The PC-SIG Library 9/The PC-SIG Library on CD ROM - Ninth Edition.iso/301_400/DISK0339/DISK0339.ZIP/CSORT.DOC

Text File | 1983-07-25 | 15KB | 276 lines

                   CHEAPSORT (level I AND II) DOCUMENTATION

                          CAPABILITIES OF CHEAPSORT
     CHEAPSORT  is a disk-based sort/merge program.  It will sort  data  files
which  have fixed-length records and fixed-length fields.  The data file being
sorted may be of any size,  but may not contain more than 32,767 records. This
should be sufficient for most uses.
     CHEAPSORT  will  not sort sequential data files,  such as  those  written
using a word processor;  nor will it sort data files written under the UCSD or
CP/M-86 operating systems, unless the data is somehow transferred to PC DOS.
     CHEAPSORT appears to work correctly under PC DOS 1.1 and 2.0, and on hard
disk,  floppy disk,  and RAM (or memory) disk. CHEAPSORT is not copy protected
in  any  way,  and  may  be backed up or transferred freely  to  other  media.
CHEAPSORT  is copyrighted and trademarked,  and should not be sold  except  as
noted in this documentation.
     You may sort data of any type commonly used in BASIC. Valid types include
ASCII data (numbers or letters,  in alphabetic order),  byte,  packed integer,
packed single precision,  packed double precision,  and any user-defined  data
type  adhering  to Microsoft's floating point format.  Many other  sorts  have
difficulty with other than ASCII data types.
     CHEAPSORT is compatible with THE CREATOR,  REPORTOR,  PROGEN, REPGEN, and
other software sold or distributed by T.N.T. SOFTWARE, INC.
     CHEAPSORT is totally compatible with Microsoft BASIC random-access files.
Other  sorts  may have some difficulties with  Microsoft  random  files,  even
appending unwanted control characters to the file output. CHEAPSORT will not.
     You  may sort unpacked numbers in numeric (rather than alphabetic)  order
by  choosing to sort that field as NUMERIC,  rather than CHARACTER.  CHEAPSORT
will   convert  any  such  numbers  into  a  packed  double-precision   format
internally; the sort will be accurate to 16 significant digits.
     CHEAPSORT  will  only output an INDEX to the file in sorted  order.  This
index will consist of a list of record numbers.  By reading the sorted file in
the order specified by the index,  you will be reading the file in the  sorted
order you specified. If you need a whole file output, you may use the index to
create  one.  See  the end of these notes for a sample program which  you  may
modify for your use.  The sample program,  incidentally, also shows how to use
the index to read a data file.
     For  your added convenience,  the sort index file can be read as a random
file  of  record length 8.  This will allow you to use the index  file  to  do
binary searches of the data,  if you wish. To help you further, CHEAPSORT will
always  tell  you exactly how many records were sorted when the sort has  been
completed. A null record is always appended to the end of the file to help you
locate  the end,  if you want to do binary searches.  Each record in the  data
file  to  be sorted can not contain more than 256 characters;  if you  have  a
requirement which exceeds that,  consider buying CHEAPSORT, level II. The cost
is $25.00,  postpaid,  from TNT SOFTWARE INC.,  34069 Hainesville Road,  Round
Lake  IL 60073.  Level II CHEAPSORT will sort input files with lengths  up  to
32767 bytes and combined key lengths of up to 32765 bytes (limited by memory).
     If  you  have  more  complex sort  requirements,  or  need  to  customize
CHEAPSORT (any level), the source code is separately available. Level I source
code costs an additional $11.00;  level II,  an additional $25.00. Examples of
customizing  CHEAPSORT to do select/exclude on fields or  field  relationships
are included with the source code.











