[attrname := ] BUILD( indexdef, dataset, [, options ] );
Form 4 allows you to separate the index from the dataset, making it easy to use the INDEX form that does not require a dataset first parameter.
The usual way of defining and building an INDEX is in terms of the dataset that is used to create it, using an INDEX declaration where the dataset is named as the first parameter of the INDEX. This is fine when the dataset is relatively simple, but there are downsides:
The INDEX can't be logically separated from the dataset it was created from.
If the dataset is very complicated (Mb of source), when the index is subsequently used in a query all the code to create it is also parsed.
This form of BUILD allows the two to be separated, so you can omit the base dataset parameter from the INDEX definition and just specify the dataset to use in the BUILD action. The fields are automatically mapped (by field name) from the dataset to the index.
Examples:
//usual way to BUILD an INDEX:
ds1 = DATASET(100, TRANSFORM({ UNSIGNED id }, SELF.id := COUNTER));
i1 := INDEX(ds1, { id }, 'myIndex'); //specifies the dataset to always use
BUILD(i1);
//Separated way to BUILD an INDEX:
ds2 = DATASET(100, TRANSFORM({ UNSIGNED id }, SELF.id := COUNTER));
i2 := INDEX({ UNSIGNED id }, 'myIndex');
BUILD(i2, ds2); //builds the i2 INDEX from the ds2 dataset