Monster Media 1996 #15/Monster Media Number 15 (Monster Media)(July 1996).ISO/prog_pas/tsfaqp31.zip/FAQPAS3.TXT

Text File | 1996-04-28 | 57KB | 1,493 lines

From [e-mail] Sun Apr 28 00:00:00 1996
Subject: FAQPAS3.TXT contents

                               Copyright (c) 1993-1996 by Timo Salmi
                                                 All rights reserved

FAQPAS3.TXT The third set of frequently (and not so frequently)
asked Turbo Pascal questions with Timo's answers. The items are in
no particular order.

You are free to quote brief passages from this file provided you
clearly indicate the source with a proper acknowledgment.

Comments and corrections are solicited. But if you wish to have
individual Turbo Pascal consultation, please post your questions to
a suitable Usenet newsgroup like news:comp.lang.pascal.borland. It
is much more efficient than asking me by email. I'd like to help,
but I am very pressed for time. I prefer to pick the questions I
answer from the Usenet news. Thus I can answer publicly at one go if
I happen to have an answer. Besides, newsgroups have a number of
readers who might know a better or an alternative answer. Don't be
discouraged, though, if you get a reply like this from me. I am
always glad to hear from fellow Turbo Pascal users.

....................................................................
Prof. Timo Salmi   Co-moderator of news:comp.archives.msdos.announce
Moderating at ftp:// & http://garbo.uwasa.fi archives  193.166.120.5
Department of Accounting and Business Finance  ; University of Vaasa
[e-mail] http://uwasa.fi/~ts BBS 961-3170972; FIN-65101,  Finland

--------------------------------------------------------------------
51) I am running out of memory when compiling my large program.
52) How do I avoid scrolling in the last column of the last row?
53) How can one hide (or unhide) a directory using a TP program?
54) How do I test whether a file is already open in a TP program?
55) How can I test and convert a numerical string into a real?
56) How can I reverse a TP .EXE or .TPU back into source code?
57) How can I calculate the difference between two points of time?
58) Is a program running stand-alone or from within the IDE?
59) Please explain Turbo Pascal memory addressing to me.
60) How do I obtain a bit or bits from a byte, a word or a longint?
61) What are Binary Coded Decimals? How to convert them?
62) How can I copy a file in a Turbo Pascal program?
63) How can I use C code in my Turbo Pascal program?
64) How do I get started with the Turbo Profiler?
65) How can I detect if the shift/ctrl/alt etc key is pressed?
66) How do I get a base 10 logarithm in TP?
67) If Delay procedure does not work properly, how do I fix it?
68) How much memory will my TP program require?
69) How to detect if a drive is a CD-ROM drive?
70) How do I convert an array of characters into a string?
71) How do I get started with graphics programming?
72) Where to I find the different sorting source codes?
73) A beginner's how to write and compile units.
74) What are and how do I use pointers?
75) How can I read another program's errorlevel value in TP?
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:00:51 1996
Subject: Out of memory in compiling

51. *****
 Q: I am running out of memory when compiling my large program. What
    can I do?

 A: If you are compiling your program from within the IDE (the
Integrated Development Environment) then select the Option from the
main menu, choose the Compiler item and set the Link buffer to
Disk. (Also make the Compile option Destination to be Disk).
   If this is not sufficient, next resort to using the TPC command
line version of the Turbo Pascal compiler instead of the IDE.  Use
the "Link buffer on disk" option.
   Divide your program into units. It is advisable anyway for
modularity when your program size grows.
   If you have extended memory, instead of TURBO.EXE use TPX.EXE, if
you have TP 7.0. If you are into protected mode programming then use
Borland Pascal BP 7.0.

 A2: If you would prefer compiling your program from within the IDE
but cannot do it for the above reason (or if you would prefer to
compile your program from within your favorite editor instead of the
TP IDE) you can use the following trick. If your editor has a macro
language like most good editors do, the assign a hotkey macro that
compiles the current file with the TPC. If you are using SemWare's
QEdit editor you'll find such macros in ("Macros and configurations
for QEdit text-editor") ftp://garbo.uwasa.fi/pc/ts/tsqed18.zip and
in ftp://garbo.uwasa.fi/ts/tstse18.zip ("SAL macro sources to extend
The SemWare Editor version 2.5").
   Also your editor must be swapped to disk during the compilation
if memory is critical. There is a very good program for doing that:
ftp://garbo.uwasa.fi/pc/sysutil/shrom24b.zip ("Shell Room, Swap to
disk when shelling to application"). For example I invoke the QEdit
editor with using the following batch:
 c:\tools\shroom -s r:\cmand -z 1024 c:\qedit\q %1 %2 %3 %4 %5 %6 %7
You'll find more about the switches in the Shell Room documentation.
The -s switch designates the swap destination (my r:\cmand directory
is on my ramdisk). The -z switch sets the shell environment size.
  An unfortunate part is that the TP 5.0 Turbo Pascal IDE is about
the only program I know that is not amenable the to Shell Room
utility, so you cannot utilize Shell Room to swap the TP IDE to
disk. Blessfully, at least TP 7.0 no more has this problem.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:00:52 1996
Subject: Last position write woes

52. *****
 Q: How do I avoid scrolling in the last column of the last row?

