Version 1.3.4 is released with the following fixes:
Fix support for ascertainment bias correction [+ASC]
For codon models, the site ranges are now counted on number of nucleotides instead of number of codons. When DNA and codon data are mixed in partition model, branch lengths are now interpreted as #nucleotide substitutions per nucleotide site. Note that for codon data, branch lengths are #nucleotide substitutions per codon site like PAML.
Fix usage of -spp with -m TESTMERGE.
Fix ML optimization of amino-acid frequencies [+FO]. Fix memory deallocation when using -m TEST for Windows version.