Basic "What am I doing wrong?" question (Tuple argument to spawn? Dead pid intead of a runtime error?)

Luke Gorrie <>
Fri Feb 14 19:17:43 CET 2003


"Jonathan Coupe" <> writes:

> Yeah, it's your basic idiotic syntax error. (Doh.) The tip about module is
> nice. What I still don't understand is why I didn't get a runtime error..?

You did, but it occured in the new process, and since it wasn't linked
with anything its crash went unreported.

Links are helpful here. For example:

  1> spawn(nosuchmodule, foo, []).
  <0.29.0>
  2> spawn_link(nosuchmodule, foo, []).
  <0.31.0>
  ** exited: {undef,[{nosuchmodule,foo,[]}]} **

There is also much more detailed error logging available via the OTP
libraries, if you start the SASL application:

  3> application:start(sasl).

Then you can use proc_lib:spawn, which is just like 'spawn' except
that it will log an error if the process crashes:

  4> proc_lib:spawn(nosuchmodule, foo, []).
  <0.53.0>
  5> 
  =CRASH REPORT==== 14-Feb-2003::19:08:35 ===
    crasher:
      pid: <0.53.0>
      registered_name: []
      error_info: {undef,[{nosuchmodule,foo,[]},{proc_lib,init_p,5}]}
      initial_call: {nosuchmodule,foo,[]}
      ancestors: [<0.33.0>]
      messages: []
      links: []
      dictionary: []
      trap_exit: false
      status: running
      heap_size: 233
      stack_size: 23
      reductions: 83
    neighbours:

Note that you don't need a link in this case, since proc_lib's spawn
function will always send a message to SASL's error logger if the
process crashes.

(NB: SASL stands for System Architecture Support Libraries -- but you
don't have to know that.)

-Luke




More information about the erlang-questions mailing list