 A: If you use write or writeln at the last column of the last row
(usually 80,25) the screen will scroll. If you wish to avoid the
scrolling you'll have to use an alternative write that does not move
the cursor. Here is a procedure to write without moving the cursor
  uses Dos;
  procedure WriteChar (Character : char; fgColor, bgColor : byte);
  var r : registers;
  begin
    FillChar (r, SizeOf(r), 0);
    r.ah := $09;
    r.al := ord(Character);
    r.bl := (bgColor shl 4) or fgColor;
    r.cx := 1;    { Number of repeats }
    Intr ($10, r);
  end;  (* writechar *)
Thus, if you wish to write to the last column of the last row, you
must first move the cursor to that position. That can be done in
alternative ways. One might get there by having written previously
on the screen (with writeln and write routines) until one is in that
position. Another alternative is using GoToXY(80,20), but then you
have to use the Crt unit. If you don't want to use it, then you can
move the cursor by employing "GOATXY As the ordinary GoToXY but no
Crt unit required" from ftp://garbo.uwasa.fi/pc/ts/tspa3470.zip.
   There is an alternative interrupt service ($0A) which does the
same as service $09, but uses the default colors instead. Just
substitute $0A for $09, and leave the r.bl assignment out of the
WriteChar routine.
   Another option for writing anyhere on the screen without
affecting the cursor is using direct screen writes:
  uses Dos;
  procedure WriteChar (c : char; x, y : byte; fg, bg : byte);
  var vidstart : word;
      regs     : registers;
  begin
    FillChar (regs, SizeOf(regs), 0);
    regs.ah := $0F;
    Intr ($10, regs);  { Color or MonoChrome video adapter }
    if regs.al = 7 then vidstart := $B000 else vidstart := $B800;
    mem[vidstart:((y-1)*80+x-1)*2] := ord(c);
    mem[vidstart:((y-1)*80+x-1)*2+1] := (bg shl 4) or fg;
  end;
To write to the last position simply apply e.g.
  WriteChar ('X', 80, 25, 14, 0);  { Yellow on black }
The foreground (fg) and the background (bg) color codes are
  Black        =   0
  Blue         =   1
  Green        =   2
  Cyan         =   3
  Red          =   4
  Magenta      =   5
  Brown        =   6
  LightGray    =   7
  DarkGray     =   8
  LightBlue    =   9
  LightGreen   =  10
  LightCyan    =  11
  LightRed     =  12
  LightMagenta =  13
  Yellow       =  14
  White        =  15
  Blink        = 128
Yet another option is the following, but it needs the Crt unit. On
the other hand, it uses the default color. This is quite a good and
easy solution. I captured this fairly frequent idea from a posting
by Robert Buergi ([e-mail]).
  uses Crt;
  procedure WriteToCorner (c : char);
  begin
    Inc (WindMax);
    GotoXY (80, 25);
    write (c);
    Dec (WindMax);
  end;  (* writeToCorner *)
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:00:53 1996
Subject: Hiding a directory

53. *****
 Q: How can one hide (or unhide) a directory using a TP program?

 A: Here is the code using interrupt programming. Incidentally,
since MS-DOS 5.0 the attrib command can be used to hide and unhide
directories.
(* Hide a directory. Before using it would be prudent to check
   that the directory exists, and that it is a directory.
   With a contribution from Jan Nielsen [e-mail]
   Based on information from Duncan (1986), p. 410 *)
procedure HIDE (dirname : string);
var regs : registers;
begin
  FillChar (regs, SizeOf(regs), 0);  { standard precaution }
  dirname := dirname + #0;           { requires ASCIIZ strings }
  regs.ah := $43;                    { function }
  regs.al := $01;                    { subfunction }
  regs.ds := Seg(dirname[1]);        { point to the name }
  regs.dx := Ofs(dirname[1]);
  regs.cx := 2; { set bit 1 on }     { to unhide set regs.cx := 0 }
  Intr ($21, regs);                  { call the interrupt }
  if regs.Flags and FCarry <> 0 then { were we successful }
    writeln ('Failed to hide');
end;  (* hide *)

 A2: An alternative method by Dr. Abimbola Olowofoyeku
[e-mail]. No paths.
  procedure HIDE (dirname : string);
  var FileInfo : searchRec;
      f        : file;
  begin
    FindFirst (dirname, Directory, FileInfo);
    while DosError = 0 do
      begin
        assign (f, FileInfo.Name);
        SetFAttr (f, Hidden);
        FindNext (FileInfo);
      end;
  end;  (* hide *)
  {}
  procedure UNHIDE (dirname : string);
  var FileInfo : searchRec;
      f        : file;
  begin
    FindFirst (dirname, AnyFile, FileInfo);
    while DosError = 0 do
      begin
        assign (f, FileInfo.Name);
        SetFAttr (f, Archive);
        FindNext (FileInfo);
      end;
  end;  (* unhide *)
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:00:54 1996
Subject: Testing file opened status

54. *****
 Q: How do I test whether a file is already open in a TP program?

 A: This question is best answered by providing the code:
  uses Dos;
  {... for non-text files ...}
  function ISFOPEN (var filePointer : file) : boolean;
  begin
    isfopen := FileRec(filePointer).mode <> FmClosed;
  end;
  {}
  {... for text files ...}
  function ISTOPEN (var filePointer : text) : boolean;
  begin
    istopen := TextRec(filePointer).mode <> FmClosed;
  end;
  {}
  procedure TEST;          { Testing a non-text file }
  const name = 'R:\TMP';
  var f  : file;
  begin
    Assign (f, name);
    writeln ('File ', name, ' is open is ', ISFOPEN(f));
    {$I-} rewrite (f); {$I+}
    if IOResult <> 0 then
      begin
        writeln ('Failed to open ', name);
        exit;
      end;
    writeln ('File ', name, ' is open is ', ISFOPEN(f));
    close(f);
    writeln ('File ', name, ' is open is ', ISFOPEN(f));
  end;
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:00:55 1996
Subject: From string to real

55. *****
 Q: How can I test and convert a numerical string into a real?

 A1: An easy task in Turbo Pascal but in standard Pascal this
frequent task is much trickier. Here are both the Turbo Pascal and
Standard Pascal versions for general edification :-).
  (* Convert and test a numerical string with Turbo Pascal *)
  function DIGVALFN (mj : string; var ok : boolean) : real;
  var k : integer;
      x : real;
  begin
    Val (mj, x, k);
    ok := k = 0;
    if ok then digvalfn := x else digvalfn := 0;
  end;  (* digvalfn *)
  {}
  (* Convert and test a numerical string with standard Pascal routines only *)
  procedure DIGVAL (mj : string; var number : real; var ok : boolean);
  label 1;
  var il, lenl, pl, kl1, kl2 : integer;
      nrol                   : boolean;
      numberdl               : real;
  begin
    ok := true; lenl := Length (mj); nrol := false; pl := 0; number := 0.0;
    if lenl = 0 then ok := false;
    for il:=2 to lenl do if (mj[il]='-') or (mj[il]='+') then ok := false;
    for il:=1 to lenl do
      case mj[il] of
        '0'..'9','+','-','.' : ; else ok := false;
      end;
    for il:=1 to lenl do
      case mj[il] of '0'..'9' : begin nrol := true; goto 1; end; end;
    1: if nrol = false then ok := false;
    for il:=1 to lenl do if mj[il] = '.' then pl := pl + 1;
    if pl > 1 then ok := false;
    kl1:=1; kl2:=lenl+1; if (mj[1]='-') or (mj[1]='+') then kl1 := 2;
    for il:=1 to lenl do if mj[il] = '.' then kl2 := il;
    if kl2-kl1 > 38 then ok := false;
    if ok then
      begin
        number:=0; numberdl:=0;
        for il:=kl1 to kl2-1 do number := (ord(mj[il])-48)+10*number;
        if kl2 < lenl+1 then
          for il:=lenl downto kl2+1 do
            numberdl := (ord(mj[il])-48)/10+numberdl/10;
        number := number + numberdl;
        if mj[1]='-' then number := -number;
      end; {if ok}
  end;  (* digval *)
  {}
  procedure TEST;
  var s : string; r : real; ok : boolean;
  begin
    s := '123.41';
    r := DIGVALFN (s, ok);
    if ok then writeln (r) else writeln ('Error in ', s);
    DIGVAL (s, r, ok);
    if ok then writeln (r) else writeln ('Error in ', s);
  end;

