Issue 85 - Don't remove duplicates for IGate RX>IS direction.

Author: wb2osz

config.c

@@ -869,6 +869,7 @@ void config_init (char *fname, struct audio_s *p_audio_config,
 	p_igate_config->tx_limit_1 = IGATE_TX_LIMIT_1_DEFAULT;
 	p_igate_config->tx_limit_5 = IGATE_TX_LIMIT_5_DEFAULT;
 	p_igate_config->igmsp = 1;
+	p_igate_config->rx2ig_dedupe_time = IGATE_RX2IG_DEDUPE_TIME;
 
 
 	/* People find this confusing. */
@@ -2234,10 +2235,18 @@ void config_init (char *fname, struct audio_s *p_audio_config,
 	        t = split(NULL,0);
 	      }
 	      else if (strcasecmp(t, "DROP") == 0) {
+	        text_color_set(DW_COLOR_ERROR);
+	        dw_printf ("Config file, line %d: Preemptive digipeating DROP option is discouraged.\n", line);
+	        dw_printf ("It can create a via path which is misleading about the actual path taken.\n");
+	        dw_printf ("TRACE is the best choice for this feature.\n");
 	        p_digi_config->preempt[from_chan][to_chan] = PREEMPT_DROP;
 	        t = split(NULL,0);
 	      }
 	      else if (strcasecmp(t, "MARK") == 0) {
+	        text_color_set(DW_COLOR_ERROR);
+	        dw_printf ("Config file, line %d: Preemptive digipeating MARK option is discouraged.\n", line);
+	        dw_printf ("It can create a via path which is misleading about the actual path taken.\n");
+	        dw_printf ("TRACE is the best choice for this feature.\n");
 	        p_digi_config->preempt[from_chan][to_chan] = PREEMPT_MARK;
 	        t = split(NULL,0);
 	      }
@@ -4160,13 +4169,13 @@ void config_init (char *fname, struct audio_s *p_audio_config,
 	        p_igate_config->igmsp = n;
 	      }
 	      else {
-	        p_igate_config->satgate_delay = 1;
+	        p_igate_config->igmsp = 1;
 	        text_color_set(DW_COLOR_ERROR);
                 dw_printf ("Line %d: Unreasonable number of times for message sender position.  Using default 1.\n", line);
 	      }
 	    }
 	    else {
-	      p_igate_config->satgate_delay = 1;
+	      p_igate_config->igmsp = 1;
 	      text_color_set(DW_COLOR_ERROR);
               dw_printf ("Line %d: Missing number of times for message sender position.  Using default 1.\n", line);
 	    }
@@ -4182,6 +4191,9 @@ void config_init (char *fname, struct audio_s *p_audio_config,
 
 	  else if (strcasecmp(t, "SATGATE") == 0) {
 
+	    text_color_set(DW_COLOR_INFO);
+	    dw_printf ("Line %d: SATGATE is pretty useless and will be removed in a future version.\n", line);
+
 	    t = split(NULL,0);
 	    if (t != NULL) {
 

doc/Successful-APRS-IGate-Operation.pdf

@@ -33,6 +33,9 @@
  *		APRS iGate properties
  *		http://wiki.ham.fi/APRS_iGate_properties
  *
+ *		Notes to iGate developers
+ *		https://github.com/hessu/aprsc/blob/master/doc/IGATE-HINTS.md#igates-dropping-duplicate-packets-unnecessarily
+ *
  *		SATgate mode.
  *		http://www.tapr.org/pipermail/aprssig/2016-January/045283.html
  *
@@ -1959,9 +1962,37 @@ static void maybe_xmit_packet_from_igate (char *message, int to_chan)
  *		There is a 1 / 65536 chance of getting a false positive match
  *		which is good enough for this application.
  *
+ *
+ * Original thinking:
+ *
+ *		Occasionally someone will get on one of the discussion groups and say:
+ *		I don't think my IGate is working.  I look at packets, from local stations,
+ *		on aprs.fi or findu.com, and they are always through some other IGate station,
+ *		never mine.
+ *		Then someone has to explain, this is not a valid strategy for analyzing
+ *		everything going thru the network.   The APRS-IS servers drop duplicate
+ *		packets (ignoring the via path) within a 30 second period.  If some
+ *		other IGate gets the same thing there a millisecond faster than you,
+ *		the one you send is discarded.
+ *		In this scenario, it would make sense to perform additional duplicate
+ *		suppression before forwarding RF packets to the Server.
+ *		I don't recall if I saw some specific recommendation to do this or if
+ *		it just seemed like the obvious thing to do to avoid sending useless
+ *		stuff that would just be discarded anyhow.  It seems others came to the
+ *		same conclusion.  http://www.tapr.org/pipermail/aprssig/2016-July/045907.html
+ *
+ * Version 1.5:	Rethink strategy.
+ *
+ *		Issue 85, https://github.com/wb2osz/direwolf/issues/85 ,
+ *		got me thinking about this some more.  Sending more information will
+ *		allow the APRS-IS servers to perform future additional network analysis.
+ *		To make a long story short, the RF>IS direction duplicate checking
+ *		is now disabled.   The code is still there in case I change my mind
+ *		and want to add a configuration option to allow it.  The dedupe
+ *		time is set to 0 which means don't do the checking.
+ *
  *--------------------------------------------------------------------*/
 
