AccountPatchReview : fix mariadb status matching

mariadb is currently using SQLException.getSQLState() (as used by
getSQLStateInt()) which returns the ANSI SQLSTATE value, not the
MariaDB/mysql error code [1].

The result is that the duplicate key detection doesn't work when using
the mariadb connector.  If you have already reviewed a file (or click
"reviewed" in the file chooser) and then open it again you get a 500
error and a long backtrace including something like

  Caused by: java.sql.SQLIntegrityConstraintViolationException:
  (conn=13) Duplicate entry '3-1-1000001-file-3.txt' for key 'PRIMARY'

We could either continue using getSQLState() and change this to
"23000" (which covers all these values) or do as MySQL connector does
and use the vendor-specific getErrorCode() call.  I've taken the later
approach here.

[1] https://mariadb.com/kb/en/mariadb-error-codes/

Change-Id: I5762118ceb5628171b991a80fe701d160ab0d959

1 files changed, 1 insertions, 1 deletions

diff --git a/java/com/google/gerrit/server/schema/MariaDBAccountPatchReviewStore.java b/java/com/google/gerrit/server/schema/MariaDBAccountPatchReviewStore.java
index b0a3370fe1..5073af15b4 100644
--- a/java/com/google/gerrit/server/schema/MariaDBAccountPatchReviewStore.java
+++ b/java/com/google/gerrit/server/schema/MariaDBAccountPatchReviewStore.java
@@ -37,7 +37,7 @@ public class MariaDBAccountPatchReviewStore extends JdbcAccountPatchReviewStore
 
   @Override
   public StorageException convertError(String op, SQLException err) {
-    switch (getSQLStateInt(err)) {
+    switch (err.getErrorCode()) {
       case 1022: // ER_DUP_KEY
       case 1062: // ER_DUP_ENTRY
       case 1169: // ER_DUP_UNIQUE;