Merge pull request #282 from basho/test/mra/gh280

Add test case and fix for #280

Reviewed-by: javajolt

src/lager.erl

@@ -469,6 +469,10 @@ add_trace_to_loglevel_config(Trace, Sink) ->
     end.
 
 %% @doc recalculate min log level
+update_loglevel_config(error_logger) ->
+    %% Not a sink under our control, part of the Erlang logging
+    %% utility that error_logger_lager_h attaches to
+    true;
 update_loglevel_config(Sink) ->
     {_, Traces} = lager_config:get({Sink, loglevel}, {ignore_me, []}),
     MinLog = minimum_loglevel(get_loglevels(Sink)),

test/zzzz_gh280_crash.erl

@@ -0,0 +1,33 @@
+%% @doc This test is named zzzz_gh280_crash because it has to be run first and tests are run in 
+%% reverse alphabetical order.
+%%
+%% The problem we are attempting to detect here is when log_mf_h is installed as a handler for error_logger
+%% and lager starts up to replace the current handlers with its own.  This causes a start up crash because
+%% OTP error logging modules do not have any notion of a lager-style log level.
+-module(zzzz_gh280_crash).
+-compile(export_all).
+
+-include_lib("eunit/include/eunit.hrl").
+
+gh280_crash_test() ->
+    application:stop(lager),
+    application:stop(goldrush),
+
+    error_logger:tty(false),
+    %% see https://github.com/erlang/otp/blob/maint/lib/stdlib/src/log_mf_h.erl#L81
+    %% for an explanation of the init arguments to log_mf_h
+    ok = gen_event:add_sup_handler(error_logger, log_mf_h, log_mf_h:init("/tmp", 10000, 5)),
+    lager:start(),
+    Result = receive
+        {gen_event_EXIT,log_mf_h,normal} -> 
+            true;
+        {gen_event_EXIT,Handler,Reason} ->
+            {Handler,Reason};
+        X -> 
+            X
+    after 1000 ->
+        timeout
+    end,
+    ?assert(Result),
+    application:stop(lager),
+    application:stop(goldrush).