⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 pointrelnotes.txt

📁 vxworks下的一个嵌入式实时数据库,可以分析源码.
💻 TXT
📖 第 1 页 / 共 5 页
字号:
	   SOLID POINT RELEASE	      RELEASE NOTES PRODUCT.....: BoostEngine 5.0VERSION.....: 05.00.002BUGS FIXED..: ENHANCEMENTS: PLATFORMS...: VxWorks, X86RELEASE DATE: Nov 27, 2006FILES INCLUDED--------------If the file names below contain "***", it represents a platform-specific 3 to 5 character sequence, e.g.    h1x = HP-UX 11 32 bit    h1x64 = HP-UX 11 64bit (PA-RISC)    hia64 = HP-UX 11 64bit (IA64)    l2x = Linux    w32 = MS-Windows NT/2000/XP    s9x = Solaris 9.xFilename extensions vary.  For example, the solidac file might be named 	solidac.dll	solidac.so	etc.Extensions for dynamic libraries: .dll, .so, .slExtensions for static libraries: .lib, .a+-----------------------------+-------------------+------------------+|  Description                |  Windows          |  Unix (usually)  |++----------------------------+-------------------+------------------+ | Binaries +-----------  Solid Server                  solid.exe           solid  Solid Data Definition List    soldd.exe           soldd  Solid Export                  solexp.exe          solexp  Solid Loader                  solload.exe         solload | Static libraries +-------------------  Solid Accelerator library     solidac.lib         solidac.a  ODBC library (ANSI)           solidimpodbca.lib   solidodbca.a  ODBC library (Unicode)        solidimpodbcu.lib   solidodbcu.a  Solid SA library              solidimpsa.lib      solidsa.a | Dynamic libraries +--------------------  Solid Accelerator library     ssolidac41.dll      ssolidac41.so  ODBC library (ANSI)           sacw3241.dll        sac***41.so  ODBC library (Unicode)        socw3241.dll        soc***41.so  Solid SA library              ssaw3241.dll        ssa***41.so | Header files				 +---------------  SQL header file		sqlext.h	    sqlext.h | JDBC	driver package			 +----------------------  JDBC Driver Java package	SolidDriver2.0.jar  SolidDriver2.0.jarPRODUCT ENHANCEMENTS--------------------SHORT DESCRIPTION: Previous logfile default mode (ping-pong) could causefilesystem corruption on VxWorks.TPR #: N/ALONG DESCRIPTION: Logwrite mode default was changed from ping-pong tooverwrite because ping-pong logwrite mode sometimes leaves wholes to thefilesystem and VxWorks filesystem cannot handle that. The ping-pong mode wasdisabled completely.EFFECTS TO OPERATION:Memory: memory consumption and usage is not affected.Performance: Overwrite mode for logwrite gives a bit better performance.================================================================================================== PREVIOUS POINT RELEASES FOR THIS PRODUCT ===========================================================================================Each point release is cumulative regardless of release platform (excluding platform specific problems).  Below are the bug fixes, feature enhancements, and documentation corrections from previous point releases for BoostEngine 4.1.SHORT DESCRIPTION: SSCStartServer() returns SSC_ERROR at the situation ofbroken netcopy when it should return SSC_CORRUPTED_DBFILE.TPR #: N/ALONG DESCRIPTION: SSCStartServer() has a return value for the situation wherethe database file is corrupted. This was not returned if server detects fileas a result of broken netcopy.EFFECTS TO OPERATION:General: only situation where the database is being started with database resulted from broken netcopy operation is affected.Memory: memory consumption and usage is not affected.Performance: performance is not affected.SHORT DESCRIPTION: Shutting down server failed at some cases due to inappropriate state of main memory engine.TPR #: N/ALONG DESCRIPTION: ss_startupstate must be set to SS_STARTUP_NORMAL statebefore shutting down because MME shuts down only if the state is SS_STARTUP_NORMAL.EFFECTS TO OPERATION:General: only shutting down the server is affected.Memory: memory consumption and usage is not affected.Performance: performance is not affected.SHORT DESCRIPTION: Memory leak in Solid.TPR #: 300065.LONG DESCRIPTION: This was not actually a bug in Solid but we added possibility to change memory allocation strategy in the Accelerator server. Allocator type can be changed using function SSCSetMemoryAllocatorType(SscMemoryAllocatorType type) which must be called before any other Solid function, type cannot be changed