1. GemStone/S 64 Bit 3.5.3 Release Notes

Overview

Supported Platforms

Platforms for Version 3.5.3

GemBuilder for Smalltalk (GBS) Versions

VSD Version

Changes in this release

Updated library version

Updated Linux platform support

GemStone on Mac is now distributed as an Apple-notarized .dmg

Improvements for using Large Pages

Adjusted values for alignment

largememorypages script on Linux

GsTlsPrivateKey methods renamed for consistent capitalization

External Sessions using Thread-safe API

Added GsObjectInventory methods

Repository scan disallowed while vote is not idle

Performance reimplementation of ReadStream >> upTo:

skip: added to BtreePlusGsReversedIndexReadStream, BtreePlusGsIndexReadStream

Behavior >> isSpecial should not be used; replaced by Behavior >> areInstancesSpecial

Other added methods

Deprecation Change

Removed Private Methods

startstone -A

Configuration Parameter Changes

Added configuration parameter

Changes in STN_STATMONITOR_ARGS with -X

Statmonitor recording some information on NetLDIs

New Cache Statistics

Added GCI Thread-Safe functions

GciTsNbExecute

GciTsNbLogout

GciTsNbPerform

GciTsNbResult

GciTsSocket

Bug Fixes

Memory leaks in NetLDI

Unnecessary LDAP calls from NetLDI in guest mode

Commit performance when using RC collections

Limiting analysis

Temporary objects unnecessarily added to RcReadSet

Symbol GarbageCollection issue

Timeouts during RPC GCI logins

Indexes with optionalPathTerms error on removing elements without the term

FFI CHeader parsing issues

The enum type was not handled correctly

Return type of void* was generated as void

Upgrade Issues

Incorrect version test during upgrade broke session methods

Upgrade did not remove #defaultValues in GcUser's UserGlobals.

Upgrade failed to add newline to DbfHistory

listReferences: results did not respect order of argument

GsObjectInventory could report 0 bytes for some collection classes

Backup made during reclaim may have object audit failures after restore

Solo login when /tmp/gemRO_<pid>_extent1.dbf exists

Slight risk of Gem fatal error on kill -USR1

Checking stack for self may incorrectly report corrupt object error

Cache Statistics Fixes