[interpreter] statement position always overwrites earlier positions.

test/unittests/interpreter/source-position-table-unittest.cc

Index: test/unittests/interpreter/source-position-table-unittest.cc
diff --git a/test/unittests/interpreter/source-position-table-unittest.cc b/test/unittests/interpreter/source-position-table-unittest.cc
index af70179ef454824d7aa2ed96edc759e56b1a2312..d4e21ab24c7c6c036bcdde2ac9e9cab29eee5ab8 100644
--- a/test/unittests/interpreter/source-position-table-unittest.cc
+++ b/test/unittests/interpreter/source-position-table-unittest.cc
@@ -37,15 +37,7 @@ TEST_F(SourcePositionTableTest, EncodeStatementDuplicates) {
   SourcePositionTableBuilder builder(isolate(), zone());
   for (int i = 0; i < arraysize(offsets); i++) {
     builder.AddStatementPosition(offsets[i], offsets[i]);
-    builder.AddStatementPosition(
-        offsets[i], offsets[i],
-        (i % 2 == 0) ? SourcePositionTableBuilder::OVERWRITE_DUPLICATE
-                     : SourcePositionTableBuilder::DISCARD_DUPLICATE);
     builder.AddStatementPosition(offsets[i], offsets[i]);

[vogelheim, 2016/04/04 16:14:31]

Hmm. Does this still do anything useful?

Each pos is added twice, so this stresses the 2nd-statement-wins logic. But
since it adds the same positions each time, the assertions in
ToSourcePositionTable cannot distinguish between an implementation where the 1st
or the 2nd statement wins.

Maybe:
  builder.AddStatementPosition(offsets[i], 2*i);
  builder.AddStatementPosition(offsets[i], 2*i + 1):

That way, the resulting table will be different, depending on what
AddStatementPosition actually does.

-    builder.AddStatementPosition(
-        offsets[i], offsets[i],
-        (i % 2 == 0) ? SourcePositionTableBuilder::OVERWRITE_DUPLICATE
-                     : SourcePositionTableBuilder::DISCARD_DUPLICATE);
   }
 
   // To test correctness, we rely on the assertions in ToSourcePositionTable().