after the first allocation is done. Possible types are Thread Local, Global and SysMem. Enum values for those are defined in sscapi.h. Thread Local is default Solid qmem behaviour. Global uses ssqmem but does not use thread local context and all allocations go to global pool. SysMem calls directly system memory allocator.SHORT DESCRIPTION: Solid to exit gracefully instead of asserting.TPR #: 300067.LONG DESCRIPTION: With this fix restricted ability to recover from fatalerrors like running out of memory or failing disk operations is added.SHORT DESCRIPTION: taskDelay(0) does not really yield in VxWorks.TPR #: N/A.LONG DESCRIPTION: taskDelay(0) relinguishes cpu but the task is put to thehead of queue of waiting tasks. This means that the same task gets cpuagain. Fixed by changing taskDelay(0) in SsThrSwitch() to taskDelay(1).SHORT DESCRIPTION: SQLGetFunctions() failed with indexes >= odbc's extended functions.TPR #: N/A.LONG DESCRIPTION: Fixed by fixing the calculation of biggest index.SHORT DESCRIPTION: Incorrect primary key unique constraint violation.TPR #: 870106.LONG DESCRIPTION: Fixed with cursor reset fixes.SHORT DESCRIPTION: Merge takes all the CPU time.TPR #: 300061.LONG DESCRIPTION: Merge operation sleeps periodically to give processor forother tasks.SHORT DESCRIPTION: admin command "bak [-s] [directory].TPR #: 453224.LONG DESCRIPTION: pathprefix added now only to relative pathnames on VxWorks. File is created with command creat() on VxWorks as with network drives filecreation with open() fails.SHORT DESCRIPTION: Sync operation reserves cpu for too long time.TPR #: 453119.LONG DESCRIPTION:  Added sleep -mechanism to master end to force yield duringsync operations.SHORT DESCRIPTION: HSB recovery modified.TPR #: 453162, 453172LONG DESCRIPTION: Added support for 'HSB phase2-recovery'. Now the hsb    recovery is performed in two phases.    In the first phase both nodes perform the recovery to the last common    durable-mark.    The second phase handles the log records after the last common    durable position. They are used only if the role changes from SECONDARY    to PRIMARY or STANDALONE. In other cases (the role is SECONDARY) the    rest of the log is cleaned.SHORT DESCRIPTION: New return codes for SSCStartServer() to handle file corruptionTPR #: 880084.LONG DESCRIPTION: Added new SSCStartServer() return code:        SSC_CORRUPTED_DBFILE - when Solid detects corrupted database file and fails to start        SSC_CORRUPTED_LOGFILE  - when Solid detects corrupted transaction log and fails to start.SHORT DESCRIPTION: iomgr_thread hogging CPU.TPR #: 453071.LONG DESCRIPTION: VxWorks 5.5 does use sync instead of close-open -sequence for syncing files to disk.SHORT DESCRIPTION: SSCStartServer to return an error code when cannot start.TPR #: 453074.LONG DESCRIPTION: Added server startup mechanism to return normally without assert exit if server startup fails. This should help Accelerator to return proper error code from SSCStartServer..SHORT DESCRIPTION: Abnormal disconnect from secondary during sync history activation.TPR #: 710069.LONG DESCRIPTION: Sleep and retry send() if the send() fails with EAGAIN error.SHORT DESCRIPTION: HSB disconnect fails to detect disconnect afterfailing connection.TPR #: 452983  and 452985.LONG DESCRIPTION: Fixed primary active side 'hsb disconnect' that hangs if secondary server is non-responsive and fix to 'hsb set primary alone'. Now itgoes directly to alone mode even when in primary active. Previously a disconnect was done. If disconnect failed server went to uncertain mode. Also it was possible for 'set alone' to last several seconds if secondary was not alive.SHORT DESCRIPTION: Fix to HSB ping.TPR #: 870065SHORT DESCRIPTION: Fixed possible memory overwrite in SsBExpand().TPR #: 300040LONG DESCRIPTION: SsBExpand() could cause memory corruption when expanding db-file.SHORT DESCRIPTION: Allocations from global memory pool while using SsSysMem.TPR #: 870051.LONG DESCRIPTION: SsSysRes -object allocated still from global memory pool.SHORT DESCRIPTION: Bug with sysmem feature (see below).TPR #: N/ALONG DESCRIPTION: Pointer for memPartLib was allocatedincorrectly. Memory pool allocation did not work correctly.SHORT DESCRIPTION: Memory Fragmentation in VxWorks.TPR #: 870035LONG DESCRIPTION: Memory fragmentation caused by Solid causes a problemallocating a block of memory requested by MFP system program.