1.  Obtained code from http://acnuc.univ-lyon1.fr/phylogeny/phylowin.html
    on 17-JAN-1997.
2.  Got it all to compile cleanly with full ANSI C checking on.  Added lots
    of (void) and minor stuff like that.  A few bigger changes.  Originals
    are all in there still .c->.c_dist, .h->.h_dist.
3.  In PHYLO_WIN.C the three symbols PROT,TREEFILE,COMB cause problems
    because there are also global prot(), treefile[], and comb[].  Renamed
    these three booleans by appending a B to each.  YN also a problem,
    conflicts with some math library entry point, changed it to YES_NO.
4.  Need to define logical PHYLO_WIN_DIR so that the program can find its
    help file.  Did that in SYLOGIN.COM, other sites can do it in the
    users login.com if they want to.
5.  A few other VMS vs. Unix changes (diff the files).
6.  The save versions stuff is bogus, it tries to write a file like
    name~version#~, but name already INCLUDES the VMS version number.
    For VMS it should strip at the ; and just do a save, a new version
    will be automatically created.  Hmm, this sort of code occurs in 4 
    places, it should be localized to a function.  All have lines like:

      sprintf(outfile, "%s~%d~", infile_old, numsave);

    localized it to newversionname().

7.  The MISC /getinfo menu is dumb, if no sequences are selected it
    tried to calculate %G, which involves a divide by zero and BANG.
    It should check to see that there is some sequence selected!!!
    THis is in PHYLO_WIN\geni_act at:

     compall[i]/=totlgcomp;

    This is a bug in the original code, let the author fix it.

8.  Interesting, phylo_win is so huge that it seems to somehow disable
    the ANSI C checking many lines (see 6) should have been flagged,
    but weren't.  Oh yuck, found it, the Xlib.h is doing this.  Ok,
    stuck in the appropriate pragmas to turn error checking back on.

9.  Odd, why won't it open an output file???  MUCH digging around.
    Oh yuck, there is a place that looks for the most recent version on
    unix by opening every version of the file until it fails, and then
    backing off one.  Problem is, it never CLOSES any of those, so that
    the process on OpenVMS eats its entire file open quota, because 
    OS takes care of the versions stuff.  Fixed that for VMS 
    version, Unix version is still broken, it will likely fail when enough
    versions accumulate (and no PURGE on Unix either.)

10. Did not implement the PRINT part for treefile, let them manually
    print the postscript from some other window.  Reason - there could be
    many different commands the user would use to print.
