It's kind of like phylip. It's closest in format to the gphocs that's written out now, with the full sequence for each locus in blocks. The 3s format only allows for 1 sequence from at most 3 taxa per block, and the sequences can come from mix-and-match individuals within a taxon. This means you'd have to use a pops file, and that it would have to be restricted to 3 populations. It's also not clear how you'd sample sequences from taxa represented by multiple individuals. Randomly? This is non-trivial, and also kind of edge-casey.
It's kind of like phylip. It's closest in format to the gphocs that's written out now, with the full sequence for each locus in blocks. The 3s format only allows for 1 sequence from at most 3 taxa per block, and the sequences can come from mix-and-match individuals within a taxon. This means you'd have to use a pops file, and that it would have to be restricted to 3 populations. It's also not clear how you'd sample sequences from taxa represented by multiple individuals. Randomly? This is non-trivial, and also kind of edge-casey.