HP 3000 Manuals HP 3000 Manuals
FILE-CONTROL Paragraph The FILE-CONTROL paragraph is used to name files to be used in your program, and to define certain properties of these files that are necessary for their use by your program. Each file named in the FILE-CONTROL paragraph must be described in the DATA DIVISION of your program. Conversely, each file described in the DATA DIVISION must be named once in the FILE-CONTROL paragraph. An overview of the types of files that can be used in HP COBOL II is presented on the following pages. Following this overview, the various clauses of the FILE-CONTROL paragraph are described. In HP COBOL II, there are five ways to access and use files: * Sequential access * Random access * Relative access * Indexed access * Sort-merge access Each type of file described must be named, and certain features specified, in the INPUT-OUTPUT SECTION. The organization of the files and their logical records must be described in an SD entry (for sort-merge files), or an FD entry (for any other type of files) in the FILE SECTION of the DATA DIVISION. Sequential Files Sequential files are generally files residing on, or being written to, a serial access device (such as a magnetic tape or a serial disk). Of course it is possible to access disk files sequentially also. A sequentially accessed file means the records of that file can only be accessed in the order in which the records were written to the file. Because of the nature of serial access devices, these types of files can only be written to or read from in a single operation. However, on direct access discs, files being accessed sequentially can be read from and written to at the same time, as well as have a record brought in, modified, and returned to the same storage area. Random Access Files Random access files must reside on disk. Through the use of a key, you can read or write a record anywhere within a random access file, regardless of whether data has been written on previous records. The only limitation on where you can write a record is the externally defined boundaries of the file. For example, if a random access file has been defined to contain a maximum of three thousand records, you cannot write data to record number 3010, although you can write data to record number 2000 without having written data to any preceding records. Record access of random access files is controlled by the data item defined in the ACTUAL KEY clause of the FILE-CONTROL paragraph. This data item is described anywhere other than in records associated with this file. It is most efficient if defined as a signed integer of five to nine digits whose usage should be described as COMPUTATIONAL SYNCHRONIZED. However, any numeric data item of a sufficient size for the records in a file can be used. The ACTUAL KEY data item is used by placing a number into it that corresponds to the logical record number in the file. Logical records in a random access file begin with record zero. Thus, to access the tenth record in a random access file, your program must move the integer 9 into the ACTUAL KEY data item, and then execute the input-output statement. Execution of an input-output statement for random access files does not update the ACTUAL KEY data item. For example, if the ACTUAL KEY data item contains a value of one before reading or writing takes place, the second record is accessed when the READ or WRITE statement is executed. Following execution of the statement, any subsequent WRITE statement or READ statement (without the NEXT phrase) also accesses record one, unless the value in the ACTUAL KEY data item has been changed. When your program writes data to a record of a random access file, and previous record areas have had no data written to them, these records are filled with blanks or zeroes. Blank fill is used when the records of the file are designated as ASCII records. Zero fill is used when the records are designated as binary records. The implication of this blank/zero filling is that you can read records of the file for which no WRITE statement has been executed. In such a case, the data moved into your program is either a blank record, or a zero-filled record. This capability does not exist for any other type of file. Relative Files Relative files are similar to random access files in that you access records of such a file through the use of a record number. The only real difference in the two keys used for random access and relative files is that record numbers on a relative file begin with one, rather than with zero as in random access files. The major functional difference between random access and relative files is that you can always reuse record areas in a random access file by simply writing new data in them. In relative files, you must use the DELETE statement to purge data from record areas. Once a record has been deleted, you can no longer access the area it occupied except to write a record into it again. Relative files opened in dynamic mode use a data item named in the RELATIVE KEY clause of the FILE-CONTROL paragraph to access records. This data item is later described in the WORKING-STORAGE SECTION. Although the ANSI standard allows only an unsigned integer value for this data item, it is most efficient if defined as a signed integer of five to nine digits whose usage should be described as COMPUTATIONAL SYNCHRONIZED. However, any numeric data item of a sufficient size for the records in a file can be used. Relative files opened in sequential mode do not need to use the RELATIVE KEY data item to access records. You can simply execute input-output operations on them as though they were sequential files. If, however, you wish to position the file by using the START statement, you must specify the RELATIVE KEY data item, since the START statement uses this data item to find the record you want. Your program can access a relative file in one of the following ways. Sequential Access. Sequential WRITE statements for a relative file release data to the file, starting with the first record on the file, and proceeding to the second, third, and so forth in turn. If the RELATIVE KEY data item has been specified, it is updated each time a sequential WRITE statement is executed.
NOTE Even though you may already have data on these records, a sequential WRITE statement will cause the new data to replace it.
