公开(公告)号：CA1290455C

公开(公告)日：1991-10-08

申请号：CA589435

申请日：1989-01-27

Applicant: IBM

Inventor： CHANG PHILIP Y ,  ENGLES ROBERT W ,  OBERMANN DAVID F ,  TRUMBLE MARY K

IPC: G06F17/30 ,  G06F20060101 ,  G06F15/40

Abstract: AT9-88-010 In the relational database system of the present invention, an extension to the SQL language is provided to allow a user/application to specify whether a column in a database table contains only single byte character set data, or mixed single byte character set data and double byte character set data. Specifying whether a column contains single byte character set data or mixed data is achieved by specifying the subtype of the character data types including CHAR, VARCHAR, and LONG VARCHAR. The user application can specify the subtype of character data within a column as the column is created or added by specifying FOR SBCS or FOR MIXED in either the CREATE TABLE statement or the ALTER TABLE statement. Along with specifying the subtype as either for single byte data or for mixed data, the user/application can further specify the codepage to be used for the column of the table in the database. Internally within the database manager of this invention, the subtype is recorded in terms of the code page attributes of the character data type as stored in the database system catalogs. This encoding scheme provides an efficient method of determining whether a column is only for single byte character set data or mixed data, and at the same time knowing the correct codepage that is to be used for that column simply by querying the code page columns within the database system catalogs.

公开(公告)号：CA1304506C

公开(公告)日：1992-06-30

申请号：CA593250

申请日：1989-03-09

Applicant: IBM

Inventor： CHANG PHILIP Y ,  COYLE DANIEL J JR ,  HOWIE LAUREN D ,  LINDSAY BRUCE G

IPC: G06F17/30 ,  G06F15/40

Abstract: AT9-88-013 SYSTEM AND METHOD FOR EFFICIENT ANALYSIS OF RELATIONAL DATABASE OBJECTS A plurality of packed descriptions is provided as a column in a table each entry containing data defining attributes associated with a different object (such as a table or view defined to the database) and all its component objects (such as related columns and indexes). Accessing the row corresponding to a particular object returns a description of all of the attributes of the object's component objects, as well as information describing the object itself. The packed description information is redundantly stored conventionally in other system catalogs for use IN at IN querying physical definitions of objects in a manner consistent with other database products. The database state information comprising the packed descriptions is retained in internal format whereas the correlative redundant information is in a user-readable external format. By denormalizing the schema for schema data whereby identifying relationships between objects is simplified, time required for compilation of statements referencing objects is substantially, reduced. All information concerning attributes of an object specified in statement and its component objects necessary, to compile the statement may thereby be fetched from the packed description field corresponding to the object in one I/O operation. Objects descriptions must otherwise be retrieved from numerous I/Os to separate system catalogs and rows collectively containing the descriptions.

公开(公告)号：CA1290456C

公开(公告)日：1991-10-08

申请号：CA589118

申请日：1989-01-25

Applicant: IBM

Inventor： CHANG PHILIP Y ,  COYLE DANIEL J JR ,  MALKEMUS TIMOTHY R ,  RODRIGUEZ REBECCA A ,  WELTI PHILIP J

IPC: G06F12/00 ,  G06F7/24 ,  G06F17/30 ,  G06F15/40

Abstract: AT9-88-011 In a relational database system, a method is utilized that increases the performance of the sort operation. An optimizer routine in the relational database manager analyzes a user's complete query to determine whether the final sorted results can be used directly by the user as sorting occurs. If the sort results can be used, the sort results are sent to Relational Data Services in the relational database manager for output to the user. Depending upon the determination made by the optimizer routine, one of two output modes for the final sorted sequence string of data are selected by the relational database manager. In disk output mode, the last pass of the final sorted sequence string is written to disk. In fast direct output mode, the records of data are sent to the user as the final sorted sequence string is being merged during the last pass. The system and method of this invention reduces the total sort time by eliminating the overhead of writing to disk during the fast direct output mode. Also, the response time, or availability for each record is reduced since each record is retrieved or sent to the Relational Data Services immediately after it has been sorted into the final sort order instead of completing the entire sort first, and then writing to disk.

公开(公告)号：CA1165449A

公开(公告)日：1984-04-10

申请号：CA402853

申请日：1982-05-13

Applicant: IBM

Inventor： CHANG PHILIP Y ,  HOFFMAN VIRGINIA M

IPC: G06F7/24 ,  G06F17/22 ,  G06F17/24 ,  G06F17/30 ,  G06F7/08

Abstract: QUALIFYING AND SORTING FILE RECORD DATA of the Invention Record data on a disk file is sorted in a text/data processor by means of an algorithm that transfers such records on the basis of rank to a sort buffer on the basis of qualifying criteria. Each qualified file record is compared with the lowest ranked record previously transferred and located in the sort buffer. When a higher ranked record is identified it is transferred into the buffer at a location based on qualification. Lower ranked records are deleted from the sort buffer if space does not permit the storing of such records within the space available. When the sort buffer has been loaded with the highest ranked records remaining in the disk file without overflowing the buffer is unloaded to an output device. The sort program recycles through a subsequent pass again transferring the highest ranked remaining records into the sort buffer. To minimize recycle time, a presort algorithm is run to set record identifying bits in a bit map section of the sort buffer. Each time the contents of the sort buffer is output the record identifying bits for the records in the sort buffer are reset to a second state. The second state of a record identifying bit indicates that that record will not be considered for future passes through the sort routine. AT9-81-021 B-14927