Performance

Remove iterating the same array twice in the pivot function

API improvements & documentation

Pivot is available from Skypack and can be found here: https://cdn.skypack.dev/@steelbreeze/pivot

API improvements

The API has seen incremental changes in 4.2 and now 4.3 releases. The following are highlights and breaking changes:

• The Predicate type has been renamed Criteria as this more accurately reflects its use in Dimension construction.
• The map function has been renamed aggregate as it does not just map the data, it also flattens the data by one dimension.
• The distinct function has been removed, as it is not strictly required, it was just provided as a utility and can easily be replicated with the code fragment below:

function distinct<TValue>(value: TValue, index: number, source: Array<TValue>): boolean {
	return source.indexOf(value) === index;
}

Documentation

All the function call documentation has now been tidied and has example code fragments. Documentation can be found here: https://steelbreeze.net/pivot/api/v4/

Tidy pivot function API and implementation

Refactored the API declaration and implementation of the pivot function for clearer and simpler use and maintenance.

Also simplified the implementations of the helper functions.

More cube creation performance

Remove the last remaining map call in pivotImplementation

Query performance

Just like the cube creation performance improvements, similar techniques applied to cube queries.

Yet more performance

Manual loop optimisation; changed:

for (var di = 0; di < first.length; ++di) {

to

for (var di = 0, dl = first.length; di < dl; ++di) {

Which makes a surprisingly large improvement.

A little more performance

4.1.8

Better typing in test

Start testing

Restructured non-core source code and started on a test suite (with a view to having coverage in place in a later release)

Minor refactor

The prior move to the partition function renders the copying of the source data redundant in the implementation of the pivot function; the latter can now just become an alias with the desired external API applied.

Minor performance tweak

Use var in place of let in for loops as this yields better performance in large data sets.