A2: The conversion can be in the other directorion as well. Here is
how to convert an integer into a string with standard Pascal
routines only.
  function CHRIVLFN (number : integer) : string;
  var il, pl, al : integer;
  cl, mj : string;
  isNeg : boolean;
  begin
    if number < 0 then begin
      isNeg := true; number := -number; end
      else isNeg := false;
    pl := 0; mj := ''; cl := '';
    repeat
      pl := pl + 1;
      al := number mod 10;
      cl := cl + chr(al+48);
      number := number div 10;
    until number = 0;
    if isNeg then begin pl := pl + 1; cl[pl] := '-'; end;
    for il := 1 to pl do mj := mj + cl[pl+1-il];
    chrivlfn := mj;
  end;  (* chrivlfn *)
  {}
  procedure TEST;
  var s : string; j : integer;
  begin
    j := 12341;
    s := CHRIVLFN (j);
    writeln (s);
  end;
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:00:56 1996
Subject: Decompiling a TP .EXE

56. *****
 Q: How can I reverse a TP .EXE or .TPU back into source code?

 A: This is simply asking too much. You cannot decompile a TP
program in a manner that would give you back the original source.
This method of reverse engineering is not on in actual practice.
Quoting Jeroen Pluimers ([e-mail]) "During the
compilation, important information get's lost about variables,
types, identifiers etc. Writing a Pascal Decompiler is impossible.
The best you can achieve is a disassembler that can help you
recognize some Pascal statements."
   You might note that this question somewhat resembles another
frequent question "How can I convert a TPU unit of one TP version to
another?" which cannot be solved without the original source code.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:00:57 1996
Subject: Calculating date/time differences

57. *****
 Q: How can I calculate the difference between two points of time?

 A: This is an unconfirmed answer so be a little careful with it.
But at the very least it shows some interesting information about
Turbo Pascal date/time conventions and how to declare and initialize
typed constants if they are records.
  program TimDifTest;
  uses Dos;
  const a : DateTime
          = (year:1992; month:10; day:24; hour:5; min:29; sec:38);
        b : DateTime
          = (year:1993; month:11; day:23; hour:6; min:30; sec:51);
  var aLong, bLong, cLong : longint;
      c : DateTime;
  begin
    PackTime (a, aLong);
    PackTime (b, bLong);
    cLong := bLong - aLong;
    UnpackTime (cLong, c);
    writeln (c.year-1980, ' ', c.month, ' ', c.day, ' ',
             c.hour, ' ', c.min, ' ', c.sec);
  end.
More generally than for dates between 1980 and 2079, or for more
accurate results, the difference between two date/times can be
calculated using Zeller's congruence (see the item "I want code that
gives the weekday of the given date"). First calculate Zeller's for
both the dates, convert them, and the hour, min, and sec into
seconds, subtract, and convert back.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:00:58 1996
Subject: Stand-alone or from IDE

58. *****
 Q: Is a program running stand-alone or from within the IDE?

 A: Not all questions have an answer yet. I posed this question to
the the late UseNet newsgroup comp.lang.pascal, but we have not
found an answer that would be general for all MS-DOS versions. The
closest we have comes from [e-mail] Duncan Murdoch
(naturally :-). I have done some slight editing of Duncan's
solution.
  uses Dos;
  type Pchar = ^Char;
  function Asciiz2Str (p : Pchar) : string;
  var
    result : string;
    len : byte;
  begin
    len := 0;
    while (p^ <> #0) and (len < 255) do
    begin
      inc(len);
      result[len] := p^;
      inc(longint(p));
    end;
    result[0] := chr(len);
    Asciiz2Str := result;
  end;
  {}
  var parentSeg : ^word;
      p         : pchar;
  begin
    if swap(DosVersion) < $0400 then
      writeln ('Requires Dos 4.0+')
    else begin
      parentSeg := ptr (prefixSeg, $16);
      p := ptr (ParentSeg^-1, 8);
      writeln ('I was launched by ', Asciiz2Str(p));
    end;
  end.
Another suggestion has been that the contents of ParamStr(0) would
show the launching program. I tested this on several configurations
and TP versions and found no evidence that the contention would
hold.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:00:59 1996
Subject: Memory Addressing

59. *****
 Q: Please explain Turbo Pascal memory addressing to me.

 A: This is far from an easy question, but let's see what we can do.
The origins of the difficulties are in the design of the 8086 chip
which still restricts all Turbo Pascal applications (which contrary
to Borland Pascal use the original real mode). The 8086 (aka real
mode) addressing is based on 16-bit registers. As you probably know
2^16 is 65536 which means that you cannot directly point to all
addresses of the lower and upper memory, which ranges from 0 to
1048575 (2^20-1). Thus all the memory addresses are pointed to into
two parts in TP programs, the segment and the offset. The following
example of the PC's memory illustrates.

