Using os:cmd to in escript to start Erlang application fails.

Question

I have an Erlang application named tb that runs fine from Erlang command line by doing application:start(tb). Whereas when I try to invoke the same application from inside escript using os:cmd, the application doesn't seem to run. When i do a 'ps | grep beam', I see the beam.smp process running. But the application is not generating any output.What might be the problem? Is there a better way to start another erlang VM from inside escript?

Here's the code snippet:

net_kernel:start([tb_escript, shortnames]),
read_config_file(FName),
Cookie = get(cookie),
Node = get(node),
N = io_lib:format("~p",[Node]),
lists:flatten(N),
C = io_lib:format("~p",[Cookie]),
lists:flatten(C),
EBIN = "~/tb/ebin",
erlang:set_cookie(tb_escript,Cookie), 
os:cmd("erl -pa " ++ EBIN ++ " -sname " ++ N ++ " -detached " ++ " -setcookie " ++ C ++ " -s application start tb").

Answer 1

This happens because the args flag to -s wraps the arguments in a list and passes that to module:function/1. -s application start tb will execute application:start([tb]), which would return {error,{bad_application,[ssl]}}. As this is just a normal return value, no error is printed by erl.

From the erl documentation:

-s Mod [Func [Arg1, Arg2, ...]](init flag)

Makes init call the specified function. Func defaults to start. If no arguments are provided, the function is assumed to be of arity 0. Otherwise it is assumed to be of arity 1, taking the list [Arg1,Arg2,...] as argument. All arguments are passed as atoms.

There are two ways to solve this:

1. Use -eval "application:start(tb)", as you already mentioned in a comment.

2. Add a start/0 (if not already present) function to tb which calls application:start(tb), and then pass just -s tb to erl. -s with a single argument will call module:start().