with the same API of a transform web stream factory asking for access to the file system and the dataset directory.

Each method is currently implemented in the same style of constant memory and nested streams.

A top-level record stream accepts a query record and returns a record. It builds a list of tablets from each query - a strategy, - and builds a pipeline of tablet streams.

A tablet stream accepts a query record and returns a record. It streams tablet lines to a line stream. During operation multiple records are queued in the tablet stream, passed to record stream and queued there for consumption.

• Data Types
• Sow & Mow
• Schema
• Insert
• Update
• Delete
• Select

To learn more about the architecture of csvs, see other User Guides, the Specifications and the Requirements.

Notes:

• remove "playing" from here
• "library" is a term from code organization, this should talk about algorithms and data structures
• "transform web stream" is too language-specific
• there should be introductory text as to why we need to describe functions at all
• "function" is a term from code organization, this should use more general terms like "step" or "stage"
• the text about the stream pipeline should be moved elsewhere
• the list should have explanation and sales pitch for each item
• right now each page under this chapter is a design sketch, not documentation.