  Decimal  Hexa-
  address  decimal  Segment  Offset  What
        0   $00000    $0000   $0000  Conventional memory starts
     1043   $00413    $0040   $0013  Base memory in Kb, a word
   655359   $9FFFF    $9000   $FFFF  Conventional memory ends
   655360   $A0000    $A000   $0000  Upper memory begins
  1048575   $FFFFF    $F000   $FFFF  Upper memory ends

To exemplify, let's look at some alternative ways we could access
the information about the amount of the base memory. It is very
straight-forward, since in a PC that information is at the fixed
memory location show by the above table. We know this in advance.
Using direct memory accessing we could write
  var memsize : word;
  memsize := MemW [$0040:$0013];
  writeln (memsize);
  {.. or ..}
  var memsize : word absolute $0040:$0013;
  writeln (memsize);
If you are not familiar with the true meaning of pointers, they may
feel confusing, but what they basically are is just what the name
indicates, pointers to memory locations. Study the following
example.
  var memsizePtr : ^word;           { A pointer to a word }
  begin
    memsizePtr := ptr ($40, $13);   { Assign the pointer a value }
    writeln (memsizePtr^);          { Write what is in the address }
  end.                              { that was pointed to }
This was relatively simple, since we knew in advance the location of
the information. Lets look at a case where we do not know that.
Consider
  var x : word;
  begin
    x := 1223;
    writeln (x);
  end.
We have a variable x somewhere in the memory, and we can refer to it
without ever needing to know where the variable actually is in the
memory. But how does one find out if one for some reason wants or
needs to know?
  var x       : word;
      Segment : word;
      Offset  : word;
      y       : ^word;
  begin
    x := 1223;
    writeln (x);
    Segment := Seg(x);
    Offset  := Ofs(x);
    writeln ('Variable x is at $', HEXFN(Segment), ':$', HEXFN(Offset));
    {... one test to ensure that the value really is in there ...}
    writeln (MemW [Segment:Offset]);
    {... another test to demonstrate that the value really is in there ...}
    y := Addr(x);
    writeln (y^);
  end.
Next consider
  var xPtr    : ^word;
      Segment : word;
      Offset  : word;
      yPtr    : ^word;
  begin
    xPtr^ := 1223;
    writeln (xPtr^);
    Segment := Seg(xPtr^);
    Offset  := Ofs(xPtr^);
    writeln ('$', HEXFN(Segment), ':$', HEXFN(Offset));
    {... a test to ensure that the value really is in there ...}
    yPtr := Ptr (Segment, Offset);
    writeln (yPtr^);
  end.
A further aspect of pointers is that you can utilize them to put a
variables onto the heap instead of the data segment so that you
won't run so easily out of space.
  var xPtr : ^word;
  begin
    { Put it onto the heap }
    New (xPtr);
    xPtr^ := 1223;
    writeln (xPtr^);
    { Get rid of it }
    Dispose (xPtr); xPtr := nil;
    readln;
  end.
Let us return to the addressing. The formulas for converting between
the addresses (sent in by Duncan Murdoch) are
  Physical := longint(segment)*16 + offset;
  {}
  Segment  := Physical div 16;
  Offset   := Physical mod 16; { This gives the normalized form }
There are multiple Segment:Offset pairs that refer to the same
address, e.g. $0000:$0413 and $0040:$0013. The normalized addresses,
with the offset in the range of 0 to $F, are, however, unique. An
example $0041:$0003.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:00:60 1996
Subject: Getting a bit from a byte

60. *****
 Q: How do I obtain a bit or bits from a byte, a word or a longint?

 A: For bit operations think of the variable as a binary number
instead of a decimal. Consider for example
  var x : word;
  x := 219;
In binary presentation it is
  The word                  0000 0000 1101 1011
  Position in the word      FEDC BA98 7654 3210

Say you need the value of bit 6 (the seventh bit) in the word. You
can "and" the following words
  0000 0000 1101 1011    (219)
  0000 0000 0100 0000    ( 64)
In decimal TP notation this amounts to
  var b : word;
  b := x and 64;
The value of b is now
  0000 0000 0100 0000    ( 64)
To get the bit value (0 or 1) you need to shift the result right by
six steps, that this the expression becomes the often seen but
cryptic
  b := (x and 64) shr 6;
which means that the value of b is finally 1 in this example.

Ok, but what then if you need the combined value of bits six and
seven. The answer is evident if you consider the binary presentation
  0000 0000 1101 1011    (219)
  0000 0000 1100 0000    (192)
hence
  b := (x and 192) shr 6;
which will give 3 as it should.

So far, so good. What if you need to turn on bit nine in a word
without interfering with the other bits. The binary presentation,
again, is the key. You'll have to "or" the following
  0000 0000 1101 1011    (219)
  0000 0010 0000 0000    (512)
that is
  x := x or 512;
This results to
  0000 0010 1101 1011    (731)

What if you wish to turn off, say bit 6, in
  0000 0000 1101 1011    (219)
  1111 1111 1011 1111  (65471)
This is achieved by
  x := 219;
  x := x and 65471;
This results to
  0000 0000 1001 1011    (155)

Consider the following application as an example. The number of a
PC's floppy disk drives (minus one) is stored in bits 6 and 7 in a
word returned by interrupt $11. This is the code to find out how
many disk drives a PC has.
  uses Dos;
  function NrOfFDiskDrives : byte;
  var regs : registers;
  begin
    Intr ($11, regs);
    NrOfFDiskDrives := ((regs.ax and 192) shr 6) + 1;
  end;

A tip from Duncan Murdoch.  You might wish to predefine the
following constants for easier handling
  const bit0  = 1;
        bit1  = 2;
        bit2  = 4;
        :
        bit15 = 32768;
        :
        bit31 = 2147483648;
Or to put it slightly differently as Dr John Stockton
[e-mail] suggests
  const
  bit00=$00000001; bit01=$00000002; bit02=$00000004; bit03=$00000008;
  bit04=$00000010; bit05=$00000020; bit06=$00000040; bit07=$00000080;
  :
  bit28=$10000000; bit29=$20000000; bit30=$40000000; bit31=$80000000;
Finally, you also might want to look at the item "Getting a nybble
from a byte".
--------------------------------------------------------------------
From [e-mail] Sun Apr 28 00:01:01 1996
Subject: Binary Coded Decimals

61. *****
 Q: What are Binary Coded Decimals? How to convert them?

