Procedure for logging in PL/SQL

Question

I am looking for a way to log sqlplus output from within a pl/sql procedure. When run through SQL*PLUS directly I can use SPOOL. Is there any way to do the same from within PL/SQL? Following is an example of what I was previously executing as a .sql script via command line SQLPLUS logon/password @instance @sqlscript.sql. Now I want to create a stored procedure to do the same...
Thanks,
Susan
spool dws_office.log
SET HEADING ON
SET ECHO ON
SET VERIFY ON
SET FEEDBACK ON
select TO_CHAR(CURRENT_DATE, 'DD-MON-YYYY HH:MI:SS AM') from DUAL;
TRUNCATE TABLE OFFICE;
INSERT INTO office
(SELECT office_id,
region_id,
office_name,
office_type,
primary_phone,
date_active,
agency,
default_email,
zip,
city,
agency_office_code,
county,
fax,
office_description,
office_JSC_id,
secondary_phone,
zip_plus_four,
state,
address1,
address2,
maintenance_user,
maintenance_date,
date_inactive
FROM office_data);
UPDATE office o
SET o.OfficeGroupID =
(SELECT og.OfficeGroupID
FROM OfficeGroup og
WHERE o.officeid = og.officeid),
o.OfficeGroupName =
(SELECT og.OfficeGroupName
FROM OfficeGroup og
WHERE o.officeid = og.officeid),
o.regionname =
(SELECT r.region_name
FROM regions r
WHERE o.regionid = r.region_id),
o.georegionid =
(SELECT g.georegionid
FROM georegion g
WHERE o.regionid = g.regionid),
o.georegionname =
(SELECT g.georegionname
FROM georegion g
WHERE o.regionid = g.regionid),
o.arearollupname =
(SELECT e.arearollupname
FROM esdrolluparea e
WHERE o.officeid = e.officeid);
commit;
select TO_CHAR(CURRENT_DATE, 'DD-MON-YYYY HH:MI:SS AM') from DUAL;
BEGIN
dbms_stats.gather_table_stats(ownname=> 'DWSCHEMA', tabname=> 'office', partname=> NULL);
END;
spool off
exit

Answer 1

Just keep in mind that SQL*Plus scripts invariably contains DDLs and SQL*Plus commands and that these cannot be simply wrapped into an Oracle DBMS_JOB.
You will need to loose all the SQL*Plus specific stuff and use dynamic SQL for all the DDLs.
In which case it is more sensible to rather wrap a script into a formal Oracle stored procedure or package instead, than it is to pass a huge block of anonymous PL/SQL code to execute to the DBMS_JOB handler.
If there are a series of jobs with dependancies, it can be done a couple of ways. Oracle 10G's DBMS_SCHEDULER is one option.
Another is to only use a single Oracle DBMS_JOB to run all the steps, and using exception handling to control/abort processing steps.
You also may not want Oracle to auto restart a job that raised an exception - and instead want to handle that manually or programatically in some other fashion.
As Justin mentioned, Oracle will auto restart a job that raises an exception. It will do so 16 times after which it will mark that continuously failing job as broken.
But then if a job fails, it will likely fail again and one may not want to keep on running that job into the same brick wall over and over again.
The bottom line in my view is to forget the how-it-was-done using the previous job scheduling system and re-look/redesign it for running using Oracle's job management systems. A simple "port" from one system to another will cause more new problems that solving existing ones IMO.