[Dwarf-discuss] RFC: DW_AT_copy_on_write (flag)

[Martin]

Pascal has ref-counted strings. If the user modifies such a string, then 
a new copy will be made unless the ref-count was equal to 1.

Would it be reasonable to request a flag to indicate data types with 
such behaviour?
(Such a flag might only be present on types that have the newly 
requested DW_AT_ref_count)

Also, does anyone know similar behaviour in other language?
I wonder if "flag" is the appropriate form.

For Pascal it would currently be enough. Since it only applies to 
strings, and strings can not be nested.

If a language had a refcounted array or structure type, then such a type 
could contain other (nested) ref-counted types (e.g. via 
pointer/reference).
There could then be a case where modifying a nested ref-counted array, 
triggers copy on write for the outer array too?
In that case a flag may not be able to express that.