Imported from http://bugzilla.roxen.com/bugzilla/show_bug.cgi?id=1137

Reported by Randall Randall, 3PSecure.com wolfkin@freedomspace.net

In Roxen 2.1.185, when the server is stopped by power failure, Roxen
sometimes (not always) repeatedly fails to start.  It will appear to start
normally, with:
: Removing old .o files (defines or pike version changed)
: Starting the Roxen WebServer.
: Using configuration from ../configurations/, storing the debug log in
../logs//debug/default.1.
: You can use the administration interface in the server to get debug info.
: Forked start script, pid 4047.

...but the Roxen processes continually restart, with this as the
"default.1" log:
12:42:15  : ----------------------------------------------------------
 0m 0.0s  : Pike v7.0 release 268, Roxen WebServer 2.1.185
          : Loading pike modules ... Roxen loader failed:
          : Wrong type of object, cannot convert to mpz.
          : Unknown program: `|(object)
          : base_server/roxenloader.pike:1473: do_main(1,0)
12:42:15  : base_server/roxenloader.pike:1206:
do_main_wrapper(1,({"/usr/local/roxen/server/base_server/roxenloader.pike"}))
 0m 0.6s  :

...and there it stops. Doing a "tail -f default.1" brings this again, but with a different time.

This may not actually be a bug, but even if it is not, it seems to me that Roxen should recover from being killed (as any other Unix program would).