Porting Jove to Silicon Graphics Iris Workstations

[Jim Frost]

In article <21152 at beta.lanl.gov> ttp at beta.lanl.gov (T T Phillips) writes:
|I have been trying to bring up Jove Version 4.9 on a Silicon Graphics
|Iris 3020 workstation.  I get a variety of error messages during the
|compilation of jove.c ending with a ***error code 1, and then it
|stops.  
|
|Has anyone out there ported Jove to Silicon Graphics workstations?
|Any information that you could give me about the problems that you
|encountered (and hopefully solved) would be most welcome.

I tried to port it a couple of months ago (I'm not sure what version)
and was unsuccessful.  Probably most of the problems you're
encountering are System V/BSD differences since jove is a BSD program.
It took me about an hour to resolve the terminal IO differences
between the two.  Unfortunately there was a problem in the jove
implementation of varargs that had me so badly stumped that I gave up
on it; this may not be a problem on the 3000's, I was porting to a 4D
and didn't try to put it on one of our 3000's.

Another thing I remember was that one of the macros was very broken;
there is a difference between the way precedence is handled between
most BSD compilers and most SysV compilers, or at least between the
Sun 3.2 compiler and the SGI compiler I was working with.  It had to
do with indirection in a complex macro; one of the macro parameters
should have been parenthesized.  It worked after I did this (to some
degree).

The version of jove I was porting had horrid code; someone has written
and told me that it's been improved.  I can't verify this without
checking the version number of jove that I was using, though.  I'm
sorry I can't be more specific but we don't seem to have a version of
jove sitting on this cluster.

Oh, could someone PLEASE tell me why varargs was completely
reimplemented?  I can understand reimplementing stdio, but VARARGS?
Eek.

jim frost
madd at bu-it.bu.edu