List of changes for portingset: 8.9a.03 ------------------------------------------------------------------------ Couperin: # 60685 (BDUX53594): blogind6.2 does not set PAM_TTY item: root login not possible when pam_securetty.so is used Date: Fri, 21 Feb 2014 15:03:14 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ 1. Linux which uses pamsecuretty.so in its PAM stack will probibit root login using blogind 2. there is no way to select a different PAM stack for Baan environments _Workaround_ remove pamsecuretty.so from the pam stack or don't use root login in the Baan Environments _Test Procedure_ BSE_PAM_TTY: 0. read defect text carefully!!! 1. create a pam stack with pamsecuretty.so 2. set BSE_PAM_TTY to baan or /dev/baan in $BSE/lib/bse_vars or in the environment where blogind6.2 is started (NOTE: blogind should be up to date!!!) 3. add a baan entry in /etc/securetty 4. try to login as root BSE_PAM_SERVICE 1. create pam entries for new pam service in /etc/pam.conf or copy /etc/pam.d/login to /etc/pam.d/bse_login 2. set BSE_PAM_SERVICE to 'bse_login' in $BSE/lib/bse_vars of the blogin environment or export BSE_PAM_SERVICE=bse_login before starting blogind6.2 3. be sure that blogind6.2 is up to date 4. login should work as usual. _Affected Executables_ blogind badmin _Motive source_ InforXtreme:439137 ------------------------------------------------------------------------ Couperin: # 60683 (BDUX53694): hostname cache does not work correctly Date: Fri, 21 Feb 2014 14:24:01 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ Followup on Create Hostname cache in ipc_name.c _Affected Executables_ all ------------------------------------------------------------------------ Couperin: # 60673 (BDUX53656): Create hostname cache in ipc_name.c Date: Thu, 20 Feb 2014 15:29:38 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ tabledef6.2 with many entries containing a hostname causes a CPU spike with the NSCD (Name Service Cache Daemon) _Test Procedure_ tests_bshell/sock/addrinfo.test _Affected Executables_ all dbc clients (bshell etc) _Motive source_ InforXtreme:439677 ------------------------------------------------------------------------ Couperin: # 60636 (BDUX53595): Bdbreconfig and bdbpost: Automatic enabling of array inserting Date: Tue, 18 Feb 2014 18:02:06 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ Bdbreconfig and bdbpost should automatically enable array inserting. It is possible to enable array inserting in the $BSE/lib/defaults/db_resource file, but this is cumbersome and mistakes are easily made. _Test Procedure_ Set DBSLOG=1570, and import a table using bdbpost. Observe that array inserting is enabled and that the array size can be changed with the table_load_array_size resource. Run the same test with bdbreconfig -Z, which forces an export and import of a table. _Affected Executables_ bdbreconfig bdbpost ------------------------------------------------------------------------ Couperin: # 60631 (BDUX53502): qcheck=nullptr and core dumps Date: Tue, 18 Feb 2014 17:04:18 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ AIX doesn't generate a core dump when expected _Affected Executables_ all _Motive source_ ScrumStory:LNTECH-1852 ------------------------------------------------------------------------ Couperin: # 60627 (BDUX53368): Sprint 2014-2: bdbreconfig Alignment none to left Date: Tue, 18 Feb 2014 16:27:22 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ A simple table reconfiguration at Snap-On tool 139 hours while doing effectively nothing: left-alignment of a field that had zero rows with spaces on the left... _Workaround_ Wait for a long time. _Test Procedure_ tests_reconf/performance/left_align (13 tests) _Affected Executables_ bdbreconfig _Motive source_ ScrumStory:LNTECH-1904 ------------------------------------------------------------------------ Couperin: # 60615 (BDUX53624): Oracle driver: Disable FETCH FIRST n ROWS ONLY for Oracle 12.1 Date: Mon, 17 Feb 2014 15:11:40 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ The FETCH FIRST n ROWS ONLY clause on Oracle version 12.1 gives problems with some queries. An example of a statement is a db.next() call on an MLE table. The problems will be reported to Oracle, and we will attempt to adopt this functionality with version 12.2. _Test Procedure_ Run SELECT statement with AS SET WITH n ROWS on Oracle 12.1, and observe that no FIRST ROWS clause is generated. _Affected Executables_ ora8_srv6.2 ------------------------------------------------------------------------ Couperin: # 60571 (BDUX53545): DB drivers: push AS SET WITH n ROWS to the database as TOP(n) or FETCH FIRST n ROWS ONLY Date: Thu, 13 Feb 2014 16:46:32 