An ALFA UDT can be declared multiple times with a fragment modifier. At compile time all fragment declarations get merged into the main UDT declaration.

Fragment declarations can add additional attributes and annotations to the main declaration and is a useful way to manage separation of concerns around writing data models.

For example consider a record declared as shown below.

record NewOrderSingle {

In order to inject additional custom attributes, a fragment can be declared as :

fragment record NewOrderSingle includes AuditInfo {

trait AuditInfo {
    CreationTime : datetime
    CreationSystem : string

At compile time, the ALFA compiler will combine these 2 declarations into a single NewOrderSingle. Any conflicts will be reported to the user ( e.g. A field cannot be redefined )

Fragments should be used with care, ideally as part a well-organised build system as it is effectively injecting attributes. Code generated for the non-fragment declaration may be incompatible with the one with the fragment declaration.