This is surely no trivial feature (both implementing and using). But when I think of the long lists of variables declared in nested procedure levels a record would come in handy.

Of course there has to be some calling convention to pass records as references, avoiding the need to copy all members to and fro. There seems to be something like that already silently working for long varchars.

FWIW, that request has been realized with SA 16 - confine the ROW (and ARRAY) data type constructors.

I'm not aware of the ORACLE implementation - so I guess the exact usage and syntax will differ in SA.

With respect to cursors with lots of fetch variables, SQL Anywhere's ANSI-compliant FOR statement might be used to reduce much of the discussed overhead with long variable lists. It also omits the need for explicit fetch statements and as such, can be schema-agnostic as long as columns are only added but not re-named or deleted (unless they aren't used in the loop).

For example:

:::SQL
begin
  for forCrs as crs cursor for
     select * from systable order by 1
     for read only
  do
    -- do something with the current row.
    -- Note: All variables are named after the according columns
    message 'table_name = ''' || table_name || '''' to client;
  end for;
end;

When you prefer particular variable names (e.g. to distinguish them from column names), you can use according alias names in the select list, such as

select table_id as @table_id, table_name as @table_name from systable order by 1

Of course, the FOR statement doesn't help to package long variable lists that are to be passed to different procedures so this is only a partial solution.