[opendx-users] 4.2 problem in script mode

[Lloyd A Treinish]

I've only just starting playing a little with 4.2. I use three platforms here, AIX, RH Linux and Windows. Since most of our production work is on AIX (on a cluster of 4-way SMP machines), I started there, especially thanks to Dave for making a tarball with binaries available. I have observed a few problems. Has anyone else tried 4.2 under AIX 4.3.3?

If I start on my programs with the ui with more than one cpu it hangs when trying to get the exec going. This does not occur with the samples. This appears to be due to the feature added after 4.0.x that affects moderate or larger size networks that has been observed on Linux. Setting DX_SOCKET_BUFSIZE (via the fix that Greg added in January) to something bigger than the 32767 default (e.g., 262144) solves the problem. On Linux, it occurred independently of the -p option.

Running in script mode with p > 1 fails, independent of DX_SOCKET_BUFSIZE and the network or script selected when the script is specified at the command line (i.e., -program or piping the program in). If script mode is started manually with multiple processors and the script is manually started via include, it works. This is clearly a problem if one wants to run production scripts on an SMP machine.

Selecting UI options in an Image window with some programs like view control or rendering mode causes the ui to crash with
Error: Object "" does not have windowed ancestor
Change viewing modes from the window menu works fine. This problem does not occur when running samples selected from the dx main menu program.