[Openmcl-devel] Faceless Background Server Behavior
gb at clozure.com
Wed Jan 8 14:46:14 EST 2003
When I do this (using tcsh), I see:
[src/ccl] gb at dervish> nohup openmcl -l tiny-server -e '(spawn-tiny-server)' &
[src/ccl] gb at dervish>
[src/ccl] gb at dervish> Welcome to OpenMCL Version (Beta: Darwin) 0.14!
 + Suspended (tty input) openmcl -l tiny-server -e (spawn-tiny-server)
[src/ccl] gb at dervish>
I don't think that it should be reading from the tty unless/until #_select
tells it that it can do so without blocking (perhaps #_select is telling
it that there's a pending EOF or some other exceptional condition; I'm
0.13.2 shows about the same symptoms.
We don't really want 0.13.2's initial process to enter a read-eval-print
loop after spawing the tiny server (or 0.14's initial process to spawn
a read-eval-print-loop after spawing the tiny server.) We could
probably do SAVE-APPLICATION with the right sort of :TOPLEVEL-FUNCTION
argument so that that read-eval-print loop is never entered; another
approach (that's not -quite- right, but is a lot closer) is to keep
the process that executes command-line-args from ever getting around
to creating/entering a read-eval-print loop.
[src/ccl] gb at dervish> nohup openmcl -l tiny-server -e '(spawn-tiny-server)' -e '(loop (sleep 5))' &
[src/ccl] gb at dervish> telnet localhost 1701
telnet: connect to address ::1: Connection refused
Connected to localhost.
Escape character is '^]'.
Connection closed by foreign host.
[src/ccl] gb at dervish> ps
PID TT STAT TIME COMMAND
418 p1 S+ 0:00.57 -tcsh (tcsh)
10803 std Ss 0:00.15 -csh (tcsh)
10898 std R 0:05.79 /usr/local/src/ccl-0.13/ccl/dppccl -l tiny-server -e
I think that there's probably a bug if the listener process is actually
trying to read from a (detached) tty; I'll try to look into that.
I think that what we'd want here is an easier way to say "don't create/enter
a read-eval-print loop", since the best we could hope for if a r-e-p-l was
entered is to see a prompt and have it wait infinitely.
A faceless background process should probably also either allow itself to be
terminated by SIGINT or ignore it completely. (Entering a break loop is
probably pretty low on the list of things it should do, and I have a hunch
[gb at dervish] kill -INT 10898
would be fun to watch, at least the first time.)
Perhaps invoking 'openmcl --server ...' should do some/all of these things
for you ?
[Note that 'nohup' is a shell builtin in tcsh; the behavior I saw may be
a little different from what you saw if you used a different shell.]
Openmcl-devel mailing list
Openmcl-devel at clozure.com
More information about the Openmcl-devel