rspec/rules/S5870/plsql/rule.adoc

== Why is this an issue?

Using ``++FORALL i IN x.first ... x++``.last or ``++FORALL i IN 1 ... x.count++`` might fail when indexed collections are sparse as Oracle tries to access non-existent element(s) of x. ``++FORALL i IN INDICES OF x++`` syntax will always work including sparse collections. Thus using ``++FORALL i IN INDICES OF x++`` should be preferred as it makes code more robust and easier to review.


=== Noncompliant code example

[source,sql]
----
FORALL i IN 1 .. l_tab.COUNT  -- Non-Compliant
    INSERT INTO forall_test VALUES l_tab(i);

FORALL i IN l_tab.first .. l_tab.last  -- Non-Compliant
    INSERT INTO forall_test VALUES l_tab(i);
----


=== Compliant solution

[source,sql]
----
FORALL i IN INDICES OF l_tab
    INSERT INTO forall_test VALUES l_tab(i);
----


== Resources

* https://asktom.oracle.com/Misc/oramag/oracle-10g-adds-more-to-forall.html[Oracle 10g Adds More to FORALL]
* https://stevenfeuersteinonplsql.blogspot.com/2019/03/using-sparse-collections-with-forall.html[Using sparse collections with FORALL] - Steven Feuerstein

ifdef::env-github,rspecator-view[]

'''
== Implementation Specification
(visible only on this page)

=== Message

Replace "IN ..." with "INDICES OF ..."


=== Highlighting

* ``++IN x.first ... x.last++``
* ``++1 ... x.count++``


endif::env-github,rspecator-view[]