These are the changes from MH's version 0.60 of dec 93 to my 0.62
Vincent Broman     broman@nosc.mil

Compilation with gcc 2.4.5 under SunOS 4.1.3
and gcc 2.4.x under Linux Slackware 1.0.3 

Linted server and bitblit library, adding many prototypes.
Added several, but not all, dec93 fixes from SAU --  and from MH.
Did not add SAU's code to allow operation with no mouse.  too dangerous.
Iconmail and a few other clients have extra options.
Fake a middle mouse button with -DEMUMIDMSBUT, mouse bug fixes.
Standout mode in color made like mono.  An inverted color was often invisible.
Colors in copyright screen < 24.
Add lots of parentheses.
Posixized some code, but #ifdef'ed on appropriate symbols.
Use ANSI/ISO memset, memcmp, memcpy, strchr, strrchr.
Use fd_set, FD_ISSET, etc for select code.
Some systems might need -DNO_STRCHR or -DNO_SETSID in Configfile MGRFLAGS
Some chars made unsigned.
Made -DDEBUG work again.  debug_flags updated.
A few source formatting problems fixed, particularly if caused by
  someone uncharitably using hardware tabs set at 3 spaces.
Fixed or removed some inappropriate/incorrect casts of 0.
Substituted O_RDONLY for 0 in open, etc.
Fixed include file names in several clients.

Integrated src/libbitblit/sunport with other blit drivers as colorport.
  One bug fix in src/libblit/colorport/blit.C, unimportant?
     No-op statement "src_map == BIT_NULL;"  made == into = .
  (colorport appears to work, line drawing problem in clock client
  may be due to semantics of erasing using xor bitblit.)
Added working cg6 support and untested cg2 support.
  Configfile needs BITBLITFLAGS to include -DCG6 or -DCG3 or -DCG2
320x200x256 VGA modes ought to be almost ready to work,
  if you can get packed mode addressing.  Bank switching needed
  to support super-vga 256-color modes.
Added colormap setting/getting to the server interface and VGA support.
Altered fast_scroll interface, so byte-alignment happens in device driver.
  fast_scrolling is still slow, however.
Moved src/libbitblit/sunport/generic.c to generic.c.old
Fixed a libbitblit bug, the def of BIT_Size, which might have
  given problems to MOVIE MGR.

Added many fonts from Yossi Gil, the X distribution, and elsewhere.
repaired a few fonts with corrupted bits.
added an MGR-HOWTO.
added tex icons and a soundeditor icon.

Updated programming manual.  man page for spot.
New man pages for some minor programs that missed manuals.

New clients mless, mvi, chess, nonmgrterm, pilot, set_colormap2, pgs,
  vgafont2mgr, xbdftomgr.
nonmgrterm used in shell scripts.
Restored snap and c_menu.
C code for resetwin written, and shell script generalized.

set_termcap recognizes the rc shell.
Texmgr bug fixes. not speedy, yet.
bitblit.3 correctly boldened.

----- Remaining problems ------

The VGA colormap setting does not yet work for colors != 0, I think.

On Linux (with my mouse), a window created in the .mgrc file appears empty
on startup, until the mouse is first moved.  Then the shell prompt appears.

The halfwin code doesn't seem to work right.  (in a system dependent way)
This breaks mless and mvi.
It seems to be related to a problem visible in mgr.c
where we loop through the windows to
find a file descriptor that select says is ready to read.
Despite the claim of select, sometimes the read fails with "I/O error",
errno == EIO.  This happens when a halfwindow is first opened,
and also happens when a ^D makes a shell exit.  The man pages for read(2) and
the POSIX specs say this error return indicates either a hardware error
or (under certain conditions) a read from one's controlling tty while in
a background process group.  But I've checked that mgr is in the foreground
process group for it's controlling terminal, and what it was reading was
not a tty, but a pty.

On the Sun, files which include both <termios.h> and <sys/ioctl.h>
have many clashing cpp definitions.  I don't know whether this
is a real problem.
