Development philosphy PL/SQL vs SQL

Question

Hi All
I talked to another fellow developer and both he and me both agrees that our company code base uses too much for loop in the PL/SQL.
(1)He says if the data is in Oracle already better avoid PL/SQL all together and just use SQL statement.
Any suggestions on (1)
This company is really data-sensitive so perhaps in the very beginning they would do it row-by-row because of that.
And this fellow engineer talks about we can use ON ERROR when in cases like that.
Anyone can give an example for using this ON ERROR ?
Or can someone show an re-startable SQL example. For example: when processing 10 million rows and it fails in the middle how we can re-start the process in SQL without re-doing the whole thing?
-Thanks for all your inputs

Answer 1

Hi,
vxwo0owxv wrote:
Hi All
I talked to another fellow developer and both he and me both agrees that our company code base uses too much for loop in the PL/SQL.
(1)He says if the data is in Oracle already better avoid PL/SQL all together and just use SQL statement.
Any suggestions on (1).Doing things in PL/SQL shat should be done in pure SQL is definitely a common mistake. Let's call this Mistake A.
Doing things in SQL that should be done in PL/SQL is also a mistake. Let's call it Mistake B. In practice, Mistake B is much less common mistake than Mistake A, but it's a mistake just the same.
FOR loops are one common way people make Mistake A. SQL was designed for efficiently hanlding sets of rows, for example, executing a query and inserting its result set into a table. A common manifestation of Msitake A is to put the query in a cursor FOR loop, and INSERT the data one row at a time. (Among the mad wags, this is called Slow-by-Slow Priocessing .)
This company is really data-sensitive so perhaps in the very beginning they would do it row-by-row because of that.What does "data-sensitive" mean?
Whatever it means, I don't see how it can justify doing soemthing in a slow, error-prone way.
And this fellow engineer talks about we can use ON ERROR when in cases like that.
Anyone can give an example for using this ON ERROR ?
Or can someone show an re-startable SQL example. For example: when processing 10 million rows and it fails in the middle how we can re-start the process in SQL without re-doing the whole thing?ON ERROR is a SQL*Plus mechanism for trapping errors.
PL/SQL's error handling is better, so this might be an occasion for committing Mistake B. (It would be more accurate to call this Mistake C, since SQL*Plus is not the same as SQL.)
Within PL/SQL, you can use many of the best features of SQL (for example, a multi-row INSERT rather than a FOR loop), and also take advantage of the good features of PL/SQL (such as EXCEPTION handling).