Mercurial > ublox > ublox8

changeset 74:e1866afc9b4d

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
print-state: don't assert that ephemeris t0 is monotonically increasing There has been at least one instance, where the most recently received ephemeris was 10 minutes older than the previous one. Signed-off-by: Josef 'Jeff' Sipek <jeffpc@josefsipek.net>
author Josef 'Jeff' Sipek <jeffpc@josefsipek.net>
date Wed, 22 Jan 2020 12:53:06 -0500
parents 0bfd3bbea386
children 84848a472137
files print-state.c
diffstat 1 files changed, 44 insertions(+), 3 deletions(-) [+]
line wrap: on
line diff
--- a/print-state.c	Wed Jan 22 13:08:34 2020 -0500
+++ b/print-state.c	Wed Jan 22 12:53:06 2020 -0500
@@ -110,7 +110,24 @@
 		return;
 	} else {
 		/*
-		 * A different ephemeris - must have a newer t0.
+		 * A different ephemeris.
+		 *
+		 * In theory it must have a newer t0.  In practice, there
+		 * have been instances where t0 has went backwards.  For
+		 * example, E31's ephemeris update on 2020-01-22 (at
+		 * 644422793 GST) went from:
+		 *
+		 *    iod = 242
+		 *    t0e = 644422200 (GST)
+		 *
+		 * to:
+		 *
+		 *    iod = 4
+		 *    t0e = 644421600 (GST)
+		 *
+		 * Therefore, we cannot assert that t0 is monotonically
+		 * increasing.  Instead, we just accept the most recently
+		 * received ephemeris but issue a warning.
 		 *
 		 * Note: It is tempting to assert that the iod has changed
 		 * (i.e., is not the same) but that would assume that we
@@ -118,9 +135,33 @@
 		 * to see two ephemeris updates in a row with the same iod
 		 * because we failed to receive all updates in between.
 		 */
-		ASSERT3U(prev->t0.gst, <, new->t0.gst);
+
+		if (prev->t0.gst >= new->t0.gst) {
+			printf("WARNING: E%02u ephemeris update is "
+			       "%u s older than previous (GST t0 %u -> %u): "
+			       "disco %f m @ prev t0, %f m @ new t0\n", new->sv,
+			       prev->t0.gst - new->t0.gst, prev->t0.gst,
+			       new->t0.gst,
+			       calc_disco(prev, new, prev->t0.gst),
+			       calc_disco(prev, new, new->t0.gst));
+			galileo_print_eph(prev);
+			galileo_print_eph(new);
 
-		prev_age = new->t0.gst - prev->t0.gst;
+			/*
+			 * Yes, we break the above rule about not assertig
+			 * that iod changed.  This is temporary, while we
+			 * try to figure out why E07 on 2020-01-20 had an
+			 * ephemeris update that was identical except that:
+			 *
+			 * omega0: 8.414103e-01 -> 8.409633e-01
+			 * i0:     9.532941e-01 -> 9.532922e-01
+			 * omega:  6.120841e-02 -> 7.349972e-02
+			 * idot:  -1.300054e-10 -> 5.571661e-11
+			 */
+			ASSERT3U(prev->iod, !=, new->iod);
+		}
+
+		prev_age = (int32_t) (new->t0.gst - prev->t0.gst);
 
 		/* calculate the discontinuity... */
 		/* ...at t0 for the new ephemeris */