 A: Let us look at full integers only and skip the even more
difficult question of BCD reals and BCD operations.
     Decimal Hexa  BCD
        1      $1    1
        :      $9    9
       10      $A   ..
        :       :    :
       12      $C   ..
        :       :    :
       16     $10   10
       17     $11   11
       18     $12   12
        :       :    :
Consider the last value, that is BCD presentation of 12. The
corresponding hexadecimal is $12 (not $C as in normal decimal to
hexadecimal conversion). The crucial question is how to convert
12BCD to $12 (or its normal decimal equivalent 18). Here is my
sample code:
  type BCDType = array [0..7] of char;
  {}
  procedure StrToBCD (s : string; var b : BCDType);
  var i, p : byte;
  begin
    FillChar(b, SizeOf(b), '0');
    p := Length (s);
    if p > 8 then exit;
    for i := p downto 1 do b[p-i] := s[i];
  end;  (* strtobcd *)
  {}
  function BCDtoDec (b : BCDType; var ok : boolean) : longint;
  const Digit : array [0..9] of char = '0123456789';
  var i, k : byte;
      y, d : longint;
  begin
    y := 0;
    d := 1;
    ok := false;
    for i := 0 to 7 do begin
      k := Pos (b[i], Digit);
      if k = 0 then exit;
      y := y + (k-1) * d;
      if i < 7 then d := 16 * d;
    end; { for }
    ok := true;
    BCDtoDec := y;
  end;  (* bcdtodec *)
  {}
  {}
  procedure TEST;
  var i  : byte;
      b  : BCDType;
      x  : longint;
      ok : boolean;
      s  : string;
  begin
    s := '12';
    StrToBCD (s, b);
    write ('The BCD value : ');
    for i := 7 downto 0 do write (b[i], ' ');
    writeln;
    x := BCDtoDec (b, ok);
    if ok then writeln ('is ', x, ' as an ordinary decimal')
      else writeln ('Error in BCD');
  end;  (* test *)
  {}
  begin TEST; end.

Next we can ask, what if the BCD value is given as an integer.
Simple, first convert the integer into a string. For example in
the procedure TEST put
  Str (12, s);

Finally, what about converting an ordinary decimal to the
corresponding BCD but given also as a decimal variable.  For example
18 --> 12?
  function LHEXFN (decimal : longint) : string;
  const hexDigit : array [0..15] of char = '0123456789ABCDEF';
  var i : byte;
      s : string;
  begin
    FillChar (s, SizeOf(s), ' ');
    s[0] := chr(8);
    for i := 0 to 7 do
      s[8-i] := HexDigit[(decimal shr (4*i)) and $0F];
    lhexfn := s;
  end;  (* lhexfn *)
  {}
  function DecToBCD (x : longint; var ok : boolean) : longint;
  const Digit : array [0..9] of char = '0123456789';
  var hexStr : string;
  var i, k : byte;
      y, d : longint;
  begin
    hexStr := LHEXFN(x);
    y := 0;
    d := 1;
    ok := false;
    for i := 7 downto 0 do begin
      k := Pos (hexStr[i+1], Digit);
      if k = 0 then exit;
      y := y + (k-1) * d;
      if i > 0 then d := 10 * d;
    end; { for }
    ok := true;
    DecToBCD := y;
  end;  (* dectobcd *)
  {}
  procedure TEST2;
  var i    : byte;
      x10  : longint;
      xBCD : longint;
      ok   : boolean;
  begin
    x10 := 18;
    writeln ('The ordinary decimal value : ', x10);
    xBCD := DecToBCD (x10, ok);
    if ok then writeln ('is ', xBCD, ' as a binary coded decimal')
      else writeln ('Error in BCD');
  end;  (* test2 *)
  {}
  begin TEST; end.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:02 1996
Subject: Copying with TP

62. *****
 Q: How can I copy a file in a Turbo Pascal program?

 A: Here is the code. Take a close look. It has some instructive
features besides the copying, like handling the filemode and using
dynamic variables (using pointers). Note that since the buffer for
the copying is places on the heap you must reserve enough heap. For
example you might have {$M 16384,0,102400}.
  procedure SAFECOPY (fromFile, toFile : string);
  type bufferType = array [1..65535] of char;
  type bufferTypePtr = ^bufferType;  { Use the heap }
  var bufferPtr : bufferTypePtr;     { for the buffer }
      f1, f2 : file;
      bufferSize, readCount, writeCount : word;
      fmSave : byte;              { To store the filemode }
  begin
    bufferSize := SizeOf(bufferType);
    if MaxAvail < bufferSize then exit;  { Assure there is enough memory }
    New (bufferPtr);              { Create the buffer, on the heap }
    fmSave := FileMode;           { Store the filemode }
    FileMode := 0;                { To read also read-only files }
    Assign (f1, fromFile);
    {$I-} Reset (f1, 1); {$I+}    { Note the record size 1, important! }
    if IOResult <> 0 then exit;   { Does the file exist? }
    Assign (f2, toFile);
    {$I-} Reset (f2, 1); {$I+}    { Don't copy on an existing file }
    if IOResult = 0 then begin close (f2); exit; end;
    {$I-} Rewrite (f2, 1); {$I+}  { Open the target }
    if IOResult <> 0 then exit;
    repeat                        { Do the copying }
      BlockRead (f1, bufferPtr^, bufferSize, readCount);
      {$I-} BlockWrite (f2, bufferPtr^, readCount, writeCount); {$I+}
      if IOResult <> 0 then begin close (f1); exit; end;
    until (readCount = 0) or (writeCount <> readCount);
    writeln ('Copied ', fromFile, ' to ', toFile,
             ' ', FileSize(f2), ' bytes');
    close (f1); close (f2);
    FileMode := fmSave;           { Restore the original filemode }
    Dispose (bufferPtr);          { Release the buffer from the heap }
  end;  (* safecopy *)

Of course a trivial solution would be to invoke the MS-DOS copy
command using the Exec routine. (See the item "How do I execute an
MS-DOS command from within a TP program?")
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:03 1996
Subject: C modules in TP

63. *****
 Q: How can I use C code in my Turbo Pascal program?

 A: I have very little information on this question, since I do not
program in C myself.  However in reading Turbo Pascal textbooks I
have come across a couple of references I can give.  They are Edward
Mitchell (1993), Borland Pascal Developer's Guide, pp. 60-64, and
Stoker & Ohlsen (1989), Turbo Pascal Advanced Techniques, Ch 4.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:04 1996
Subject: Using Turbo Profiler

64. *****
 Q: How do I get started with the Turbo Profiler?

