|
 |
Hi,
since pvmpov still is version 3.1g from a FreeBSD port, I self-compiled
pvmpov 3.5. Unfortunately, this dumps core on my slave(s). I searched
for the reason and it turned out that the error occurs while trying to
init the display (which I thought I turned off by using -D on the
command line and Display=Off in global povray.ini).
The reason seems to be that ttyname(0) in unix.cpp's SVGAMode() returns
a NULL pointer (on FreeBSD 4.10-RELEASE-p3) which makes strstr() fail.
The attached patch checks for that so that I don't get coredumps anymore
and it works... ;-)
--- unix.cpp.orig Fri Sep 24 15:40:44 2004
+++ unix.cpp Fri Sep 24 15:33:02 2004
@@ -1838,7 +1838,7 @@
return (false);
}
/* if this isn't actually a TTY, no need for SVGA */
- if (!strstr(ttyname(0), "tty"))
+ if (ttyname(0) != NULL && !strstr(ttyname(0), "tty"))
{
return (false);
}
HTH & HAND,
bye, Rocco
--
:wq!
Post a reply to this message
|
|
|
|
> Thanks for pointing this out. pvmpov still is based on 3.5 which is also
> quite old... so I wonder if there're people using pvmpov at all (or if
> there're better alternatives).
I would personnaly prefer a MPI patch of 3.6 (over PVM, which,
to make it simple, is basically wasting one node).
I'm not aware of such a port yet, 3.6 is still rather new.
Moreover the fundamental problems with e.g. the radiosity feature
are far from being solved.
The best attempt towards improving the radiosity results on
multiple CPUs that I know of is in Francois "Wozzeck" Dispot's
PvMegaPOV patch:
http://www.wozzeck.net/images/pmp/
which is another PVM patch based on MegaPOV 0.7 (therefore
it's older than POV 3.5 too, but more "featured" than 3.1g).
- NC
Post a reply to this message
|
|
