Login  |  Register
E-mail:
Password:
Forgot your password?
E-Mail:

Authorization

DBMS Linter Modifications List of 13.04.10

13.04.2010

Platforms:

  • Maemo support added.
  • New access management functions added for installation support in Windows Vista.
  • New message exchange variant for QNX corrected.

Interfaces:

Perl:
  • Functions for working with several BLOB columns and some previously absent functions for working with BLOB added.
  • BLOBGetSize function modified: now it can be called in two ways: as before, or mentioning the column number.
  • Interface for Version 6 redesigned: function returning connection info as an array added; function for setting the encoding added; function using batch row reading command added; functions for getting row info as an array and as hash, as well as information about return rows, added; more column info returned as hash in GetColInfo; the .pm file modified (description of work with extension; some module export errors (constant names) corrected).
  • Broad response processing optimized.
  • Immediate first row retrieval added.
  • Optional behavior for BLOB implemented (when entering the CO_FETCH_BLOBS_AS_USUAL_DATA option, BLOB is retrieved as a whole, like regular data).
  • Asynchronous query support implemented.
  • Operations with parameters in DBI optimized (the old callback mechanism removed from intlib; bind corrected in host variables (now the data is taken directly at the moment of execute); поправлена работа с параметрами процедур working with procedure parameters optimized).
ADO.NET:
  • LINQ provider code enhanced.
  • Nhibernate support added.
  • Nhibernate compatibility problems eliminated (if the buffer is empty, GetBytes returns its length for BLOB; conversion of some data types corrected; getOrdinal now searches first case-sensitive, then case-insensitive).
  • MONO developer 2 support added.
  • .NET 4 RC support added.
CALL interface (intlib):
  • Transfer of the inter() function of an empty buffer for a long value corrected.
OleDB:
  • Getting the result column description in case of a long query corrected.
  • Retrieval of a long row (over 8K) by an ordinary select query corrected.
  • Problem of extra characters at the end of a query eliminated.
  • Buffer size for NULL indicators increased to avoid errors caused by its small size.
DBExpress:
  • Problem with double loading of the driver after uploading eliminated.
ODBC:
  • Buffer (translation, result, batch row reading) size increased to 64KB for new builds. These corrections affect the DBExpress interface as well.
JDBC:
  • getbytes function processing corrected (during a call with a 0 length argument, the BLOB size is ignored, as well as the fact that the 1 position does not exist in an empty BLOB).
LinQT:
  • Correct build using win32-g++ provided.

Utilities:

Lindeskx:
  • BOOLEAN data loading problem eliminated.
  • Visual representation of dialog boxes corrected. Some corrections made in SQLEditor;: CALL word added to the list of words starting a query; problem with non-execution of "Minimize/maximize everything in the query" (if the query starts in the 0 row) eliminated.
  • Query editor, stored procedure editor and stored procedure debugger significantly enhanced (parsing delay problem eliminated; missing key words added).
  • Possibility to insert a breakpoint on a procedure prior to its calling added; WAIT mode completely redesigned (in particular, modal window added).
  • Error processing corrected to prevent hang-up when disconnecting from the kernel; asynchronous query stopping corrected.
  • Displaying of local variables in recursive calls corrected.
  • Logging the same error prevented if the user has not performed any actions.
  • Comment parsing enhanced.
  • SYSTEM_TIME and USER_TIME graphs included in the statistics window.
Linadm:
  • Automatic start of the replication server during the automatic database startup corrected.
  • Replication server management corrected.
  • Default replication startup for the database corrected.
Loarel:
  • For new builds, additional quick data loading mode with a large buffer using the "-s 2" key added.
Dbstore:
  • Problems with saving the CHECK limitation text in a wrong encoding and with a double '%' character eliminated.
  • For new builds, additional batch data uploading mode with a large buffer using the "-o M" key added.
Lhb:
  • Checkpoint transfer processing corrected to eliminate the log file accumulation problem.
  • Temporary checkpoint transfer problem eliminated in the non-incremental mode using the lhb –movecp option.
  • In the scripting language, inequality operation for row variables corrected; message returning in case of external command execution problems corrected.
Inl:
  • Utility shutdown in case of Linux kernel shutdown eliminated.
hresctl (hot backup copying):
  • Command line syntax hint corrected (for –t key).
Asynchronous replication server (lreplserver):
  • Work with databases with owners other than SYSTEM corrected.
testdb:
  • Problem with testing tables with rows of over 4K containing BLOB values eliminated.
  • Time representation precision in TESTDB messages enhanced (new –showticks key added).