 A: Borland's separate Turbo Profiler is a powerful tool for
improving program code and enhancing program performance, but far
from an easy to use. It is an advanced tool. In fact setting it up
the first time is almost a kind of detective work.
   Let's walk through the steps with Turbo Profiler version 1.01 to
see where a running Turbo Pascal program takes its time.
Assume a working directory r:\
1. Copy the target .PAS file to r:\
2. Compile it with TURBO.EXE using the following Compiler and
   Debugger options. The standalone debugging option is crucial.
     Code generation
      [ ] Force far calls        [X] Word align data
      [ ] Overlays allowed       [ ] 286 instructions
     Runtime errors             Syntax options
      [ ] Range checking         [X] Strict var-strings
      [X] Stack checking         [ ] Complete boolean eval
      [ ] I/O checking           [X] Extended syntax
      [ ] Overflow checking      [ ] Typed @ operator
                                 [ ] Open parameters
     Debugging
      [X] Debug information     Numeric processing
      [X] Local symbols          [ ] 8087/80287
                                 [ ] Emulation
     Debugging         Display swapping
      [X] Integrated    ( ) None
      [X] Standalone    () Smart
                        ( ) Always
3) Call TPROF.EXE
4) Load the .EXE file produced by compilation in item 2.
5) Choose from the TPROF menus
     Statistics
       Profiling options...
         Profile mode
          () Active    ( ) Passive
          Run count
           1
          Maximum areas
           200
6) Choose from the TPROF menus
      Options
       Save options...
      [X] Options
      [ ] Layout
      [ ] Macros
     Save To
      r:\tfconfig.tf
7) Press Alt-F10 for the Local Menu. Choose
     Add areas
       All routines
and so on.
8) Choose Run from the TPROF menus (or F9)
9) Choose from the TPROF menus
      Print
       Options...
     Width
      80
     Height
      9999
      ( ) Printer     ( ) Graphics
      () File        () ASCII
     Destination File
      r:\report.lst
10) Print
       Module...
         All modules
         Statistics
           Overwrite
Also see Edward Mitchell (1993), Borland Pascal Developer's Guide.
It has a a very instructive chapter "Program Optimization" on the
Turbo Profiler. The material in the Turbo Profiler manual is so
complicated that additional guidance like Mitchell's is very much
needed.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:05 1996
Subject: Detecting shift status

65. *****
 Q: How can I detect if the shift/ctrl/alt etc key is pressed? I
know how to get the scan codes with the ReadKey function, but I
can't find the procedure for detecting these keys.

 A: Detecting pressing the special keys or getting the toggle status
cannot be done with ReadKey. You'll need to access the Keyboard
Flags Byte at $0040:$0017. You can do this either by a direct "Mem"
access, or using interrupt $16 function $02. For more details
including the bitfields for the shift flags see in Ralf Brown's
interrupt list ftp://garbo.uwasa.fi/pc/programming/inter49a.zip (or
whatever is the current version). For example to see if the alt key
is pressed you can use
  uses Dos;
  function ALTDOWN : boolean;
  var regs : registers;
  begin
    FillChar (regs, SizeOf(regs), 0);
    regs.ah := $02;
    Intr ($16, regs);
    altdown := (regs.al and $08) = $08;
  end;
For the enhanced keyboard flags see interrupt $16 function $12. It
can distinguish also between the right and the left alt and ctlr
keys.
   A tip from Martijn Leisink [e-mail]. Be careful [if
you use the $0040:$0017 memory position to set a toggle]: On several
computers you have to call int 16h after the new setting is shown by
the LED's on the keyboard. Not doing so might give the user wrong
information.
   A tip from Dr John Stockton [e-mail]. Going via a
BytePointer set to Ptr(Seg0040, $0017) is almost as easy as "Mem",
and also works in Protected mode.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:06 1996
Subject: Base 10 logarithm

66. *****
 Q: How do I get a base 10 logarithm in TP?

 A: Just define
     function log (x : real) : real;
     begin log := ln(x) / ln(10); end;
This result is based on some elementary math. By definition
y = log(x) in base 10 is equivalent to x = 10^y (where the ^
indicates an exponent). Thus ln(x) = y ln(10) and hence
y = ln(x) / ln(10).
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:07 1996
Subject: Replacing Delay procedure

67. *****
 Q: If Delay procedure does not work properly, how do I fix it?

 A: The Delay procedure in the Crt unit delays a specified number of
milliseconds. It is declared as "procedure Delay(MS: Word);". There
are two problems. The procedure requires using the Crt unit and
there is a bug in it in TP 6.0, at least. The alternative is to use
the procedure GetTime(var Hour, Minute, Second, Sec100: Word) as
shown by the skeleton below
  GetTime (...)
  initialTime := ...
  repeat
    GetTime (...)
    interval := ... - initialTime;
  until interval >= YourDelay;
There are two things you will have to see to. You will have to
convert the time to sec100, and you will have to take care of the
possibility of the interval spanning the midnight. If you do not
wish to program the alternative Delay procedure yourself, you can
use "DOSDELAY Delay without needing the Crt unit" from TSUNTD.TPU
from ftp://garbo.uwasa.fi/pc/ts/tspa3470.zip.

 A2: Dr John Stockton [e-mail] suggested procedure that is
expanded below. It has the advantage of being concise and working in
the protected mode. The disadvantage is that it requires a later TP
version. The solution is quite instructive.
  uses Dos;
  {... John's procedure ...}
  procedure WAIT (SecondsDelay : real) ;
  Var Tptr : ^longint ; Finish : longint ;
  begin
    Tptr := Ptr(Seg0040, $006C) ;
    Finish := Tptr^ + Round(18.2*SecondsDelay) ;
    repeat until Tptr^ > Finish ;
  end;
  {... now let's test it ...}
  var h1, m1, s1, sa100 : word;
      h2, m2, s2, sb100 : word;
  begin
    GetTime (h1, m1, s1, sa100);
    WAIT (3);
    GetTime (h2, m2, s2, sb100);
    writeln (h1, ':', m1, ':', s1, '.' ,sa100);
    writeln (h2, ':', m2, ':', s2, '.' ,sb100);
  end.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:08 1996
Subject: TP program memory requirement

68. *****
 Q: How much memory will my TP program require?

 A: Get MAPMEM.EXE from ftp://garbo.uwasa.fi/pc/memutil/tsrcom35.zip
and put the following code within your Turbo Pascal program:
  Program faq;
  uses Dos;
  :
  SwapVectors;
  Exec (GetEnv('comspec'), '/c mapmem');
  Swapvectors;
Then you'll see a MAPMEM output something like this
  Psp  Cnt   Size Name       Command Line        Hooked Vectors
  ---- --- ------ ---------- ------------------- --------------
        2  26,896 DOS
  0694  2   3,392 COMMAND                        2E
        1      64 ---free---
  0776  2   1,488 MARK       scrollit
  07D6  2  70,816 FAQ                            FF
  1923  3   2,752 command                        22 23 24
  19D2  2 549,712 ---free---
          655,344 ---total--
The memory requirement of your program FAQ.PAS is 70,816. Do not
confuse this figure with the physica size of your program. The
memory requirement affected among other things by the Memory
Allocation Sizes Directive. For example you might have
{$M 16384,0,50000}

-Date: Sun, 12 Jun 1994 10:22:18
-From: [e-mail] (Duncan Murdoch)
-Newsgroups: comp.lang.pascal
-Subject: Re: How much memory will my TP program require?

