Documentation of Pipeline
Pipeline library defines functions for a variety of purposes (e.g. generating, filtering, transforming, ordering, etc) that operate on ranges of elements. Note that a range is defined as
[begin, end) where end refers to the element past the last element to inspect or modify.
Anatomy of Pipeline
To make it easier for understanding how pipeline works and what each function does, this documentation classifies functions based on their behavior and defines some terminologies along the way.
A pipe is a function which either generates (or prepares) data or processes them.
A pipe which generates or prepares data is called generator pipe.
A pipe which processes data is called processor pipe.
A pipeline consists of many pipes separated by pipe operator |. It could have zero or one generator pipe, and zero or more processor pipes.
A pipeline with zero generator pipe is called unbounded pipeline.
A pipeline which has one generator pipe is called bounded pipeline. In a bounded pipeline, the generator pipe is always the first pipe.
Consider this pipeline,
auto evens = from(v) | filter(e % 2 == 0) | transform(e * 2); ^^^^^^ ^ ^^^^^^^^^^^^^^^^^ ^ ^^^^^^^^^^^^^^^^ | | | | | | | | | procoessor-pipe | | | pipe-operator | | processor-pipe | pipe-operator generator-pipe
As indicated using ascii diagram above, from, filter and transform are called pipes. from is a generator pipe which only prepares the data. filter and transform are called processsor pipes which process the data one after another. The symbol | between two pipes is pipe-operator. In a typical pipeline, the first pipe is a generator pipe, and rest are processor pipes, separated by pipe operators.
//this is how a typical pipeline looks like auto results = generator-pipe | processor-pipe1 | processor-pipe2 | processor-pipe3;
There are lots of pipes belonging to each group. So it is easier to understand if we know which group they belong to.
Most of the signature has return type mentioned as implementation-defined.