Victor, The mmio.c I sent to you was obtained few years ago. Your mmio.c diff from the current mmio.c on diff Victor_mmio.c mmio.c 10d9 < #include <stdlib.h> 12a12

#include <malloc.h> 251c251 < if (fprintf(f, "%d %d\n", M, N) < 0)

if (fprintf(f, "%d %d\n", M, N) != 2)

393c393 < if (ret_code < 0 ) ---

if (ret_code !=2 )

441c441,453 < ---

/** * Create a new copy of a string s. strdup() is a common routine, but * not part of ANSI C, so it is included here. Used by

mm_typecode_to_str().

* */ char *strdup(const char *s) { int len = strlen(s); char *s2 = (char *) malloc((len+1)*sizeof(char)); return strcpy(s2, s); } 446a459 char *strdup(const char *);

It seems the bug is on when to raise error flag? I have been using several matrices converted from matrix market using their mmio.c. I just want make sure the bug has not affected the correctness of my converted matrices. I'll put your mmio.c in petsc repository. Thanks, Hong On Thu, 7 Apr 2005, Victor Eijkhout wrote:

At 1:41 PM -0500 4/5/05, Hong Zhang wrote:

...

The files mm2petsc.c, mmio.c and mmio.h are attached.

Years ago I found a couple of bugs in mmio.c and reported them to Roldan Pozo. Apparently they haven't been fixed. The good boy is too busy with other things. -- Victor Eijkhout <[email protected]>, 329 Claxton, Comp Sci, UT, Knoxville TN 37996. tel: 865 974 9308 (W), 865 673 6998 (H), 865 974 8296 (F) http://www.cs.utk.edu/~eijkhout/