DBMS Kernel and SQL Translator:

  • 'TABLE ' construction now supported according to the SQL standard.
  • CREATE UNIQUE INDEX syntax support added.
  • LINTER_VERSION function returning information about the DBMS version added.
  • Number conversion added to display numbers in the exponential format by the TO_CHAR function.
  • USER_TIME and SYSTEM_TIME columns added to $$$SYSINFO virtual table to calculate user time (kernel operation time) and system time (input-output operation time).
  • In CASE constructions, constant expressions of numeric types (integer, real, or both) are automatically converted to CASE type.
  • Corrections made for uniform deletion of log files during hot backup.
  • Processing of IS [NOT] NULL predicate after OUTER JOIN optimized for anti-join of tables (find something in one operand which is absent from the other).
  • Some additional optimizations for calculating aggregative functions in trivial cases implemented.
  • Some additional optimizations for calculating the LIKE predicate in trivial cases implemented.

Problems Eliminated:

  • Logging and operation of the kernel in the hot backup copying mode:
    • in operations with BLOB values;
    • in parallel work of different cursors in one connection;
    • when working with stored procedures; o in trigger channel rollback;
    • when deleting compound keys;
    • in operations: PRESS, REBUILD, TRUNCATE etc.;
    • with Error 6708 in case of ROLLBACK in the main channel after COMMIT in the cursor (in operations with BLOB values);
    • in quantization of additional operations according to the log;
    • when working with AUTOINC columns;
    • looping during transaction rollback;
    • damage of table converter pages in the backup database during hot backup.
  • Query processing problems:
    • problems when using indexes to calculate predicates for sub-query columns (intermediate result conversion algorithm changed);
    • calculating aggregative functions from bulk row fetching (UNION etc.) ;
    • processing CONTAINS full text search predicate in advanced queries;
    • wrong choice of processing strategy for the LIKE predicate in case it contains an expression in its right part;
    • incorrect conversion of BETWEEN/NOT BETWEEN predicate in case aggregative indexes are used;
    • calculating a query with an advanced sub-query in SELECT containing an external link;
    • channel memory leak during execution of a certain category of advanced queries containing the OR predicate;
    • quantization of advanced query processing.
  • Incorrect operations with records longer than 4K and long queries:
    • execution of a hierarchic query with a result record size over 4K;
    • execution of a query selecting a record over 4K in length and containing a BLOB value (in case this BLOB value is used further);
    • creating an aggregative index in the BY APPEND mode in case RECORD SIZE LIMIT is over 4K;
    • processing large-size translated queries.
  • Problem with possible omittance of a page to be transferred to the disk in the DBMS kernel due to a predefined error code.
  • Setting the BLOB value type when adding it via SQL.
  • Required checks added to building a phrasal index for a BLOB value.
  • Work of global time tables corrected (support of BLOB tables added; creation of phrasal indexes forbidden; operations with compound PRIMARY KEY corrected).
  • Problem that sometimes occurred when transferring NULL values via parameters eliminated.
  • A rare problem with returning Error 816 in an index search in case of using the index for a long time corrected.
  • Incorrect truncation of end spaces for constant CHAR arguments eliminated.
  • Concurrent debugging of the same stored procedure by several users forbidden.
  • Problems with parallel operation of several BACKUP processes eliminated.
  • Additional information added to diagnostic messages (search for a key in an index failed; unknown device detected).
  • LINTER_TMP environment variable now supported by loltp.
  • Description of BIGINT parameters in views for ODBC support corrected.
Main | Back to the list
ѕрокрутить вверх