[
https://issues.apache.org/jira/browse/ORC-203?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16555947#comment-16555947
]
ASF GitHub Bot commented on ORC-203:
------------------------------------
Github user omalley commented on a diff in the pull request:
https://github.com/apache/orc/pull/292#discussion_r205180384
--- Diff: java/core/src/java/org/apache/orc/impl/ColumnStatisticsImpl.java
---
@@ -517,10 +519,14 @@ public int hashCode() {
protected static final class StringStatisticsImpl extends
ColumnStatisticsImpl
implements StringColumnStatistics {
+ public static final int MAX_STRING_LENGTH_RECORDED = 1024;
--- End diff --
We should probably use the number of bytes instead of the number of
characters. I'd propose that we always truncate a whole number of characters,
with a maximum of 1024 bytes.
For ASCII strings, it will have the same result, but larger characters may
end up at 1021 to 1024 bytes long.
> Modify the StringStatistics to trim minimum and maximum values
> --------------------------------------------------------------
>
> Key: ORC-203
> URL: https://issues.apache.org/jira/browse/ORC-203
> Project: ORC
> Issue Type: Bug
> Reporter: Owen O'Malley
> Assignee: Sandeep More
> Priority: Major
>
> Currently the StringStatistics will record the entire value for minimum or
> maximum. It creates large protobuf objects and serves very little value. I
> think we should trim long strings to 1024 characters and record the fact that
> they were trimmed.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
