Look for the EDITOR with autoconf.

Do something with the stuff in the util/ subdirectory.

Check the return value of all strdup() calls; don't assume strdup will
work or that we're running a special version that will call RTError.

Clean up the code so that gcc -O -Wall is (almost) silent.

More syslog() calls.  Lots more.

More conversion utilities to go from unix files to SAT databases.

pipeopen() call need to go back to the "old" way.

satd should be enhanced to allow it to be spawned off from inetd.

satd should ALWAYS call syslog() when it exits.

/etc/satd.pid is not always what we want; /var/run/satd.pid is another
choice.

The documentation needs to be converted to texinfo.

We should consider including send-pr with the distribution.

We need a test to distinguish the SysV setpgrp(void) call from the BSD
setpgrp(pid, pgrp) call.  If <limit.h> exists and defines PID_MAX, we
could check the return value from setpgrp({-1,PID_MAX+1}, getpid()), as
if the return value is not -1 then the system must obviously (hollow
laugh) be SysV, as it ignored our bogus argument.

Raphael and I (and anybody else with either a stake in the outcome or an
opinion) should talk to see if we can come up with a "standard"
directory for SAT and its files to live in.  We do want to rdist some
stuff, and RemoteRun based on relations in "known" places.

The original SAT distribution seemed to require a separate copy of the
source on each target machine, although I got the impression that the
object files could still be put in another directory.  The original
distribution also used some scripts to recompile SAT on all of the
"known" architectures.  While I have not tried to make this capability
work in this release, it should be at worst straightforward, and we'll
have the benefit of being able to compile each architecture from a
single copy of the source.

I could not get autoheader to run under HPUX 8.07 or 9.01 (probably due
to shell problems, but bash didn't help either).

I had a problem with autoconf under IRIX; I had to install and use GNU tr,
as the IRIX version was not properly up-casing the macro names.  If I
were more motivated, I could probably write an autoconf test and find a
suitable alternative to the tr arguments...

autoconf required GNU m4, as advertised!