-#define RX2IG_DEDUPE_TIME 60		/* Do not send duplicate within 60 seconds. */
 #define RX2IG_HISTORY_MAX 30		/* Remember the last 30 sent to IGate server. */
 
 static int rx2ig_insert_next;
@@ -1982,6 +2013,12 @@ static void rx_to_ig_init (void)
 static void rx_to_ig_remember (packet_t pp)
 {
 
+// No need to save the information if we are not doing duplicate checking.
+
+	if (save_igate_config_p->rx2ig_dedupe_time == 0) {
+	  return;
+	}
+
        	rx2ig_time_stamp[rx2ig_insert_next] = time(NULL);
         rx2ig_checksum[rx2ig_insert_next] = ax25_dedupe_crc(pp);
 
@@ -2031,8 +2068,21 @@ static int rx_to_ig_allow (packet_t pp)
 	  dw_printf ("rx_to_ig_allow? %d \"%s>%s:%s\"\n", crc, src, dest, pinfo);
 	}
 
+
+// Do we have duplicate checking at all in the RF>IS direction?
+
+	if (save_igate_config_p->rx2ig_dedupe_time == 0) {
+	  if (s_debug >= 2) {
+	    text_color_set(DW_COLOR_DEBUG);
+	    dw_printf ("rx_to_ig_allow? YES, no dedupe checking\n");
+	  }
+	  return 1;
+	}
+
+// Yes, check for duplicates within certain time.
+
 	for (j=0; j<RX2IG_HISTORY_MAX; j++) {
-	  if (rx2ig_checksum[j] == crc && rx2ig_time_stamp[j] >= now - RX2IG_DEDUPE_TIME) {
+	  if (rx2ig_checksum[j] == crc && rx2ig_time_stamp[j] >= now - save_igate_config_p->rx2ig_dedupe_time) {
 	    if (s_debug >= 2) {
 	      text_color_set(DW_COLOR_DEBUG);
 	      // could be multiple entries and this might not be the most recent.

doc/User-Guide.pdf

@@ -39,6 +39,9 @@ struct igate_config_s {
 	char t2_passcode[8];		/* Max. 5 digits. Could be "-1". */
 
 	char *t2_filter;		/* Optional filter for IS -> RF direction. */
+					/* This is the "server side" filter. */
+					/* A better name would be subscription or something */
+					/* like that because we can only ask for more. */
 
 /*
  * For transmitting.
@@ -69,6 +72,11 @@ struct igate_config_s {
 					/* We allow additional flexibility of 0 to disable feature */
 					/* or a small number to allow more. */
 
+/*
+ * Receiver to IS data options.
+ */
+	int rx2ig_dedupe_time;		/* seconds.  0 to disable. */
+
 /*
  * Special SATgate mode to delay packets heard directly.
  */
@@ -82,6 +90,10 @@ struct igate_config_s {
 #define IGATE_TX_LIMIT_5_DEFAULT 20
 #define IGATE_TX_LIMIT_5_MAX     80
 
+#define IGATE_RX2IG_DEDUPE_TIME 0		/* Issue 85.  0 means disable dupe checking in RF>IS direction. */
+						/* See comments in rx_to_ig_remember & rx_to_ig_allow. */
+						/* Currently there is no configuration setting to change this. */
+
 #define DEFAULT_SATGATE_DELAY 10
 #define MIN_SATGATE_DELAY 5
 #define MAX_SATGATE_DELAY 30