[ https://issues.apache.org/jira/browse/GEODE-2578?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15898168#comment-15898168 ]
ASF GitHub Bot commented on GEODE-2578: --------------------------------------- Github user PivotalSarge commented on the issue: https://github.com/apache/geode-native/pull/44 The unit tests failed on Linux but not other platforms. It turns out that DataOutput has some not-very-obvious (and apparently platform-specific) behavior around buffer management that results in the return values from getBufferLength() and getRemainingBufferLength() depending on the behavior of other unit tests that may precede them. Hence, the test of ginormous query strings was causing the return values of those methods to differ from what would be returned if that test was *not* run. So I reworked how the DataOutput cursor advancement unit tests work and created GEODE-2598 to understand and test how DataOutput's buffer management works. > Query string limited to 64 KiB > ------------------------------ > > Key: GEODE-2578 > URL: https://issues.apache.org/jira/browse/GEODE-2578 > Project: Geode > Issue Type: Bug > Components: native client > Reporter: Michael Dodge > Assignee: Michael Dodge > > The serialization of query strings uses a 16-bit unsigned integer to > represent the length of the query string. Query strings with more than 65535 > characters are silently truncated. Use of a 32-bit unsigned integer to > represent the length would greatly increase the size of query strings that > may be used. -- This message was sent by Atlassian JIRA (v6.3.15#6346)