I agree that Text::CSV is almost certainly the best way, but here's a reinvented wheel (note the leading "null" field introduced into the example):

c:\@Work\Perl\monks>perl -wMstrict -MData::Dump -le
"my $s = ',1,,3,4,,,';
 ;;
 my @ra = map { m{ \A \s* \z }xms ? undef : $_ } split ',', $s, -1;
 dd \@ra;
"
[undef, 1, undef, 3, 4, undef, undef, undef]
[download]