[rrd-users] why does/should rrd_create() alter argv?

[Rick Jones]

I have code that is creating a new RRD via rrd_update() (1.4.3 via
Ubuntu) and I'm at a loss as to why rrd_update is mucking with the argv
I pass it.  Here is the code snippet:

    optind = 0;
    opterr = 0;

    rrd_clear_error();

    tell_them(LOG_INFO,
	      "%s creating new RRD %s\n",
	      __FUNCTION__,
	      argv[1]);

    rc = rrd_create(position,argv);

    if (rrd_test_error()) {
      char *err = rrd_get_error();
      tell_them(LOG_WARNING,
		"%s() rrd_create() returned '%s'\n",
		__FUNCTION__,
		err);
      context->rrd_errors++;
      line = __LINE__;
      goto fail;
    }

    tell_them(LOG_INFO,
	      "%s added new rrd %s\n",
	      __FUNCTION__,
	      argv[1]);

and here is the resulting output:

2011-04-27-13:19:27.447202: stat_and_create_generic_counter_db creating
new RRD /tmp/counters/192.168.1.7/0/32/generic.rrd
2011-04-27-13:19:27.627944: stat_and_create_generic_counter_db added new
rrd --start

before the call, argv[1] was the filename.  after the call it is
"--start" that was in argv[2].  As I am filling argv with dynamically
allocated strings I wish to later free() I'm concerned there is a memory
leak since rrd_create() is shuffling things around.

It seems all the more strange given that rrd_update() does not seem to
do the same thing - at least not to the filename argument.  Here is
output of a similar code snippet bracketing rrd_update():

2011-04-27-13:28:16.785560: update_generic_counter_db updating
RRD /tmp/counters/192.168.1.7/0/19/generic.rrd using arrival time
1303936096.785111 truncated to millseconds
2011-04-27-13:28:16.785646: update_generic_counter_db updated
RRD /tmp/counters/192.168.1.7/0/19/generic.rrd using arrival time
1303936096.785111 truncated to millseconds

rick jones