selection depth sometimes not properly calculated for holds

[Galen Charlton]

Consider a title-level hold request created open-ils.circ.holds.test_and_create.batch where the desired selection depth is not provided.

Depending on various permutations of boundary settings, the depth returned by the possibility check (ultimately open-ils.circ.title_hold.is_possible) may be 0.

That in turn can get passed into the logic that later creates the action.hold_request row via lines like this:

  $params->{'depth'} = $res->{'depth'} if $res->{'depth'};

However, there's a problem here: if $res->{'depth'} is 0, $params->{'depth'} is unchanged. If that key is missing, or contains an undef value, the selection_depth gets set to null.

Normally, ahr.selection_depth has a default value applied at the database layer in that case. In a stock system, that value is 0, but I've run across at least one system where that value was instead changed to 1.

IOW, depth = undef/NULL is not equivalent to depth = 0. $res->{'depth'} should be tested for existence and having a defined value, not just Perl truthiness.

Evergreen main

[Galen Charlton (gmc)]

(I grant that this is a pretty obscure bug.)