Sequential READ statements for relative files start with a particular record, read it, and proceed to the next existing record. The record that is read depends upon the type of the last input-output statement executed before the READ statement is encountered. If a DELETE or READ statement is executed, the READ statement reads the next existing record following the record just read or deleted. If an OPEN statement is executed, the first existing record is read. If a successful START statement is executed, the record pointed to is read. Sequential DELETE statements require the use of the READ statement to position the file to the record to be deleted. This READ statement must be the last input-output operation performed on the file before the DELETE statement is encountered. Sequential REWRITE statements require the use of the READ statement to position the file to the record to be rewritten. This READ statement must be the last input-output statement performed on the file before the REWRITE statement is encountered. Random Access. Random access input-output statements use the required RELATIVE KEY data item to select the record for the READ, WRITE, REWRITE, and DELETE statements. Thus, to perform a random access input-output operation on a relative file, you must place the number of the record to be accessed into the RELATIVE KEY data item before executing the input-output statement. When the random access input-output operation is executed, an implicit seek is performed to find the record, and the specified input-output operation is performed if possible. Dynamic Access. A relative file open in dynamic access mode allows you to access your file in either random or sequential mode. The only permissible sequential access that can be performed on a relative file opened in dynamic access mode is the READ NEXT form of the READ statement. This statement allows you to access the records of the file, starting with the record pointed to (if valid) by the current record pointer. If the record is invalid (that is, has been deleted), the next valid record in the file is read. Indexed Files Indexed files use data items that are integral parts of the records to control accessing of the records. For a given indexed file, each record contains a single prime record key, and zero or more alternate record keys. Each key must be described as alphanumeric within the record description entry of the associated indexed file. To indicate which key is the prime record key and which, if any, are the alternate keys, you must specify their names in the RECORD KEY and ALTERNATE RECORD KEY clauses, respectively, in the FILE-CONTROL paragraph of the ENVIRONMENT DIVISION. The prime record key is used in writing, deleting, or updating records of an indexed file. Alternate record keys are used only in reading records. The values of both prime and alternate record keys may be duplicated within an indexed file. Note however, that since the prime record key must be used for all input-output operations except reading, you should be very careful to make sure that if you are deleting a record whose prime key has duplicates, it is the record you wish to delete, and not some other record with the same prime record key value. When a prime record key has duplicates, and you use such a key to access a record on an indexed file, the file is searched in a chronological order. That is, the file is searched according to which record was written first. Thus, the record accessed is the first record containing the specified record key value that is still active. As with relative files, an indexed file can be accessed in random, dynamic, or sequential mode. The actions taken for a specific access mode and a specific input-output operation are listed below. Sequential Access. A sequential access READ statement for an indexed file uses the current record pointer to read records from the file. The record selected to be read is determined in essentially the same way as for relative files opened in sequential mode. See the description of sequential access of relative files, above. A sequential access WRITE statement for an indexed file uses the prime record key to place records in the file. Records must be written in ascending order according to these keys. Since the prime record key (and all alternate keys as well) must be alphanumeric, this means that the records must be written using the ASCII collating sequence to determine ascending order. A sequential access REWRITE or DELETE statement requires that a READ statement be the last executed input-output statement for the file being referenced. Random Access. The READ ... KEY IS form requires that you place a key value into one of the record key data items (prime or alternate). This data item is then specified in the READ statement, and the indexed file is searched until a record having the same value in the same record key is found. This record is then brought into your program. A random access WRITE, REWRITE, or DELETE statement uses the contents of the RECORD KEY data item to select the record to be written or deleted. Dynamic Access. When dynamic access is used for indexed files, you can use either the READ NEXT or the READ...KEY IS form of the READ statement. The READ NEXT form is used when you wish to read records in a sequential manner. This statement allows you to access the records of the file starting with the record pointed to by the current record pointer (if valid). If the record is invalid, the next valid record in the file is read. A dynamic access WRITE, REWRITE, or DELETE statement uses the contents of the RECORD KEY data item to select the record to be written or deleted. As an extension to ANSI COBOL'85, HP COBOL II allows the use[REV BEG] of alphanumeric, computational, numeric[REV END] display (without the optional SIGN clause) and COMPUTATIONAL-3 data types for RECORD KEY and ALTERNATE RECORD KEY clauses respectively, in the FILE-CONTROL paragraph of the ENVIRONMENT DIVISION. Indexed files are processed with Hewlett-Packard's KSAM subsystem in HP COBOL II. Refer to the KSAM/XL Reference Manual for information such as creation and deletion of this file type. Programmatic creation and deletion of these files can be accomplished by calling the appropriate intrinsics described in the above manual using the CALL statement. Opening, closing, and reading of these files is accomplished through the normal COBOL OPEN, CLOSE, and READ statements. Sort-Merge Files The ability to arrange records in a particular order is often required in COBOL applications. This ability is provided by the sort and merge features of HP COBOL II. The sort facility allows you to arrange the records of one or more files in a specified sequence. The merge facility[REV BEG] merges two or more previously sorted files.[REV END] Sort-merge files are the files acted upon by the sort and merge operations. These files can never be accessed directly, except in input and output routines associated with a SORT or MERGE statement. When a SORT or MERGE statement is issued, the records are taken from the named input file or procedure and then placed in the sort/merge workfile. Finally, the sorted/merged records are[REV BEG] placed in the output file, files, or procedure.[REV END] Refer to Chapter 12 , "SORT-MERGE Operations", for more information on sort-merge files and operations. File Status Every type of file described above[REV BEG] (except sort-merge files)[REV END] may have a file status data item associated with it. This data item can be used to check on the success or failure of an input-output operation involving the file with which it is associated. File status data items are described in more detail later in this chapter. Fixed Length Records Fixed length records must contain the same number of character positions for all the records in the file. Only one record size can be processed by all input-output operations on the file. Fixed length records may be explicitly selected by specifying format 1 of the RECORD clause in the file description entry for the file, regardless of the individual record descriptions. Refer to Chapter 7 for more information on the RECORD clause. Variable Length Records Variable length records can contain differing numbers of character positions among the records on the file. To define variable length records explicitly, specify the VARYING phrase in the RECORD clause of the file description entry or the sort-merge file description entry for the file. The length of a record is either affected by: the data item referenced in the DEPENDING phrase of the RECORD clause, the DEPENDING phrase of an OCCURS clause, or the length of the record description entry for the file.