Union

Construct with a schema list in tuple form, for example (Int32 Nat8) Union.

Fields

• Items: list of supported schemas.
• data: raw aligned storage for the selected interpretation.
• pad: remaining storage needed to reach the maximum item size.

Methods

• get (key -- ref): returns one Ref to the overlapping storage interpreted as the schema at key.

Alignment and size helpers

• MAXIMUM_ALIGNMENT: greatest alignment among the supported schemas.
• MAXIMUM_SIZE: greatest storage size among the supported schemas.

Getter result schemas

"Union" use
"control" use

{} () {} [
  u: (Int32 Nat8) Union;
  0 @u.get printStack _:;
  1 @u.get printStack _:;
] "main" exportFunction

Expected Output During Compilation

Int32 Ref
Nat8 Ref

Write through typed refs

"Union" use
"String" use
"control" use

{} Int32 {} [
  u: (Int32 Nat8) Union;
  123 0 @u.get set
  ("int=" 0 @u.get new LF) printList
  7n8 1 @u.get set
  ("nat=" 1 @u.get new LF) printList
  0
] "main" exportFunction

Expected Output

int=123
nat=7

Reinterpret the same storage through another key

"Union" use
"String" use
"control" use

{} Int32 {} [
  u: (Int32 Nat32) Union;
  -1 0 @u.get set
  ("asInt=" 0 @u.get new LF
   "asNat=" 1 @u.get new LF) printList
  0
] "main" exportFunction

Expected Output

asInt=-1
asNat=4294967295

Storage and reference semantics

• Union has no tag and does not track which schema was written last.
• Every get call returns one typed view of the same overlapping storage.
• Reading the storage through a different key than the one used for writing reinterprets the same bytes with no validation.
• The returned references point into fixed storage and remain valid while the union object itself remains alive.
• Writing through one returned reference changes the bytes observed through every other key.
• The union itself provides no branch-specific initialization or destruction. Schemas that need manual lifecycle management must be managed by the caller.
• MAXIMUM_ALIGNMENT and MAXIMUM_SIZE are derived from the supported schema list and determine the storage layout.

See also

• Variant: Tagged multi-schema storage and helper operations.
• Dict: Struct schema with named items.
• Tuple: Struct schema with unnamed items whose item schemas are not required to match.