获取使用DBIx :: Class :: ResultSet :: populate（）插入的记录的标识的最佳方法是什么？

Question

As we know, the populate() method in DBIx::Class::ResultSet provides efficient bulk inserts.

However, we also need to get the identity of the inserted records for further processing.

We have something like this:

my @created_records = $self->schema->instrument_rs->populate(\@records_to_create_rs);

foreach my $record ( @created_records ) {
    $id_of->{ $record->Name } = $record->Id;
}

But, when populate() is used in a non-void context, list context in this case, it just acts as a wrapper to the create() method and not as an efficient bulk insert.

Given this limitation, what do you recommend as the best way to get identity records bulk-inserted using DBIx::Class::ResultSet::populate()?

Answer 1

If you have an auto-incremented primary key,
and there are no other insertions going on at the same time,
would it not be a solution to simply bulk insert N-1 records, then insert the last via normal create?

For example ( worked for me ):

my $last_info = pop @records_to_create_rs;
my $items = scalar(@records_to_create_rs);
$rs->populate( \@records_to_create_rs );
my $last = $rs->create( $last_info );
my ( $first_id, $last_id ) = ( $last->id - $items, $last->id );
my @inserted_ids = ( $first_id .. $last_id );

Edit: if you wanted to check whether the insertion went as planned ( instead of ensuring it, or anyway ), one could expand upon this by also shifting and inserting the first info and comparing whether the id count is off. If it is, the method could default to get the records via search ( no speed gain then, sorry ), if it is not, all went well and we are done.

my $first_info = shift @records;
my $items = scalar( @records );
my $last_info = pop @records;
$rs->create( $first_info );
$rs->populate( \@records );
my $last = $rs->create( $last_info );
my $expected = $first_id +1;
if ( $expected == $last->id - $items ){
   return [ $first_id .. $last->id ];
}else{
   # get the records another way, I did not check this
   @records = $rs->search( \@records )->all;
   unshift @records, $first;
   push @records, $last;
   return [ map { $_->id } ];
}

I actually put the first version into one of my add_to methods ( given array, bulk_insert ), it appears to be working fine so far.

I will test the second version one of those days and report if it fails...

Answer 2

This is what I've done but the answer from @bytepusher seems to be more efficient. Posting for documentation of all possible answers.

$self->schema->instrument_rs->populate(\@records_to_create_rs);

$created_records_rs = $self->schema->instrument_rs->search(
    { 'InstrumentTypeId' => $inst_type_id, },
    { order_by           => { -desc => 'Id' } }
)->slice(0,$records_to_create_count - 1);