clustalw-mpi-0.15

1. bug fix version 0.15 released on May 23, 2008. In the previous version,
memory crash may occur if the length of a sequence's name is longer than 60
characters. This was defined in "clustalw.h" by the typedef macro of stree.

/***** Original code ***************************************************/
typedef struct node {           /* phylogenetic tree structure */
        struct node *left;
        struct node *right;
        struct node *parent;
        float dist;
        sint  leaf;
        int order;
        char name[64];
} stree, *treeptr;
/***********************************************************************/

In version 0.15, we have changed the definition to "char names[MAXNAMES]".
It means you must re-define macro MAXNAMES in order to process FASTA file
with sequence names longer than the default length.

/***** New code ***************************************************/
typedef struct node {           /* phylogenetic tree structure */
        struct node *left;
        struct node *right;
        struct node *parent;
        float dist;
        sint  leaf;
        int order;
        char name[MAXNAMES];
} stree, *treeptr;
/***********************************************************************/



 clustalw-mpi-0.14

1. bug fix version 0.14 released on dec 19, 2005.

   (1) in the old version, "mpirun -np 3 ./clustalw-mpi two.seq" would hang
       provided the input file "two.seq" contains only two sequences. This bug has
       been fixed. 

       Details: in pairalign_new.c: if the number of slaves is greater than
       N*(N-1)/2, we will MPI_Send the sequence data to N*(N-1) slaves only.  This
       is to prevent the idled slaves not being able to get out of their while(1)
       loop, see "parallel_compare.c". 

clustalw-mpi-0.12

1. Fixed a bug that would hang the program when using "mpirun -np 2 ./clustalw-mpi".

clustalw-mpi-0.11

1. Fixed a bug that would hang the program when using "mpirun -np 1 ./clustalw-mpi".
