I used a simplistic script to convert GB to EMBL format and manual inspection reveals strange behaviour

#!/usr/bin/perl -w
use strict;
use Bio::SeqIO;

if (@ARGV != 2) { die "USAGE: gb2embl.pl <GB_file> <EMBL_name>\n"; }

my $seqio = Bio::SeqIO->new('-format' => 'genbank', '-file' => "$ARGV[0]");
my $seqout = new Bio::SeqIO('-format' => 'embl', '-file' => ">$ARGV[1]");
while( my $seq = $seqio->next_seq) {
  $seqout->write_seq($seq)
}

a piece of input GB formatted

 mRNA            complement(join(280..5840,5989..6007))
                 /gene="YRF1-7"
                 /locus_tag="YPL283C"
                 /gene_synonym="YRF1"
                 /product="Y' element ATP-dependent helicase protein 1 copy
                 7"

the same piece in the embl output

FT   mRNA            join(complement(280..5840),complement(5989..6007))
FT                   /locus_tag="YPL283C"
FT                   /gene_synonym="YRF1"
FT                   /gene="YRF1-7"
FT                   /product="Y' element ATP-dependent helicase protein 1 copy
FT                   7"

I do not think that this is OK

complement( join ( -1-> -2-> ) ) = <-2- <-1-
join( complement( -1-> ) complement ( -2-> ) ) = <-1- <-2-

the order gets inverted by this logic, leading to chaotic sequence fusion

Anybody agrees and/or has comments?

Bioperl used: 1.007001