                                     1


                             HOW TO RUN CHEAPSORT
     Simple:  while at the DOS prompt (A>,  B>,  etc.) just type in CSORT.  If
CHEAPSORT  is not on your current default drive,  preface CSORT with the drive
letter and a colon. For example:
A>B:CSORT
B>A:CSORT
     CHEAPSORT will ask you for the name of the file to be sorted,  the record
length,  the name of the output index file,  and the beginning position, type,
and length of each of the sort keys.
     The  beginning  position of each key must  be  carefully  specified.  The
easiest  way to understand how this should be done is by means of an  example.
Suppose you want to sort a data file whose record length is 85;  the fields in
each record are of lengths 15,20,5,20,10,10,  and 5.  Suppose you want to sort
each record by field 4, then 3, then 6, then 1.
     In  that case,  the starting position of the first key (field 4) would be
15+20+5+20=60,  plus  1 (skip the first 60 characters,  start the sort at  the
61st character). The key length would be 20, if you want to sort by all of the
fourth field.  Likewise,  the second key begins at 15+20+1=36,  the third  key
begins at 15+20+5+20+10+1=71, and the last key begins at position 1.
     You need not use all of a key field as a sort key.  If your key field  is
very  long (a name field,  with 80 characters allowed,  for example),  you may
wish to use only the first few characters of the field as the key.  Just enter
whatever length you decide to use.
     CHEAPSORT will remind you of the permissible data types;  please read the
possibilities  carefully.  Many of the possible data types (the packed numeric
ones, especially) are called "Other".
     You  may use up to 50 keys,  but the total length of all of the keys  may
not exceed 253 bytes. CHEAPSORT will inform you of the remaining key space and
the number of remaining keys after each key is entered.
     When you are done entering the keys,  you may go on to the next step  (as
CHEAPSORT  will inform you) by entering the number 0 for the starting position
for the next key.
     CHEAPSORT will then ask for the drive letter of the drive to use for work
files.  Many  times,  a  work file will not be  necessary;  if  one  is,  this
temporary file will be placed on the drive you specify.
     You  may  use any drive for your work files.  If you're sorting a  small-
enough data file (under 1000 records) and you have a small enough  total   key
length (depends on memory available), you may not actually need any work file.
     In  any event,  the work file space necessary will not exceed your  total
key lengths plus two bytes,  times the number of records in the data file. For
example,  if  your  total of key lengths is 50 bytes and you are sorting  2000
records,  the space required for your work file will be 52*2000=104000  bytes.
Be  sure  you have enough space available on the drive you designate for  your
work files. It is disheartening to run the sort almost to completion and get a
'disk full' error message.
     It  is  interesting to note,  in this  connection,  that  CHEAPSORT  uses
generally  less work file space than other sorts.  You will probably find that
CHEAPSORT will be capable of sorting data files other sorts cannot touch.
     CHEAPSORT is quite quick,  but is probably not the fastest possible  sort
for the IBM PC. It is, however, probably the least expensive good sort package
you will ever buy.  Here are some comparative sort times for the IBM PC, where
the  sort file and work files were on separate (floppy) disks.  The file being
sorted consisted of 3000 records, each of 25 bytes in length. There were three
fields:  a  15-character alphabetic name,  an 8-byte  packed  double-precision
amount, and a two-byte packed integer field. Print level 4 was used. The times
quoted  are  from  the last operator entry to CHEAPSORT until the  DOS  prompt
reappeared.  All data was generated by using BASICs RND function, so no fields
were in any particular sorted order.
     The  times quoted are from the time of the last operator input until  the
system prompt returns, and include all disk read and write times.



                                     2


               SORT TIMES, 3000 RANDOM RECORDS (25 BYTES EACH)
                   ALL TIMES QUOTED ARE MINUTES AND SECONDS
ASCENDING      DESCENDING     KEY(S)
2:47           2:51           Packed integer.
3:10           3:56           15 character alphabetic.
3:20           3:45           Packed double precision.
3:26                          15 character alphabetic + packed integer.
3:37                          Packed double precision + packed integer.
4:04                          15 character alphabetic + packed double prec.
4:22           5:39           15 char. alpha + packed double + packed int.

     Each  run  required  a merge of 3000 records;  this merge  took  about  a
minute.  If you were to sort fewer than a thousand records, a merge may not be
required; generally, sort times for such small data files will be well under a
minute.  The merge function has been highly optimized, and will take advantage
of  memory buffering as much as possible.  Disk reads of the merge  work  file
have been reduced to the minimum.
     Sorts in descending order generally take about 2-3% longer than ascending
sorts for each byte in the descending key field. Merge times are not affected.
Sorts  of packed single or double precision data take about 10% longer than  a
sort  on  an  alphabetic key of the same size.  Again,  merge  times  are  not
affected. Merges will take somewhat longer if the key lengths are longer.
     CHEAPSORT  will  run  more slowly if the disk data is  in  reverse  order
compared  to  the sort order requested.  I have done no checking on times  for
this,  but it should be possible to double the sort times if the data were  in
exactly reverse order.  On the other hand, if the data were in close to sorted
order  to  begin with,  CHEAPSORT could run appreciably faster than  indicated
here. Sort times in that instance would probably be primarily limited by disk
transfer times.
     On  some machines,  it is possible to appreciably speed up  CHEAPSORT  by
specifying print level 1.  On the IBM PC, it makes a difference of only two or
three  seconds on the above sort.  Therefore,  there is no real penalty if you
like to watch how the sort is progressing.
     You  may find it hard to believe (especially considering the  speed  with
which  CHEAPSORT  operates),  but CHEAPSORT was written entirely in  Microsoft
BASIC.   This  was  done  quite  deliberately,  so  that  CHEAPSORT  would  be
transportable to as many new machines as possible,  and would be adaptable  to
any  new operating system releases from either Microsoft or IBM.  If there  is
sufficient demand,  I may produce a version of CHEAPSORT for CP/M-86. However,
I consider this rather unlikely; most CP/M-86 software tends to get written in
and for the Digital Research languages, for which I have no love whatsoever.
     System  builders and developers might want to incorporate CHEAPSORT  into
their software; or, a user may obtain a copy of either version of CHEAPSORT in
order to 'test' its capabilities.  In any case,  if you wish to sell CHEAPSORT
or  'buy' a copy in order to make your sale or possession legal,  you need  do
the following:
     (1) Remit $5 for CHEAPSORT level I, or $10 for CHEAPSORT level II;
     (2)  Send the name and full address of the ultimate user of the  software
to T.N.T. SOFTWARE, INC. at 34069 Hainesville Road, Round Lake IL 60073.
     Others  using  or  selling CHEAPSORT do so  without  any  possibility  of
support and with the possibility of legal action against them for such illegal
use or sale.
     T.N.T.  SOFTWARE  has a very easy distribution policy,  and very low cost
software. It is our belief that software prices are entirely too high; but, if
our software is pirated and we obtain little gain from our policies,  we  will
have  only  two choices:  increase prices or leave the  market.  Neither  will
result  in any long-term advantage to the software buyer.  Please help us keep
prices low, and obey our requests for sale and distribution of our software.





                                     3


               SAMPLE PROGRAM TO READ A DATA FILE VIA AN INDEX
                        AND CREATE A WHOLE FILE OUTPUT

10 LINE INPUT"FILE TO READ:";F$
20 LINE INPUT"FILE TO WRITE:";W$
30 LINE INPUT"INDEX FILE NAME:";I$
40 INPUT"RECORD LENGTH:";R:'IF RECORD LENGTH >255 YOU WILL NEED TO ALTER
50 OPEN"R",1,F$,R:'      LINES 60 AND 80 FOR A MORE COMPLICATED
60 FIELD #1,R AS A$:'    FIELD STATEMENT. USE AS MANY VARIABLES AS NEEDED
70 OPEN"R",2,W$,R:'      TO USE UP ALL THE CHARACTERS IN EACH RECORD.
80 FIELD #2,R AS B$:'    FOR EXAMPLE:
90 OPEN"I",3,I$:'        FIELD #1, 255 AS A1$,255 AS A2$,255 AS A3$
91 '                     WILL HANDLE A RECORD OF 765 BYTES. YOU WILL
92 '                     NEED TO FIELD #2 SIMILARLY, AND DO THE LSETS IN LINE
93 '                     110 AS WELL. THE LSETS CAN BE DONE LIKE THIS:
94 '                     LSET B1$=A1$:LSET B2$=A2$:LSET B3$=A3$
95 '                     FURTHER NOTE: IF YOUR RECORD LENGTH EXCEEDS 128, YOU
96 '                     SHOULD BE SURE TO ENTER BASIC WITH THE /S: SWITCH
97 '                     SET APPROPRIATELY. SEE YOUR BASIC MANUAL.
100 PRINT"TRANSFERRING":ON ERROR GOTO 500
110 INPUT #3,A:GET 1,A:LSET B$=A$:COUNT=COUNT+1:PUT 2,COUNT
120 GOTO 110
500 PRINT"DONE.";COUNT;"RECORDS TRANSFERRED."
510 CLOSE:END


THANKS FOR PURCHASING CHEAPSORT!
FOR A CATALOG OF OUR OTHER INEXPENSIVE,  HIGH-QUALITY SOFTWARE FOR THE IBM PC,
CP/M, AND RADIO SHACK MACHINES, WRITE TO:

                            T.N.T. SOFTWARE, INC.
                            34069 HAINESVILLE ROAD
                             ROUND LAKE, IL 60073
                                [phone]


IBM PC is a registered trademark of The IBM Corporation;  CP/M and CP/M-86 are
trademarks  of  Digital  Research;   Radio  Shack  is  a  trademark  of  Tandy
Corporation;   THE  CREATOR,  REPORTOR,  PROGEN,  REPGEN,  and  CHEAPSORT  are
trademarks of T.N.T. SOFTWARE, INC.























                                     4



        This disk copy was originally provided by "The Public Library",
        the software library of the Houston Area League of PC Users.

        Programs are available from the Public Library at $2 per disk
        on user-provided disks.  To get a listing of the disks in the
        Public Library, send a self-addressed, stamped envelope to

             Nelson Ford,  P.O.Box 61565,  Houston, TX 77208.