Creating a function for your needs can be as simple as:

1. Write this code in a file called myfun.test:

CREATE OR REPLACE PROCEDURAL LANGUAGE plperlu;
ALTER PROCEDURAL LANGUAGE plperlu OWNER TO postgres;
CREATE OR REPLACE FUNCTION mytest(text) RETURNS text
  LANGUAGE plperlu
  AS $_X$
  my $data = $_[0];
  my $string = $data." -->from mytest\n"; 
  return $string;
$_X$;
[download]

(Of course you can have as many arguments and return types you want in your function's definition.)

2. Insert/replace this function in your db by issuing the command line:

3. Run the test (given a table called data and a text field called f1):

And that's it... When your function will be ready you can dump/restore it permanently into your db if you want.

I seems rather more complicated, not easier. In what way does this make the queries faster? Benchmark?

I would try something like:

my %ipH;
my $sth = spi_query("SELECT * FROM data");
while (defined ($row = spi_fetchrow($sth))) { 
  delete fetched or hashed;
}
[download]

but it's not tested.

And deletion is done by a command like:

$req = spi_exec_query("DELETE FROM data WHERE id = $id",1);
[download]

A cavall donat, no li miris el dentat.