embulk-input-jdbc 0.7.0 → 0.7.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 5f5b564e877d141a046fdb04fc281bc9730af498
-  data.tar.gz: aab75d9db9406b59446617ce03240b615db1f090
+  metadata.gz: fdf58a6bf2822d6fc08367c222f2cd6eab203d56
+  data.tar.gz: f68be4ab3068b33d4c35b824c4fed9068bd0c4a4
 SHA512:
-  metadata.gz: 3924ef7b5698e9e9aaea6ed11763d3ff38acdb6bd963e352adca85912c6efc8ca12d0d159bf92d2b8e340e93e5585bfee3f668ef895bf7995c7eaf3d29155596
-  data.tar.gz: cadb17a65f0b6a9e658f4d4e830841e7d44662bc668c4964a9c0f67918d836c8f88b9260bfe21ca4991c59a96cc002ba7b51260335a8d28dcd9f9478a0a43f59
+  metadata.gz: 7c52c12c901c33c8e8b1e4dfeb4a7ae68246f9257fcb7a9bf933f63dcb258e7e83e51167709c7ea21bbce92cf595095cb2efbe123cacd498b56359f75c9429f8
+  data.tar.gz: 237ab455d727b5f4831119534860448f06fb1f8ca90286c6f65d6efa232db7c975b3976f40bc8e660d294c9390636549d8d63117139f7b34906e26bc317fccdf

data/src/main/java/org/embulk/input/jdbc/AbstractJdbcInputPlugin.java

@@ -252,31 +252,36 @@ public abstract class AbstractJdbcInputPlugin
         BufferAllocator allocator = task.getBufferAllocator();
         PageBuilder pageBuilder = new PageBuilder(allocator, schema, output);
 
-        try {
-            List<ColumnGetter> getters = newColumnGetters(task, querySchema, pageBuilder);
-
-            try (JdbcInputConnection con = newConnection(task)) {
-                try (BatchSelect cursor = con.newSelectCursor(getQuery(task, con), task.getFetchRows(), task.getSocketTimeout())) {
-                    while (true) {
-                        // TODO run fetch() in another thread asynchronously
-                        // TODO retry fetch() if it failed (maybe order_by is required and unique_column(s) option is also required)
-                        boolean cont = fetch(cursor, getters, pageBuilder);
-                        if (!cont) {
-                            break;
-                        }
+        try (JdbcInputConnection con = newConnection(task)) {
+            try (BatchSelect cursor = con.newSelectCursor(getQuery(task, con), task.getFetchRows(), task.getSocketTimeout())) {
+                List<ColumnGetter> getters = newColumnGetters(task, querySchema, pageBuilder);
+                while (true) {
+                    // TODO run fetch() in another thread asynchronously
+                    // TODO retry fetch() if it failed (maybe order_by is required and unique_column(s) option is also required)
+                    boolean cont = fetch(cursor, getters, pageBuilder);
+                    if (!cont) {
+                        break;
                     }
                 }
-
-                if (task.getAfterSelect().isPresent()) {
-                    con.executeUpdate(task.getAfterSelect().get());
-                    con.connection.commit();
-                }
             }
-
+            pageBuilder.finish();
+
+            // after_select runs after pageBuilder.finish because pageBuilder.finish may fail.
+            // TODO Output plugin's transaction might still fail. In that case, after_select is
+            //      already done but output plugin didn't commit the data to the target storage.
+            //      This means inconsistency between data source and destination. To avoid this
+            //      issue, we need another option like `after_commit` that runs after output plugin's
+            //      commit. after_commit can't run in the same transaction with SELECT. So,
+            //      after_select gets values and store them in TaskReport, and after_commit take
+            //      them as placeholder. Or, after_select puts values to an intermediate table, and
+            //      after_commit moves those values to the actual table.
+            if (task.getAfterSelect().isPresent()) {
+                con.executeUpdate(task.getAfterSelect().get());
+                con.connection.commit();
+            }
         } catch (SQLException ex) {
             throw Throwables.propagate(ex);
         }
-        pageBuilder.finish();
 
         TaskReport report = Exec.newTaskReport();
         // TODO

data/src/main/java/org/embulk/input/jdbc/getter/AbstractColumnGetter.java

@@ -57,7 +57,10 @@ public abstract class AbstractColumnGetter implements ColumnGetter, ColumnVisito
     }
 
     @Override
-    public void jsonColumn(Column column) { to.setNull(column); }
+    public void jsonColumn(Column column)
+    {
+        to.setNull(column);
+    }
 
     @Override
     public void timestampColumn(Column column)

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: embulk-input-jdbc
 version: !ruby/object:Gem::Version
-  version: 0.7.0
+  version: 0.7.1
 platform: ruby
 authors:
 - Sadayuki Furuhashi
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2016-03-29 00:00:00.000000000 Z
+date: 2016-05-06 00:00:00.000000000 Z
 dependencies: []
 description: Selects records from a table.
 email:
@@ -42,7 +42,7 @@ files:
 - src/main/java/org/embulk/input/jdbc/getter/TimeColumnGetter.java
 - src/main/java/org/embulk/input/jdbc/getter/TimestampColumnGetter.java
 - src/test/java/org/embulk/input/EmbulkPluginTester.java
-- classpath/embulk-input-jdbc-0.7.0.jar
+- classpath/embulk-input-jdbc-0.7.1.jar
 homepage: https://github.com/embulk/embulk-input-jdbc
 licenses:
 - Apache 2.0