[comp.lsi.cad] SunOS-4.0.3 breaks Magic

seeger@helios.iec.ufl.edu (F. L. Charles Seeger III) (08/08/89)

Has anyone successfully compiled Magic (from the UCB '86 VLSI Tools) under
SunOS-4.0.3 and SunView?  I have gotten what appears to be a good compilation
on a 3/110C-4, but upon execution it fails to draw any color to its initial
window (not even its name-stripe and scrollbars).  The cursor and box appear
just fine (and the mouse manipulates them OK), and the background color in
the window looks OK (i.e. its working in the color bit-planes).  After
creating and sizing the box, a ":paint red" command causes no painting to
the screen, but the program starts chewing up all available cpu resources
(with no disk access and no system calls).  Typing ":" causes the cpu
activity to stop, but it resumes after hitting <CR>.  We had Magic running
properly on this machine (and a 3/60FC-8) for the past 2 years under 3.3
and 3.5, and we have all the pertinent (I think) fixes from Bob Mayo.

It looks like the old SunWindows 2.0 code that Magic is written with has
finally been broken by a new release of SunView.  The symptoms perfectly
match with this description from the SunView 1 Programmer's Guide (p. 483,
Appendix C, "Converting SunWindows Programs to SunView"):

Signals
	If you are catching signals, then you should read the documentation
	on signals in Section 17.2, "Restrictions", in the "Notifier"
	chapter.  There are several that the Notifier now catches on your
	behalf.

	You should no longer be catching SIGWINCH signals.  If you do, your
	program may never appear on the screen as it will start catching
	the signals and redrawing endlessly on the screen, which may not be
	visible.


Fortunately, Sun has provided a way to interface to the SunView Notifier
without completely converting the control flow of the program to be
notification based.  Nonetheless, the port looks like a good 1-4 week
project for a novice SunView programmer (e.g. me), depending on how much
you want to modernize the program.  With a new release of Magic in the
works and with everyone (including Sun) moving away from SunView, I don't
want to spend a lot of time "improving" and modernizing the code.  Another
problem with the port is that I have no docs on SunWindows 2.0 (the SunView
manual suggests keeping the old manuals, but we never had them).

I would like to hear from anyone who has or has not seen this problem with
Magic under 4.0.3.  Maybe (I hope), there is something very fundamental
that is escaping me.  I would *love* to hear from anyone who has either
solved this problem or is working on it.  Any fix that I might generate
will be massive and will be unlikely to appear for at least a couple of
weeks.

Regards,
Chuck
--
  Charles Seeger            216 Larsen Hall             +1 904 392 8935
  Electrical Engineering    University of Florida       "Bye, Opus.
  seeger@iec.ufl.edu        Gainesville, FL 32611       It's been fun."