Partial for for lp bug 796647 - handle DELETED records in ha_rnd_next of backing table

1 files changed, 2 insertions, 9 deletions

diff --git a/storage/oqgraph/ha_oqgraph.cc b/storage/oqgraph/ha_oqgraph.cc
index 8cd15dec2e8..5251d1600eb 100644
--- a/storage/oqgraph/ha_oqgraph.cc
+++ b/storage/oqgraph/ha_oqgraph.cc
@@ -1050,7 +1050,7 @@ int ha_oqgraph::fill_record(byte *record, const open_query::row &row)
 
 int ha_oqgraph::rnd_init(bool scan)
 {
-  edges->file->info(HA_STATUS_VARIABLE); // Fix for bug 1195735 - ensure we operate with up to date count!
+  edges->file->info(HA_STATUS_VARIABLE); // Fix for bug 1195735, hang after truncate table - ensure we operate with up to date count
   edges->prepare_for_position();
   return error_code(graph->random(scan));
 }
@@ -1060,14 +1060,7 @@ int ha_oqgraph::rnd_next(byte *buf)
   int res;
   open_query::row row = {};
 
-  // Problem: bug 1195735 - mysqld hang if we delete * from the underlying table, we get an infinite loop through here
-  // fetch_row() --> fetch_row() --> num_edges() --> _table->file->stats->records
-  // _table is actually a handle on the backing table we just deleted everything from.  so the statistics are out of date.
-  // so we really need to force a refresh on the backing store statistics, when starting a new query
-  // So we probably need to fix this bug in rnd_init
-  // Note, close() never called in between; this would otherwise clean things up
-
-  if (!(res= graph->fetch_row(row))) // FIXME - this called after DELETE FROM graph_base; hangs...
+  if (!(res= graph->fetch_row(row)))
     res= fill_record(buf, row);
   table->status= res ? STATUS_NOT_FOUND: 0;
   return error_code(res);