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ If a query contains an AS SET WITH n ROWS clause, then this restriction is not pushed to the database. The advantages of doing so are: * the database may choose a more optimal execution plan because it knows that only a few rows must be returned to the application. * if the result set is big and only a few rows are fetched, but the cursor is not closed, then MS SQL Server continues populating the cursor in the background. This unnecessarily consumes CPU and memory resources. _Test Procedure_ On one of the database stated above, write a query with AS SET WITH n ROWS. Run the query with DBSLOG=400 and observe that the resulting query contains the clause that limits the result set. See tests in tests/as_set_with_n_rows. _Affected Executables_ All database drivers _Motive source_ ScrumStory:LNTECH-1855 ------------------------------------------------------------------------ Couperin: # 60562 (BDUX53563): DB drivers: Support array inserts for all bufferable inserts (db.retry flag) Date: Thu, 13 Feb 2014 14:13:01 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ If array inserting is enabled, then only inserts with the flags db.retry and db.skip.dupl are buffered. All other inserts are flushed immediately to the database. It would be good to use array inserts for all bufferable inserts. An insert is bufferable if it has the db.retry flag. Buffering inserts may not have an impact on performance of typical applications, but it does improve the performance of bdbreconfig if the export/import method is used. _Test Procedure_ Write a small program that inserts multiple rows with the db.retry flag. Enable array inserting and set DBSLOG=1570. Observe that array inserting is used. _Affected Executables_ All database drivers ------------------------------------------------------------------------ Couperin: # 60512 (BDNT53479): domain constraint violations are incorrectly logged. Date: Tue, 11 Feb 2014 14:23:35 +0100 (CET) Created on: Danzi Type: bugfix _Problem Description (Customer terms)_ Event viewer shows: The primary key is: _index1={'3501 ',2003,322,9001} but should show: The primary key is: _index1={'3501 ',2003,32,2,9001} _Workaround_ N/A _Affected Executables_ Bshell _Motive source_ InforXtreme:438004 ------------------------------------------------------------------------ Couperin: # 60501 (BDUX53359): Db2 driver: Unicode: Incorrect query results after adding single-byte string field to table Date: Mon, 10 Feb 2014 17:17:14 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ If on Db2 using Unicode a table is extended with a column, subsequent queries may give incorrect results. _Workaround_ Force an export/import of the table (using bdbreconfig -Z for example). _Test Procedure_ Add a column without a default to a table (Unicode mode) and observe that the column is properly padded. Use Db2 tools or use the following statement: bsql6.2 -p16 -q "select '>' & column & '<' from table" There should be spaces between the pointy brackets. _Affected Executables_ db2v5_srv _Motive source_ InforXtreme:429156 ------------------------------------------------------------------------ Couperin: # 60418 (BDNT53315): Wrong expansion with bic -z in combination with BAAN_SCM_GRP in a fd file Date: Wed, 5 Feb 2014 14:36:47 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ Expansion of fd file entries are incorrect when running on Windows. _Workaround_ NA _Affected Executables_ bic, bshell and all other binaries that perform file name expansion using fd files. _Motive source_ InforXtreme:432098 ------------------------------------------------------------------------ Couperin: # 60347 (BDUX53301): bdbpost crashes due to missing ttocm999 Date: Wed, 29 Jan 2014 13:57:18 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ bdbpost crashed on a bse where no COBO table is defined _Affected Executables_ bdbpost6.2 ------------------------------------------------------------------------ Couperin: # 60328 (BDUX53010): WF: support fixed and separated dumps Date: Mon, 27 Jan 2014 17:37:21 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ Add support for fixed and separated pre/post dumps in combination with Workflow _Affected Executables_ bdbpre bdbpost _Motive source_ ScrumStory:LNTECH-1485 ------------------------------------------------------------------------ Couperin: # 60326 (BDUX53291): AIX: remove nullptr check from -qcheck Date: Mon, 27 Jan 2014 17:28:28 +0100 (CET) Created on: Couperin Type: porting specific change _Problem Description (Customer terms)_ No core dumps are produced when calling ABORT_PROGRAM This is needed to analyze bshell problems _Affected Executables_ all ------------------------------------------------------------------------ Couperin: # 60233 (BDUX52995): Bdbvalidate: -H option disables high_ascii_tolerance_strict_mode Date: Fri, 17 Jan 2014 15:44:39 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ In high_ascii_tolerance_strict_mode, bdbvalidate -H may give unexpected errors. _Test Procedure_ Set up a Unicode environment with high_ascii_tolerance_strict_mode. Insert high-ASCII characters into a single byte column of a table that are valid in the installation locale. Next, run bdbvalidate -H on this table and observe that all high-ASCII characters are reported (and that no unexpected errors are given). _Affected Executables_ bdbvalidate ------------------------------------------------------------------------ Couperin: # 60222 (BDNT53158): D399425: Unicode: Use proper replacement for meta_max character (follow-up) Date: Fri, 17 Jan 2014 13:15:42 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ In Unicode on Microsoft SQL Server Easy Filtering gives unexpected results when searching for strings that match an initial portion of the string. For example, when searching for any string that starts with "L", the string "LIEFERABRUF" may not be found. Note: an attempt to fix this problem was included in the 8.9a.02 porting set (project number BDUX51974). _Test Procedure_ Observe that proper results are returned by the Easy Filtering functionality. _Affected Executables_ msql_srv _Motive source_ D399425 ------------------------------------------------------------------------ Couperin: # 60149 (BDUX53047): defect 429479: field can not be left alligned on report. Date: Fri, 10 Jan 2014 16:11:46 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ Left Alignment not working anymore after installing LN portingset on BaanIV on report compilation. _Workaround_ Compile with a 6.1c portingset _Test Procedure_ Check a report with a left aligned field and compile with LN portingset in BaanIV mode. _Affected Executables_ repgen _Motive source_ InforXtreme:429479 ------------------------------------------------------------------------ Couperin: # 60142 (BDUX53027): SQL: SELECT statement ending on = SUM(...) may create corrupt RDBMS SQL text Date: Fri, 10 Jan 2014 14:38:23 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ If a SQL SELECT statement ends on "= SUM(..)" (or any other aggregate function), then a corrupt SQL statement text may be generated by the database driver. An example of such a statement is the following: SELECT edlevel FROM dbtst120 GROUP BY edlevel HAVING 46500 = SUM(salary) _Workaround_ Reformulate the expression as "SUM(..) = ", or add a dummy expression such as "and 1=1". _Test Procedure_ See Problem Description (Customer terms). _Affected Executables_ All database drivers ------------------------------------------------------------------------ Couperin: # 59968 (BDUX52801): D432095: SQL: Correlated subquery with UNION gives incorrect result Date: Mon, 16 Dec 2013 10:45:22 +0100 (CET) Created on: Couperin Type: bugfix _Problem Description (Customer terms)_ A SQL SELECT statement with a correlated subquery having a UNION, fails if the branches of the UNION have outer column references. An example of such a statement is the following: SELECT count(*) FROM cisli240 a WHERE {a.slcp,a.ortp,a.koor,a.slso,a.oset} >= {0,0,0,'',0} AND EXISTS ( SELECT 1 FROM cisli245 WHERE cisli245.slcp = a.slcp and cisli245.ortp = a.ortp and cisli245.koor = a.koor and cisli245.slso = a.slso and cisli245.oset = a.oset and (cisli245.stat = cisli.stat.on.hold) UNION ALL SELECT 1 FROM cisli245 WHERE cisli245.slcp = a.slcp and cisli245.ortp = a.ortp and cisli245.koor = a.koor and cisli245.slso = a.slso and cisli245.oset = a.oset and (cisli245.stat = cisli.stat.confirmed) ) Note that the subquery has a UNION (ALL), and that both branches contains outer column references (a.slcp, a.ortp, a.koor, a.slso and a.oset). _Test Procedure_ Write a query such as in the Problem Description (Customer terms), and observe that the correct result is returned. A regression test has been added. _Affected Executables_ All database drivers/dlls _Motive source_ D432095 ------------------------------------------------------------------------ Couperin: # 59812 (BDNT52457): Validating Windows 2012 cluster Date: Thu, 28 Nov 2013 14:17:42 +0100 (CET) Created on: Couperin Type: porting specific change _Problem Description (Customer terms)_ Installing multi-BSE on cluster may not completely work as expected. _Workaround_ NA _Test Procedure_ Tested with 2-node cluster on Windows Server 2012 with 2 BSEs. _Affected Executables_ bmscsset.exe bmscscfg.dll envres.dll _Motive source_ ScrumStory:LNTECH-1095