HP 3000 Manuals HP 3000 Manuals
INPUT (SORT/XL)
Within the SORT/XL subsystem, the >INPUT command specifies the input
file(s) to be sorted. Refer to the MERGE/XL >INPUT command for
information on how to use the command within that subsystem.
SYNTAX
{$STDIN [ X ] }
>I[NPUT] {* }[,#records][,rec size]
{fname }
{(fname1, fname2,...fnameN )}
PARAMETERS
$STDIN [ X ] Specifies that the input records are to be read from $STDIN[
X ]. In interactive mode a question mark (?) prompt is
displayed.
* Entering an asterisk (*) in an interactive session specifies
that the input records are read from the text file TEXT. In
this case, the input records are to follow the >END command,
and the list of records is to be terminated with :EOD.
Recall that TEXT is the formal file designator of the file
containing the SORT/XL commands and that it defaults to
$STDIN. Therefore, if no file equation has been entered
against the file TEXT, the input records are entered from
the terminal in interactive mode and are included in the
stream file in batch mode. (The prompt "?" is displayed for
each record in interactive mode.) If a file equation has
been issued, then the records should be part of the file
equated to TEXT (again following the >END command).
fname Specifies the actual file designator. $NULL is not a valid
input file.
#records The #records parameter should be specified only if one or
more of the input files is not on disc. It is a positive
integer specifying the upper limit of the number of records
sorted. If multiple input files are specified, it is the
total number of records from all input files. When all
input files are on disc, the current end-of-file (EOF)
values are used and #records is ignored. If all input files
are not on disc and #records is not specified, a default
value of 10,000 is assumed by SORT/XL. This parameter cannot
be used to extract a subset of the input file.
Table 6-0. (cont.)
rec size A positive integer specifying the maximum allowable number
of bytes in a record. This parameter may be used to set the
record size of the output file, but is used mainly for files
containing variable-length records. When sorting such
files, this parameter should be set to the size of the
largest record present in the input. If rec size is not
specified when sorting variable-length record files, SORT/XL
will use the block size as the maximum record size. This
could result in more space than necessary being used for the
scratch file, as well as causing some degradation of
performance.
MPE XL SORT has two sets of scratch files. If the sort takes place in
compatibility mode, there is one scratch file for which the size is
computed as in MPE V/E SORT. If the sort takes place in native mode,
there are two (mapped) scratch files. Both computations are described
below.
Compatibility Mode Scratch File Size
If you want to extimate the scratch file record size (SFRS) and the
scratch file size (SFS), use the following equations:
SFRS+((rec size + 7)/2) + 4
where rec size is the input record size in bytes. (You must add the
length of the keys to the rec size if the keys are of the type, BYTE, and
ALTSEQ is used.) SFRS is in words.
SFRS+((SFRS*#records )/128) + 1
SFS is in sectors.
Compatibility Mode Scratch Filename
You can issue a file equation for the scratch file only to specify a
particular logical device which must be a disc. For example:
FILE SORTSCR; DEV=2
Native Mode Scratch File Size
Native mode scratch files contain two types of records: Work Records and
End-of-Subfile Records. The following algorithm calculates the size of
one native mode scratch file.
Work Record Length (#Work Records + #End-of-Subfile Records)
= #Bytes in 1 scratch file
Where:
#Work Records = #Input Records
#End-of-Subfile Records = #Input Records / 50
and Work Record Length = Input Record Length + Expansion Bytes
The value of Expansion Bytes depends on the number and type of keys that
the user specifies. Expansion Bytes is expressed as:
#key 5's + #Key 9's + 2(#key 4's + #key 6's + 3key 7's +#key 8's +#key 12's) + 3(#key 13's) +3
The space used for a sort occuring in native mode will be two times the
value returned by this formula.
NOTE This formula illustrates a worst case situation.
Native Mode Scratch Filenames
You cab issue file equations for the native mode scratch files only to
specify a particular device which must be a disc. For example:
FILE HPSORTS1; DEV=2
FILE HPSORTS2; DEV=2
DISCUSSION
When specifying more than one input file to SORT/XL, the list of files
must be enclosed in parentheses. This differs from the use of the >INPUT
command for MERGE/XL, where parentheses cannot be used. If more than one
>INPUT command is entered, only the last command is effective. Thus, all
the files to be sorted must be specified in a single >INPUT command.
This command can be entered any time before the >END command. In the
absence of the >INPUT command, any disc file with the formal designator
>INPUT is considered the input file. Also, file equations may be issued
before entering or during either subsystem. Thus, if the >INPUT command
refers to the same file as specified in a file equation, the file's
characteristics are determined by the file equation. The user issues the
>RESET command before entering SORT/XL or MERGE/XL if the default values
for the parameters of the file are desired. The same holds for the
>OUTPUT command for SORT/XL and the >INPUT and >OUTPUT command for
MERGE/XL.
EXAMPLE
In the following example, the file EMPLOYEE is to be sorted with a
maximum of 30 characters from each record:
:SORT
HP32214A.01.00 SORT/3000 THU, JUN 4, 1987, 9:50 AM
(c) HEWLETT-PACKARD CO. 1986
>INPUT EMPLOYEE, 30
ADDITIONAL DISCUSSION
See the >INPUT command for MERGE/XL in this chapter.