How to generate synthetic rows (raw(16) guid cols) in one SQL statement?

Question

We're populating a table containing two GUID columns:
create table object
( object_guid raw(16) primary key
, project_guid raw(16)
)All object GUIDs are unique (thus the PK), and each object belongs to a given project (should be a FK to some project table). We want N objects / rows, belonging to only 100 projects, i.e. 1% of the rows of the object table belong to the project #1, 1% to #2, etc...
Right now we're using about 25 lines of C++/OCI code to do that (one query doing a "select sys_guid() from dual", and using the generated GUIDs to do inserts into object), but I suspect it's possible to do this using a single SQL statement using mysterious connect by or some other Oracle SQL magic. (our OCI code does quite a few round-trips to do the equivalent).
Would anyone please demonstrate how to generate the rows as explained above, and possibly describe how it works for the non-initiated?
Thanks, --DD
PS: I'm sure it can be done in PL/SQL as well, but I'm interested in a SQL version if one's possible.

Answer 1

I've found two ways, both taking a few SQL statements, but somehow I think this ought to be possible without intermediary tables... I'm sure there's a better way.
#1drop   table project_tmp;
create table project_tmp
as select rownum pid, sys_guid() guid from dual
connect by level <= 100;
drop   table object_tmp;
create table object_tmp
as select mod(rownum, 100) + 1 pid, sys_guid() guid from dual
connect by level <= 1000;
drop   table object;
create table object
as select o.guid object_guid, p.guid project_guid
from object_tmp o, project_tmp p
where o.pid = p.pid;
drop table project_tmp;
drop table object_tmp;#2:drop table project;
create table project
as select mod(rownum, 100) + 1 prj_id, sys_guid() guid from dual
connect by level <= 100;
drop table object;
create table object
as select mod(rownum, 100) + 1 prj_id, sys_guid() object_guid from dual
connect by level <= 1000;
alter table object add project_guid raw(16);
update object o set o.project_guid = (select guid from project p where p.prj_id = o.prj_id);
drop table project;
alter table object drop column prj_id;Verification:select count(distinct project_guid) from object;
select project_guid, count(OBJECT_GUID) from object group by project_guid;