Add //third_party/sqlite to dirs_to_snapshot, remove net_sql.patch

third_party/sqlite/sqlite-src-3070603/test/delete2.test

Index: third_party/sqlite/sqlite-src-3070603/test/delete2.test
diff --git a/third_party/sqlite/sqlite-src-3070603/test/delete2.test b/third_party/sqlite/sqlite-src-3070603/test/delete2.test
new file mode 100644
index 0000000000000000000000000000000000000000..1668a3a9ae05c82ece49e3d0642531731c5f9135
--- /dev/null
+++ b/third_party/sqlite/sqlite-src-3070603/test/delete2.test
@@ -0,0 +1,119 @@
+# 2003 September 6
+#
+# The author disclaims copyright to this source code.  In place of
+# a legal notice, here is a blessing:
+#
+#    May you do good and not evil.
+#    May you find forgiveness for yourself and forgive others.
+#    May you share freely, never taking more than you give.
+#
+#***********************************************************************
+# This file implements regression tests for SQLite library.  The
+# focus of this script is a test to replicate the bug reported by
+# ticket #842.
+#
+# Ticket #842 was a database corruption problem caused by a DELETE that
+# removed an index entry by not the main table entry.  To recreate the
+# problem do this:
+#
+#   (1) Create a table with an index.  Insert some data into that table.
+#   (2) Start a query on the table but do not complete the query.
+#   (3) Try to delete a single entry from the table.
+#
+# Step 3 will fail because there is still a read cursor on the table.
+# But the database is corrupted by the DELETE.  It turns out that the
+# index entry was deleted first, before the table entry.  And the index
+# delete worked.  Thus an entry was deleted from the index but not from
+# the table.
+#
+# The solution to the problem was to detect that the table is locked
+# before the index entry is deleted.
+#
+# $Id: delete2.test,v 1.8 2008/07/08 15:59:52 danielk1977 Exp $
+#
+
+set testdir [file dirname $argv0]
+source $testdir/tester.tcl
+
+# Create a table that has an index.
+#
+do_test delete2-1.1 {
+  set DB [sqlite3_connection_pointer db]
+  execsql {
+    CREATE TABLE q(s string, id string, constraint pk_q primary key(id));
+    BEGIN;
+    INSERT INTO q(s,id) VALUES('hello','id.1');
+    INSERT INTO q(s,id) VALUES('goodbye','id.2');
+    INSERT INTO q(s,id) VALUES('again','id.3');
+    END;
+    SELECT * FROM q;
+  }
+} {hello id.1 goodbye id.2 again id.3}
+do_test delete2-1.2 {
+  execsql {
+    SELECT * FROM q WHERE id='id.1';
+  }
+} {hello id.1}
+integrity_check delete2-1.3
+
+# Start a query on the table.  The query should not use the index.
+# Do not complete the query, thus leaving the table locked.
+#
+do_test delete2-1.4 {
+  set STMT [sqlite3_prepare $DB {SELECT * FROM q} -1 TAIL]
+  sqlite3_step $STMT
+} SQLITE_ROW
+integrity_check delete2-1.5
+
+# Try to delete a row from the table while a read is in process.
+# As of 2006-08-16, this is allowed.  (It used to fail with SQLITE_LOCKED.)
+#
+do_test delete2-1.6 {
+  catchsql {
+    DELETE FROM q WHERE rowid=1
+  }
+} {0 {}}
+integrity_check delete2-1.7
+do_test delete2-1.8 {
+  execsql {
+    SELECT * FROM q;
+  }
+} {goodbye id.2 again id.3}
+
+# Finalize the query, thus clearing the lock on the table.  Then
+# retry the delete.  The delete should work this time.
+#
+do_test delete2-1.9 {
+  sqlite3_finalize $STMT
+  catchsql {
+    DELETE FROM q WHERE rowid=1
+  }
+} {0 {}}
+integrity_check delete2-1.10
+do_test delete2-1.11 {
+  execsql {
+    SELECT * FROM q;
+  }
+} {goodbye id.2 again id.3}
+
+do_test delete2-2.1 {
+  execsql {
+    CREATE TABLE t1(a, b);
+    CREATE TABLE t2(c, d);
+    INSERT INTO t1 VALUES(1, 2);
+    INSERT INTO t2 VALUES(3, 4);
+    INSERT INTO t2 VALUES(5, 6);
+  }
+} {}
+do_test delete2-2.2 {
+  set res [list]
+  db eval {
+    SELECT CASE WHEN c = 5 THEN b ELSE NULL END AS b, c, d FROM t1, t2
+  } {
+    db eval {DELETE FROM t1}
+    lappend res $b $c $d
+  }
+  set res
+} {{} 3 4 {} 5 6}
+
+finish_test