n3nash commented on a change in pull request #674: MOR read query fixes and
performance improvement
URL: https://github.com/apache/incubator-hudi/pull/674#discussion_r283475558
##########
File path:
hoodie-hadoop-mr/src/main/java/com/uber/hoodie/hadoop/realtime/AbstractRealtimeRecordReader.java
##########
@@ -323,13 +340,21 @@ private static Schema addPartitionFields(Schema schema,
List<String> partitionin
* the base split was written.
*/
private void init() throws IOException {
- writerSchema = new AvroSchemaConverter().convert(baseFileSchema);
- List<String> fieldNames =
writerSchema.getFields().stream().map(Field::name).collect(Collectors.toList());
+ Schema schemaFromLogFile = null;
+ HoodieTableMetaClient metaClient = new HoodieTableMetaClient(jobConf,
split.getBasePath());
+ HoodieTimeline completedTimeline =
metaClient.getActiveTimeline().getCommitsTimeline()
+ .filterCompletedInstants();
+ Collections.reverse(split.getDeltaFilePaths());
Review comment:
It is since we get it in the reverse order from HoodieRealtimeInputFormat
class which calls the fileslice.getSplits() which in turn has sorted it
correctly. But that's too much context to keep and can mess up anytime, so
sorting it again here anyways.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
