[720] in Zephyr_Bugs
Zephyr stuff
daemon@ATHENA.MIT.EDU (Dejan Vucinic)
Wed Nov 1 09:36:19 1995
Date: Wed, 01 Nov 1995 08:35:39 -0600
From: Dejan Vucinic <dejan@mit1.fnal.gov>
To: bug-zephyr@MIT.EDU
Hello,
Here's a little potpourri for a good morning... :)
1. I successfully built Zephyr 2.0.2 on three different platforms here at
Fermilab, it works like a charm. I then tried the same exhausting procedure
(configure; make; make install) on a Solaris 2.4 machine, and guess what.
I didn't try to figure it out (I copied executables from SunOS 4.something
and they work) but in case it makes any sense to you, configure dies with:
checking size of short... configure: error: can not run test program while
cross compiling
I can send you more details if you tell me where to look for them.
2. In ZOpenPort.c, you create a socket with
if ((__Zephyr_fd = socket(AF_INET, SOCK_DGRAM, 0)) < 0)
Strictly speaking, this is wrong. A socket doesn't know anything about
address families. It should be either PF_INET or PF_UNIX. The fact that
PF_INET and AF_INET are the same number is an accident. However, this
error is so common, if someone changed AF_INET to a different number
e would probably break 80% of networking code in existence today...
3. I wrote a simple Zephyr client for a VxWorks real-time operating system.
We're going to use it as an error-reporting and logging system in the data
acquisition system of the CDF detector. Are you interested in
incorporating it in the source tree? VxWorks is rather peculiar, it's a
single-user multithreaded un*x lookalike without memory protection and with
global namespace. I put the code in a separate directory under 'clients'.
I'm using zephyr_err.h from a SunOS configuration, and I cut and pasted
parts of zephyr.h. The OS is too different from unix to use the complete
zephyr.h from a different target. I suppose the right way to do it would
be to make a configure target, but that's a bitch, every installation of
VxWorks is slightly different, and there's no standard compiler.
The alternative would be to make source available separately from the
Zephyr distribution, since it's a very specialized application. I'll
certainly do this, but it's a tiny piece of code and it would be a pity not
to include it with other clients. I'd even write a decent README. :)
Let me know what you think...
Regards,
Dejan Vucinic
dejan@mit.edu
dejan@mit.fnal.gov