(Unable to allocate a requested block of memory). The solution has been implementedusing VxWorks's memPartLib (see VxWorks ref.manual). Usage: before startingthe server call function SsSysMemGlobalInit(<size>). For exampleSsSysMemGlobalInit(30000000).PRODUCT.....: BoostEngine 4.1VERSION.....: 04.10.253BUGS FIXED..: ENHANCEMENTS: 1 (#870081)PLATFORMS...: hia64RELEASE DATE: January 16, 2006SHORT DESCRIPTION: HSB connect is rejected even when the netcopy completed.TPR #: 870081LONG DESCRIPTION: Now 'status copy' returns ACTIVE also when secondary is      loading the database. Eg. changed state string from      'Secondary database starting' to 'ACTIVE'.PRODUCT.....: BoostEngine 4.1VERSION.....: 04.10.251BUGS FIXED..: ENHANCEMENTS: 1 (#870081)PLATFORMS...: hia64RELEASE DATE: December 23, 2005SHORT DESCRIPTION: HSB connect is rejected even when the netcopy completed.TPR #: 870081LONG DESCRIPTION: The problem was that 'hsb status copy' returned     SUCCESS when the actual copy was done. Loading the database was     not included in the copy process.  Now 'hsb status copy' will     return ACTIVE until secondary has received and loaded the     database and is ready for 'hsb connect' and user activity.     Eg. when 'hsb status copy' returns SUCCESS user can issue 'hsb     connect' command and applications can connect to secondary node.PRODUCT.....: BoostEngine 4.1VERSION.....: 04.10.248BUGS FIXED..: 1 (#710151)ENHANCEMENTS: 2PLATFORMS...: hia64RELEASE DATE: October 25, 2005BUG FIXES---------SHORT DESCRIPTION: Unusual occurrence of 'OLD_OFFICIAL_UNIQUE_DELETE' in a    normal refresh.TPR #: 710151LONG DESCRIPTION: When a row was moved from bonsai tree to permanent tree     update information was lost. Because of this bug update trigger was not     able to raise UPDATE value for operation type if row was merged.PRODUCT ENHANCEMENTS--------------------SHORT DESCRIPTION: Profile outputTPR #: noneLONG DESCRIPTION: Added profile output with profile limit. It prints profile     output from selected profile points when profile time limit is exceeded.     This can be enabled using server command line using option         -D/LOG/NOD/PROF:100/THR/LIM:8000000    The output will be printed into file ssdebug.out. In the example    all profile points that take more than 100 milliseconds are    printed. General syntax is               /PROF:<milliseconds>SHORT DESCRIPTION: New pmon counters	TPR #: noneLONG DESCRIPTION: Added new pmon counters to count different transaction lock     operations. The counted operations are ok, timeout, deadlock and    wait. It can be used e.g. to monitor MME lock manager behaviour.PRODUCT.....: BoostEngine 4.1VERSION.....: 04.10.245BUGS FIXED..: 1 (#452810)ENHANCEMENTS: PLATFORMS...: w32, hia64, jdbcRELEASE DATE: September 5, 2005SHORT DESCRIPTION: Database engine crash.TPR #: 452810LONG DESCRIPTION: Fix to a problem where RPC broken was not detected properly.     It caused server crash in case of batch statement execute.PRODUCT.....: BoostEngine 4.1VERSION.....: 04.10.243BUGS FIXED..: 4 (#12031, 12099, 710069, 710137)ENHANCEMENTS: PLATFORMS...: w32, l2x, jdbcRELEASE DATE: August 31, 2005BUG FIXES---------SHORT DESCRIPTION: Displaying APPINFO information in the solmsg.out.TPR #: 12031LONG DESCRIPTION: In addition to messages 30001 and 30009, a APPINFO is    added to messages 30002, 30003, 30004, 30007 and 30008. Now all    messages are added. Previous versions add only some of them.   SHORT DESCRIPTION: Abnormal disconnect from secondary during sync history     activationTPR #: 12099, 710069, 710137LONG DESCRIPTION: Primary is sending data to secondary when the write    fails with errno EAGAIN. Now this error code is handled and write    is retried instead of just disconnecting the socket.SHORT DESCRIPTION: Solid cores on Fedora 3 (glibc 2.3.4) when server    is not found.TPR #: 452890LONG DESCRIPTION: This is a problem with glibc compatiblity, that    was caused by a change in the size of pthread_cond_t data    structure.SHORT DESCRIPTION: Solid ODBC driver cores on Fedora 3 (glibc 2.3.4)     when SQLConnect() fails.TPR #: 452892LONG DESCRIPTION: This is a problem with glibc compatiblity, that    was caused by a change in the size of pthread_cond_t data    structure. Same as #452890.PRODUCT.....: BoostEngine 4.1VERSION.....: 04.10.235BUGS FIXED..: 1 (#870067)ENHANCEMENTS: 

⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -