Trailing-Edge - PDP-10 Archives - decuslib10-02 - 43,50231/change.doc
There are 3 other files named change.doc in the archive. Click here to see a list.





                                   David Richard Kiarsis

                                   Date:  June 30,1973

                                   Updated July 1982

                                   By Peter J.Plourd II
CHANGE -- Character set converter                      Page 0
           Command syntax


     CHANGE is  a  program  to  aid  in  the  conversion  of
character  sets  foreign to the DECsystem-10.  It is capable
of using any i/o device on the DECsystem-10, but  is  mainly
designed  for  users  with magnetic tapes and disks.  CHANGE
will   perform   blocking,   duplication,   character    set
conversion,  unblocking,  and  reading  and  writing of tape


     CHANGE may be run by typing the following:

          R PUB:CHANGE

CHANGE will respond with the date and time of day.  An angle
bracket indicates that CHANGE is ready for user input.


     CHANGE uses the traditional DECsystem-10 command string
format.  It will accept one input file specification and one
output file specification separated by a back arrow  (_)  or
equal sign (=), Example:


Each output  or  input  file  specification  consists  of  a
device, file name, extension, and project-programmer number.
The device is terminated by a colon (:), the file name by  a
period (.), and the project-programmer number is enclosed in
square brackets ([]).  Example:


The complete file specification is optional and will take on
default  values  if any part of it is omitted.  For example,
the device will default to generic disk  (dsk:),  the  input
file  name  to  INPUT,  the  output file name to OUTPUT, the
extensions to DAT, and the project-programmer numbers to the
current disk area.

CHANGE -- Character set converter                     Page 1
           Command syntax

     In addition to the file  specification  each  input  or
output  file  specification includes switches.  Switches are
descriptive in nature, and allow CHANGE  to  determine  more
about the files being handled.  A switch begins with a slash
and terminates on the next slash, if  the  switch  takes  no
argument,  or  on  a  colon if the switch takes an argument.
Note:  the return and altmode also  terminated  switches  as
well as the input line.  Example:


If a switch takes an argument the argument follows the colon
as  in  the second example above.  In all cases switches may
be abbreviated to the number of characters that  will  allow
that switch to be unique.

1.1.3     SWITCHES

     The following is a list of switches  that  CHANGE  will
accept.   In  the  following  descriptions  "x" represents a
decimal number, and "arg" represents a key word that may  be


     Designates that CHANGE should set up x buffers for  the
device.   This switch is used to speed up the i/o process at
the expense of core.  The size of a buffer is calculated  by
the following formula:



     Designates that  CHANGE  should  advance  x  EOF  marks
before  processing a file.  This switch has meaning only for
magnetic tapes and is ignored for all other  devices.   Note
that if a magnetic tape has labels, there may be several EOF
marks per file, depending on the label type.


     This switch is the opposite of the advance switch.   It
tells  CHANGE to backspace the magnetic tape x EOF marks and
then forward space one EOF mark if the tape is not  at  load
point.   This switch is only used with magnetic tapes and is
ignored for other devices.  CHANGE  resolves  BACKSPACE  and
CHANGE -- Character set converter                     Page 2
           Command syntax

ADVANCE  for  the  same  device by subtracting the BACKSPACE
counter from the ADVANCE counter;  if the result is positive
CHANGE  moves  the  tape  forward, if negative, CHANGE moves
that tape backwards.  /BLOCK:x    /BLKSIZE:x

     This switch tells  CHANGE  that  there  are  x  logical
records  in  a  logical block.  For variable ebcdic, this is
the maximum number of records  in  the  block.   Note  IBM's
blocking  factor  is  the  total number of characters in the


     This switch tells CHANGE that there are x characters in
a logical record.


     This switch is use to set the density on magnetic tapes
and  is  ignored  for other devices.  The argument is one of
the following:  200, 556, 800.


     This switch saves the current command string for  later
editing.   See the RETAIN command for a complete description
of the RETAIN function.


     For this switch CHANGE performs a  RETAIN  command  and
then  executes  the  current  command  string.   See the RUN
command for a complete description.


     CHANGE prints a list of its commands and switches  with
a  short  explanation of their function.  If the argument is
present, CHANGE will attempt to print the help file for that


     The LABEL switch informs CHANGE that  a  magnetic  tape
has  labels  or  that labels should be written.  In any case
the argument specifies the type of label as follows:

     NONE      The magnetic tape has no labels.
CHANGE -- Character set converter                     Page 3
           Command syntax

     MINE      The  magnetic  tape   has   special
               DECsystem-10 labels.
     DIGITAL   The magnetic tape has  DECsystem-10
               COBOL labels.
     BURROUGHS The  magnetic  tape  has   standard
               BURROUGHS labels.
     IBM       The  magnetic  tape  has  IBM   360
     GE635     The  magnetic  tape   has   GENERAL
               ELECTRIC 635 labels.

Labels are written in the character set specified except for
IBM  and  GE635  labels.   IBM labels are written in BCD for
7-track drives and EBCDIC for 9-track drives.  GE635  labels
are always written in GEBCD.



  ASCII   File character set is 7-bit ascii.
  HPASCII File character set is 8-bit ascii.
  GEASCII File character set is 9-bit ascii.
  IMAGE   File  is  read  and  written  as  36-bit
  SIXBIT  File  character  set  is   sixbit   with
          control words.
  FIXSIX  File character set  is  sixbit  with  no
          control words.
  BCL     File character set is bcl.
  BCD     File character set is bcd.
  GEBCD   File character set is  GENERAL  ELECTRIC
  HONBCD  File character set is HONEYWELL bcd.
  EBCDIC  File character set is ebcdic.
  VEBCDIC File character set is variable ebcdic.

The end of the input record is determined by the mode of the
input file as described below:

  ASCII     Terminated   by   a    carriage-return
            character or record count.
  SIXBIT    Determined by the header word in front
            each record.
  FIXSIX    Always  copies  the  number  of  bytes
            specified by the record size.
  BCL       Always  copies  the  number  of  bytes
            specified by the record size.
  BCD       Always  copies  the  number  of  bytes
            specified by the record size.
  EBCDIC    [fixed] Always copies  the  number  of
            bytes specified by the record size.
CHANGE -- Character set converter                     Page 4
           Command syntax

  EBCDIC    [variable] Determined  by  the  header
            word in front of each record.


     Sets the parity of a magnetic tape  file,  ignored  for
other  devices.   The  argument  is either ODD or EVEN.  The
default parity is ODD.


     Sets the password for GE635  labels  to  the  argument.
The  argument can be up to twelve characters.  This argument
is never checked it is only written in the output  label  as


     Sets the reel serial number in labels  that  have  this
feature.   Note  for  some  labels the reel serial number is
retained when writing  over  a  label.   For  instance,  for
BURROUGHS  labels  the  output  tape is read to retrieve the
serial number and then a new  label  is  written  with  this
number.   To  override this feature the REEL switch is given
and the output tape is not  read  first  (i.e.   the  output
label is purged).


     This switch initializes the magnetic tape for  industry
compatible  mode.  Usually set for EBCDIC and HPASCII modes.
In this mode 32-bit words are read or written.   The  32-bit
word is broken down into four eight bit bytes and written in
four frames on the tape.  The low-order 4-bits are not used.
Note  this  mode  is  only  used for 9-track drives and will
cause unexpected results on 7-track drives.


     On labeled tapes, this switch  forces  CHANGE  to  skip
down  the tape, in a forward direction, until the input file
name is found or an end of tape condition is detected.  This
switch is used only for labeled magnetic tapes.

CHANGE -- Character set converter                     Page 5
           Command syntax

     Specifies that CHANGE should abort the current  job  on
either an input or output parity error.


     Specifies that records are  allowed  to  cross  logical
block  boundaries.   The normal case is that a record always
will terminate on a logical block boundary even if a partial
word will be wasted.

/REWIND:arg *

     Specifies that the magnetic  tape  should  be  rewound.
This  switch  is only used for magnetic tapes and is ignored
for other devices.  The arguments are as follows:

   BEFORE Rewind before the operation.
   AFTER  Rewind after the operation.
   ALWAYS Rewind always [default].
   OMIT   Rewind neither before or after.


     Specifies that the magnetic  tape  should  be  unloaded
after the operation is completed.


     Specifies that  CHANGE,  during  a  wild  card  search,
should type out the file names as they are found [default].


     Tells  CHANGE  to  perform  a  list  of  the  devices's
directory.   This  switch  is  valid  for disk, DECtape, and
magnetic tapes.  Note this  switch  will  alter  a  retained
command and will transfer no data.


     Specifies that headers should be printed if the  output
device is a line printer [default].

CHANGE -- Character set converter                     Page 6
           Command syntax

     Specifies that carriage-return line-feed sequences  are
included in ASCII, HPASCII, and GEASCII files.

     Switches flagged with an asterisk (*) can be turned off
or  have the opposite effect noted, by concatenating NO with
the switch.  Example:




     The DATA command specifies  a  file  name  that  CHANGE
should  use  to  find  the  conversion tables that it needs.
Note this command is ignored  if  a  RETAIN  command  is  in
effect.  Example of use:


The data command has only  been  tested  with  disk  as  the


     The EXIT command informs CHANGE that the user wants  to
return to monitor level.  CHANGE upon receiving this command
will reset all I/O and return to the monitor.


     The BYE command is similar to the EXIT  command  except
that  CHANGE  will logout the user off the system instead of
returning the  user  to  monitor  level.   This  command  is
equivalent to typing the following:

          EXIT   typed to CHANGE
          KJOB/F typed to the monitor


CHANGE -- Character set converter                     Page 7
           Command syntax

     This command is used to save subsequent commands  typed
to  CHANGE.   In  this mode the commands are not immediately
executed.  Instead, CHANGE saves the command line and allows
the  user  to  edit it.  To edit the current command string,
once RETAIN has been typed, the user need  only  retype  the
part  of the command he wishes to change.  Items that follow
the back arrow or equal sign will be  edited  to  the  input
side of the command while things that precede the back arrow
or equal sign will be edited  to  the  output  side  of  the


     Once a command is retained the user may delete it  from
memory  by  typing  ERASE.   CHANGE  will  erase the current
command and return to its immediate mode of  interpretation;
that is CHANGE will now execute commands as they are typed.


     The  print  command  is  used  to  examine  a  retained
command.   As  the user is editing the current command it is
useful to print it out to see if the edits are correct.   To
do  so  the  user  types  PRINT  and CHANGE will display the
current command string


     To have CHANGE perform  a  retained  command  the  user
types  RUN.  CHANGE will then execute the current command as
if the user had typed it directly to CHANGE.   If  the  user
finds  that  the  command is incorrect he may stop CHANGE by
typing a control/c.  CHANGE will return to command level  to
accept  another  command.   Note a retained command is never
forgotten until the user types ERASE.


     The DIALOG command allows the user to carry on a dialog
with  CHANGE.   CHANGE will ask the user questions about the
input file specification and the output file  specification.
When  CHANGE  feels  that  it has enough information it will
execute the command it has compiled.  Note  the  command  is
not  retained in this mode and any command that was retained
is erased.  The DIALOG mode may also be  entered  by  typing
only an altmode to CHANGE.

3.0       TAPE FORMATS
CHANGE -- Character set converter                     Page 8
           Command syntax

     There are  great  number  of  different  tape  formats,
however,  there  are  some  rules to use when handling tapes
from certain vendors.   IBM  tapes,  that  are  recorded  on
9-track  drives,  usually  use  EBCDIC as the character set.
9-track IBM EBCDIC  tapes  always  use  industry  compatible
mode.   That  is, they always write 32-bit words on the tape
in four frames.  Another  tape  format  that  uses  industry
compatible  mode  is  HPASCII.  In addition, data written in
this mode usually does not have  return-line-feed  sequences
in  the text.  In other character sets, trial and error must
be used to determine whether the tape is written in industry
compatible mode.

     Although some computers have the  capability  to  write
words  shorter  than  36-bits (or 32-bits in industry mode),
the DECsystem-10 can only read and write 36-bit words.  This
creates  a problem for tapes written on the DECsystem-10 and
read on another vendors machine.  Note that the  reverse  is
not  true.   CHANGE  is written to handle short words if the
record size and block factor are accurate.  Note this can be
determined  by a trial and error procedure.  If records seem
to be cut short try a larger record size, or if records seem
to  be  completely  missed try a bigger blocking factor.  In
any case, tapes written with CHANGE on the DECsystem-10, and
read  on  another  computer may find extra characters in the
record.  There is no way around  this,  unless  the  program
that reads the tape on the other computer is smart enough to
handle this condition.

     Definition of terms:


     The smallest unit of data  that  can  be  processed  by
CHANGE.  In CHANGE this is also called a record.


     The smallest unit of data that can be processed by  the
hardware  (e.g.   128  words  for  disk,  80 columns for the
card-reader, the  data  between  record  gaps  for  magnetic


     An area of core memory into which the monitor reads, or
from which the monitor writes, a physical record.

CHANGE -- Character set converter                     Page 9
           Command syntax

     The number in the "/BLOCK:x" switch.  If  there  is  no
blocking switch the blocking factor is said to be zero.


     Those  buffers  required  to  contain   a   number   of
contiguous  records,  that number being the blocking factor.
A logical block may extend over  many  buffers,  but  always
uses  an  integral number of buffers;  any unused portion of
the last buffer is wasted.  If the smallest record of a file
is  much  smaller  than  the  largest record, there could be
several wasted buffers, since the number of buffers required
is  always  determined  by  the  size  of the largest record
multiplied by the number of logical records contained in the
logical block.


     An ordered collection of  contiguous  logical  records;
the largest unit of data that can be processed by CHANGE.

     A file is considered blocked if the blocking factor  is
non-zero;  it is considered unblocked if the blocking factor
is zero or if the SPAN switch is specified.



     An ascii record  is  a  set  of  contiguous  characters
terminated  by a return-line-feed sequence.  Word boundaries
have no significance, and the last character of  the  record
is  immediately  followed by the first character of the next
record.  The amount of buffer space required is  the  number
of  characters  in  the  record  plus two (cr-lf).  A record
terminates with the  first  EOL  character  or  a  satisfied
character  count.   If the record count was satisfied before
the EOL character was detected, the next record starts  with
the  next  character.  If the EOL character comes before the
record count is satisfied then the record terminates with no
space fill.  Ascii null characters are always discarded.


     These modes are a contiguous  set  of  characters  that
terminate only on a record count.  They are similar to ascii
in that they are independent of word boundaries (e.g.   when
a  record  ends  the  next  starts  in  the  next  character
CHANGE -- Character set converter                    Page 10
           Command syntax

position).  If a record ends before the  count  has  expired
the  record  is  space filled to the character count.  In no
case will the size of the record written be unequal  to  the
record  size.   The  buffer  space required is the number of
characters in the record.


     A sixbit record is a  set  of  contiguous  words.   The
first  word has, in the right half, the number of characters
in the record.  The last word may be padded to  ensure  that
the  record  boundary  coincides  with a word boundary.  The
amount of buffer space required is the number of  characters
in the record plus six characters for the character count in
the first word plus the number of padding characters.


     An ebcdic record is a  contiguous  set  of  characters.
The  first  word of a logical block contains a block control
word in the first two bytes, and  spaces  in  the  next  two
bytes.  This block control word contains the number of bytes
in the block.  The first word of each record is  the  record
control  word.   The record count is stored in the first two
bytes and  spaces  in  the  second  two.   The  record  then
follows.  Unlike sixbit, the ebcdic block and record control
words include the size of the  control  words.   The  buffer
space require is the number of characters in the record plus
four for the record control word,  and  plus  four  for  the
block control word.


     Only magnetic tapes have labels written  out  with  the
data.   A  mag-tape  file may have 2 or more labels.  If the
labeled file is a multi-reel-file, it has 2 labels for  each
reel.   A  labeled  file  contained entirely on one reel has
only two labels.  The  beginning  file  label  occupies  the
first  block  on  the  tape, and is followed by an EOF mark.
(if DIGITAL labels are used this tape mark is omitted.)  The
data  follows  this tape mark and is terminated with another
EOF mark.  The ending file label occupies the last block  of
the file and is followed by another EOF mark.


     CHANGE does not keep all of its  conversion  tables  in
core.   It  reads  a data file, after the user has specified
the  input  and  output  character  set,  to  retrieve   its
CHANGE -- Character set converter                    Page 11
           Command syntax

conversion  tables.   CHANGE  will try to open a file called
"SYS:CHANGE.DAT"   and   if   that   fails   it   will   try
"DSK:CHANGE.DAT".   If  both  of  these attempts fail CHANGE
will type out an error message an do no more.  The user  can
alternately  tell  CHANGE  the name of the data file to use.
To do this the user must first erase  any  retained  command
through the use of the "ERASE" command.  Then he may use the
"DATA" command to specify the file where CHANGE is  to  find
the  conversion  tables.   Note the format of the conversion
tables is defined in CHANGE at  assembly  time  and  is  not
taken from the data file.

     The conversion tables are really  two  tables  in  one.
That  is,  each  table has one table in the left half of the
word and one in the right half.  The left half table is  the
table  that  converts  to  ascii, while the right half table
converts to the output character set.  Note this means  that
CHANGE  internally  converts  all  character  sets to ascii.
Because CHANGE does this, character sets with more than  128
characters  (such  as  EBCDIC)  cannot be fully handled, and
some characters may be translated incorrectly.   The  manner
that  CHANGE  accesses  these  tables  is as follows.  First
CHANGE reads from the input file and extracts a byte of  the
proper size.  It then takes this byte and indexes by it into
the input conversion table.  From the left half of the input
conversion   table   CHANGE   extracts  an  ascii  character
equivalent to the input  character.   Then  with  the  ascii
character  CHANGE  indexes  into the output conversion table
and extracts a character from the right  half  table.   This
character  is  then the converted character and is passed to
the output file.  Note if a character can not be  translated
CHANGE  places  a  back  slash  (\)  in that position of the

     There is a program called  TABLE  which  generates  the
conversion  file  for  CHANGE.   TABLE  contains  all of the
conversion tables which  are  mapped  in  CHANGE  with  IOWD
pointers.   Thus  to  modify  a table, the user need only to
modify the program TABLE.  To do this, TABLE must be edited,
reassembled, and then run again.

6.0       EXAMPLES

     The following are examples of commands to CHANGE:

     A user has a magnetic-tape that was written on  an  IBM
360  system.   He  knows  that  the  tape  is a 9-track tape
written in IBM fixed ebcdic  and  wants  to  convert  it  to
sixbit  for  a COBOL program he is working on.  In addition,
the user knows that the input tape is blocked 1 and  has  80
character records.  He wants the output to be blocked 20 and
CHANGE -- Character set converter                    Page 12
           Command syntax

have 120 character records.  To solve this problem the  user
types the following command to CHANGE:


In the above example only user input was  shown.   He  first
retained  the  following  commands  and  then  typed  in his
command string.  Once having typed in the command string the
user had CHANGE type it back to him.  Having decided all was
o.k. the user had CHANGE do the command by typing  "RUN"  to
CHANGE.  The complete output may look as follows:


     CHANGE here at 12:21 02/02/73
     > RETAIN



     > PRINT


     > RUN

     CHANGE    12:25     02/02/73

     12.021 secs. 1234 i/o units.
     17777 octal locations used.
     400 rec. 400 blk. read.
     400 rec. 20 blk. written.


     Another user has a sixbit file that his  COBOL  program
just  generated.  He wants to convert it to ascii so that he
may edited it.  To do this he types the following command to

CHANGE -- Character set converter                    Page 13
           Command syntax

Note in this example the input record size did not  need  to
be specified because CHANGE will determine it from the data.
In general if the device has a fixed  buffer  size  and  the
character  set  has  control words or other control features
than the record size may be omitted.
CHANGE -- Character set converter                    Page 14


COMMAND SYNTAX . . . . . . . . 0
CONVERSION TABLES  . . . . . . 10

EXAMPLES . . . . . . . . . . . 11
EXTENDED COMMANDS  . . . . . . 6

SWITCHES . . . . . . . . . . . 1

TAPE FORMATS . . . . . . . . . 7