   I think this is a hard question, and probably needs a longer
answer than you gave.  Yours isn't quite right, because TP will
allocate memory that it doesn't need if you set the heapmax
parameter too high.  Your program will run in less memory than
MAPMEM reports. Here's a quick attempt at it:
   TP DOS programs use memory in 4 or 5 blocks:  fixed code, static
data, the stack, sometimes overlaid code, and the heap.  TP Windows
programs add a local heap to this list, but don't use overlays.  The
discussion below deals with real mode DOS programs.
   The size of the code is determined by which procedures and
functions you use in your program.  In DOS, if you don't use
overlays, this is all fixed code, and the size is reported as "Code
size" in the Compile| Information listing in the IDE.  The ways to
reduce it are to use fewer procedures or make them smaller, or move
them to overlays.
   Static data consists of all the global variables and typed
constants in every unit.  It is reported as "Data size" in the
Compile|Information listing.  You can reduce it by declaring fewer
or smaller variables.
   If you use the $O directive to move code to overlays, then those
units won't count as part of your fixed code needs.  You will need
an overlay buffer at run-time; by default, it's the size of the
largest unit you use, but normally you'll change the size with
OvrSetBuf.  It's difficult to work out the best size of this block
except by trial and error:  if your program spends too much time
swapping, then make it larger; if you run out of memory, make it
smaller.  You'll need to use the .MAP file (see the Options| Linker
dialog to create one) to find the size of each unit.  Remember to
subtract the size of overlaid units from the reported "Code size"
when working out the size of fixed code.
  The stack is used for local variables in procedures.  Its size is
controlled by the first parameter to the $M directive; the default
size is 16K.  It's hard to predict exactly how much stack space your
program will use.  One way is to keep reducing the value until your
program aborts with a stack overflow, then use a slightly larger
value.  Another way is to fill the stack with a fixed value at the
start of your program, and at the end, see how many values were
changed.  Again, it's a good idea to allow for a margin of safety,
because hardware interrupts will use this space, and their size is
hard to predict.
   The heap is where New and Getmem get their allocated memory.  The
size is controlled by the 2nd and 3rd parameters to the $M
directive.  The heapmin value will always be allocated; if extra
memory is available, your program will ask for as much as possible,
up to heapmax.  If not enough memory is available to load all your
fixed code, data, stack and heapmin, DOS will refuse to load your
program.  You have nearly complete control over the size of the heap
that you need, determined by how much you use New and Getmem. The
only exception is that some of the standard units use heap space;
GRAPH and all the TurboVision units are examples.  To find how much
your program actually uses, you can reduce Heapmax until it fails,
fill the heap with a special value and look for changes, or monitor
the value of HeapPtr as your program progresses.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:09 1996
Subject: Detecting a CD-ROM drive?

69. *****
 Q: How to detect if a drive is a CD-ROM drive?

 A: There are several methods to do this. Here is one option.
  (* Is a drive a CD-ROM with MSCDEX driver installed *)
  function CDROMFN (drive : char) : boolean;
  var regs : registers;
  begin
    cdromfn := false;
    if swap(DosVersion) < $0200 then exit;
    drive := UpCase(drive);
    if (drive < 'A') or (drive > 'Z') then exit;
    FillChar (regs, SizeOf(regs), 0);
    regs.cx := ord(drive) - ord('A');
    regs.ax := $150B;
    Intr ($2F, regs);
    cdromfn := (regs.ax <> 0) and (regs.bx = $ADAD);
  end;  (* cdromfn *)
The other relevant $2F interrupt functions you can use are $1500,
$1501, and in particular $150D.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:10 1996
Subject: Array of chars into string

70. *****
 Q: How do I convert an array of characters to a string? More
specifically, I haven't been able to convert an array of characters
into a string, so that I can write it to a file. The only way I have
been able to do it, is writing 1 char at a time.

 A: Carefully study these two simple test examples.  Note the
difference in the array's dimensions in the tests.

  type atype = array [0..20] of char;
  type stype = string[20];
  var s : stype;
      a : atype absolute s;
  begin
    FillChar (a, SizeOf(a), '*');
    s[0] := chr(20);
    writeln (s);
  end.

  type atype = array [1..20] of char;
  var s : string;
      a : atype;
  begin
    FillChar (a, Sizeof(a), '*');
    Move (a, s[1], 20);
    s[0] := chr(20);
    writeln (s);
  end.

Of course, you could also assign the array's characters one by one
to the string using a simple for loop (left as an exercise), but the
above methods are more efficient.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:11 1996
Subject: Graphics programming primer

71. *****
 Q: How do I get started with graphics programming?

 A:
  (* This simple test shows the rudiments of getting started with Turbo
     Pascal graphics programming *)
  uses Crt, Graph;
  var grDriver : integer;
      grMode   : integer;
      ErrCode  : integer;
      i, j     : integer;
      xm, ym   : integer;
  const CharSize : integer = 3;
  begin
    { Request graphics driver autodetection }
    grDriver := Detect;
    { Initialize graphics system and put hardware into graphics mode }
    { The relevant .bgi driver is needed in the current directory
      for example egavga.bgi }
    InitGraph (grDriver, grMode, ' ');
    { Return an error code for the previous graphic operation }
    ErrCode := GraphResult;
    { Test for initialialization success }
    if ErrCode <> grOk then begin
      Writeln ('Graphics error:', GraphErrorMsg(ErrCode)); halt; end;
    { Clear the output device and home the current pointer }
    ClearDevice;
    {}
    { Use your own coordinates }
    xm := Round (GetMaxX / 100.0);
    ym := Round (GetMaxY / 100.0);
    {}
    { Set the current line width and style, optional }
    SetLineStyle (SolidLn, 0, ThickWidth);
    { Set the drawing color }
    SetColor (Yellow);
    { Draw a line }
    Line (70*xm, 50*ym, 90*xm, 80*ym);
    {}
    { Drawing bars }
    { Set the fill pattern and color }
    SetFillStyle (SolidFill, Red);
    Bar (0, 0, 25*xm, 25*ym);
    {}
    SetColor (Magenta);
    SetFillStyle (SolidFill, Blue);
    Bar3D (30*xm, 20*ym, 50*xm, 60*ym, 8*xm, TopOn);
    {}
    { Writing text in the graphics mode }
    { Set the drawing color }
    SetColor (LightCyan);
    { Set the current background color }
    SetBkColor (Black);
    { Set style for text output in graphics mode }
    SetTextStyle(DefaultFont, HorizDir, CharSize);
    OutTextXY (0, 80*ym, 'Press any key');
    {}
    repeat until KeyPressed;
    {}
    { Restore the original screen mode before graphics was initialized }
    RestoreCrtMode;
    writeln ('That''s all folks');
    { Shut down the graphics system }
    CloseGraph;
  end.
For an example what you can do with graphics, see
 111673 Oct 8 1993 ftp://garbo.uwasa.fi/pc/ts/tsdemo16.zip
 tsdemo16.zip Assorted graphics demonstrations of functions etc
(or whatever is the current version).
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:12 1996
Subject: Sorting it out

72. *****
 Q: Where to I find the different sorting source codes?

