3 Port of Gibbs package to VMS, completed 20-MAR-1995    David Mathog mathog@seqvax.bio.caltech.edu ? Manager, sequence analysis facility, biology division, Caltech      # 0.  Edit Makefile.. to make_vms.com A 1.  dheap.c.  All uses of NULL were wrong, changed all NULL to 0. I     Renamed original to dheap.c_dist (ditto for all changes that follow.) = 2.  finalsites.c.  One use of NULL where 0 needed.  Line was:   7        for(mean = 0.0 (site = delminHeap(H)) != NULL ){        became:    4        for(mean = 0.0 (site = delminHeap(H)) != 0 ){  ( 3.  data.c,  three improper uses of NULL    <            if(SeqI(P->entity[i]) !=NULL) PutPopE(fptr, i,P);            if(SeqI(E) != NULL){ 9            if(item==NULL) pop_error("shufflepop2 error");        Also, it had the line:  :                      insrtHeap(item,((keytyp)random()),H);  /     Whereas all other routines had things like:   >          { S[i]=E->S[i];insrtHeap(i,((keytyp)RandomNum()),H);}  C     The linker couldn't find "random".  Replaced it with RandomNum.   D 4.  stdinc.h, added a couple of #includes in stdinc.h (stdlib, time),     so that standard functions were defined.  H 5.  Randomseed, RandomNum generating tons of implicit function warnings.D     Aha, random.h isn't getting included.  Put that too in stdinc.h.   6.  This leaves 3 errors:      GetFreqProb 
     seq_error      GetFinalSites   J    all are declared int, but return no value.  However, they are used like8    voids, so it doesn't matter.  Live with the warnings.  N 7.  Now, why isn't define/user sys$output whatever  creating an output file???I     Ugh, some of the output is going to stdout, the rest to stderr.  The  G     stuff going to stderr is the "this cycle, that value" information,  B     whereas the true output gotes to stdout.  Best to define/user 0     sys$output and sys$error to different files.     I 8.  Now set compiler back to ansi89 from relaxed. Do it again.  Picked up H     only one new implicitly defined function, lgamma in models.c.  Hmm, H     this is part of the C RTL, need to include <math.h>. It is included.I     Hmm, oh, ok, it's one of those __HIDE_FORBIDDEN_NAMES problems, undef 2     that on the command line.  Yes, that fixes it.  @ 9.  Put together GIBBS.COM, a procedure that wraps the GIBBS.EXE     program.  J 10. Put togher DEMO_VMS.COM (from demo..).  This just runs all of the demo     data sets.  J 11. Checked with author - the variation seen in results (slight shifts) isL     expected, can reduce this variation by increasing some of the parameters     like cycles and such.   # 12. Ok, distribute this like it is. 