use DBI;
use strict;

my $db = DBI->connect("DBI:CSV:");

# create table
$db->do("create table testdb (NAME CHAR(20),POSTAL CHAR(20),TELEPHONE 
+CHAR(20))");

# create dummy data
my $sth=$db->prepare("insert into testdb (NAME,POSTAL,TELEPHONE) VALUE
+S (?,?,?)");

for (1..10)
{
        $sth->execute("name $_","postal $_","telephone $_");
}
# update 2 lines of data with different format statements
my $sth1 = $db->prepare("update testdb set NAME=?,POSTAL=?,TELEPHONE=?
+ where name=?");
my $sth2= $db->prepare("update testdb set POSTAL=?,TELEPHONE=? where n
+ame=?");

$sth1->execute("name 9","postaladdress","123456","name 9");
$sth2->execute("postaladdress","123456","name 10");

# display the contents of testdb
my $sth3 = $db->prepare("select * from testdb");

my $result = $sth3->execute();

my $line;
while ($line = $sth3->fetchrow_arrayref)
{
        for my $attrib (@$line)
        {
                print "$attrib\t";
        }
        print "\n";
}
[download]

(See below for output)

If I change the connect line as follows:

my $db = DBI->connect("DBI:SQLite:dbname=testfile","","");
[download]

I get a different result. This is the output using DBD::CSV

name 1  postal 1        telephone 1
name 2  postal 2        telephone 2
name 3  postal 3        telephone 3
name 4  postal 4        telephone 4
name 5  postal 5        telephone 5
name 6  postal 6        telephone 6
name 7  postal 7        telephone 7
name 8  postal 8        telephone 8
name 9  name 9  postaladdress
name 10 name 10 postaladdress
[download]

and this is the output using DBD::SQLite

name 1  postal 1        telephone 1
name 2  postal 2        telephone 2
name 3  postal 3        telephone 3
name 4  postal 4        telephone 4
name 5  postal 5        telephone 5
name 6  postal 6        telephone 6
name 7  postal 7        telephone 7
name 8  postal 8        telephone 8
name 9  postaladdress   123456
name 10 postaladdress   123456
[download]

As you can see, the lines that have been changed in the DBD::CSV test are corrupt, whereas the corresponding lines in the DBD::SQLite test are correctly updated.

Am I tripping over an incorrectly formatted statement that DBD::SQLite forgives me for, and DBD::CSV doesn't, or is there a bug in DBD::CSV?

thanks,

Update: I'm running perl 5.8.7, and installed both DBD::CSV and DBD::SQLite using cpan install last night, so they should both be bang up to date.

Update:Versions:

DBD::CSV - 0.22
SQL::Statement - 1.14
DBI - 1.48
DBD::File - 0.33
Text::CSV_XS - 0.23

OS is linux