 A: I'll answer very briefly by giving two references:
 303771 May 2 1991 ftp://garbo.uwasa.fi/pc/turbopas/nrpas13.zip
 nrpas13.zip Numerical Recipes Pascal shareware version
and
 Gary Martin (1992), Turbo Pascal, Theory and Practice of Good
Programming, Chapter 15.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:13 1996
Subject: TP units

73. *****
 Q: A beginner's how to write and compile units.

 A1: Many of the text-books in the bibliography section of this FAQ
discuss using units in Turbo Pascal. For example see Tom Swan
(1989), Mastering Turbo Pascal 5.5, Chapters 9 and 10 for a more
detailed discussion than the rudiments given in the current item.
Likewise see your Turbo Pascal (7.0) User's Guide Chapter 6, "Turbo
Pascal units".
   You can and need to write your own units if you need recurring or
common routines in your programs and/or your program becomes so big
that it cannot be handled as a single entity.
   A Turbo Pascal unit is a separate file which you compile. The
following trivial example to calculate the sum of two reals
illustrates the basic structure of a unit.
  { The name of this file must be faq73.pas to correspond. }
  unit faq73;
  {}
  { The interface section lists definitions and routines that are }
  { available to the other programs or units. }
  interface
  function SUMFN (a, b : real) : real;
  {}
  { The implementation section contains the actual unit program }
  implementation
  function SUMFN (a, b : real) : real;
  begin
    sumfn := a + b;
  end;
  {}
  end.
When you compile the file FAQ73.PAS a unit FAQ73.TPU results. Next
an example utilizing the faq73 unit in the main program.
  uses faq73;
  {}
  procedure TEST;
  var x, y, z : real;
  begin
     x := 12.34;
     y := 56.78;
     z := SUMFN (x, y);
     writeln (z);
  end;
  {}
  begin
    TEST;
  end.

 A2: Most often you would be compiling a Turbo Pascal program
using the IDE (Integrated Development Environment). If you have
precompiled units you must see to it that you have informed the IDE
of the path to them.
   Press F10 and invoke the "Options" menu (or press alt-O). Select
"Directories...". Press tab two times to get to "Unit directories"
and edit the path accordingly. Here is what I have entered myself
  EXE & TPU directory  r:\
  Include directories  r:\
  Unit directories     f:\progs\turbo70\tpu70
  Object directories   f:\progs\turbo70\tpu70
As you see I keep all my precompiled Turbo Pascal 7.0 units in the
f:\progs\turbo70\tpu70 directory.
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:14 1996
Subject: Beginners' pointers

74. *****
 Q: What are and how do I use pointers?

 A: This is a beginner's simplified introduction. A pointer is a
variable type used to hold the address of another variable, that is
to point to it. Pointers are used to
 1) To refer to and manipulate variables indirectly.
 2) In Turbo Pascal to obtain access to the heap storage area, which
    is not restricted to 64Kbytes.
Consider the following example
  {$M 16384,0,80000}
  var yPtr : ^real;
  begin
    New(yPtr);
    yPtr^ := 3.14159;
    writeln ('2 times pi = ',  2.0 * yPtr^);
    Dispose(yPtr);
    yPtr := nil;
  end.
Before we can discuss pointers we have to consider some rudiments of
what a kind of a memory model a compiled Turbo Pascal program uses.
This is a highly simplified presentation. For a more detailed
presentation of the TP memory model see for example Tischer (1990b).
  +-------------------------+
  | Heap                    |
  |-------------------------|
  | Data Segment            |
  |-------------------------|
  | Code                    |
  |-------------------------|
  | Program Segment Prefix  |
  +-------------------------+
When you write and compile a Turbo Pascal program it usually
consists of (this is a simplification!) of the three lowest parts.
When you define a global variable, it goes to the Data Segment. For
example defining at the beginning of your program
 var x : real;
requires 6 bytes from the data segment. (Local variables are placed
on the stack.)
   Now, the catch is that because of the underlying 16-bit nature of
MS-DOS, the size of the data segment cannot exceed 64Kb. On occasion
the 64Kb is insufficient. However, if you use pointers, the
corresponding variable values are held on the heap instead of the
data segment or the stack. Before you can use the heap, you have to
reserve it for your program. The following compiler directive makes
a heap of 80000 bytes available to your program {$M 16384,0,80000}.
(The syntax is {$M Stack size, Low heap limit, High heap limit}).
   With pointers you do not refer to a variable directly, but you
point to it. For example, define
  var yPtr : ^real;
Before you can use this pointer, you have to create this new dynamic
variable as follows:
  New(yPtr);
The New(yPtr) statement "Creates a new dynamic variable and sets a
pointer variable to point to it." This pointer, yPtr, will point to
the actual value, which the program puts on the heap. In your
program you can write, for example
  yPtr^ := 3.14159;
Think about the difference between yPtr and yPtr^. The former
contains the value of the memory address where you now have put the
value 3.14159. The latter gives that value. Hence yPtr^ can be used
like any ordinary real variable. The difference is that it is on the
heap, not on the data segment (or stack). Thus you can now use this
pointer. For example you n write
 writeln ('2 times pi = ',  2.0 * yPtr^);
When you do not need the pointer any more in your program you can
dispose of it to release the memory allocated for other purposes:
  Dispose(yPtr);
  yPtr := nil;
"After a call to Dispose, the value of yPtr is undefined and it is
an error to reference yPtr. The reserved word nil denotes a pointer
type constant that does not point to anything." Setting yPtr := nil
is just good programming practice, because then you can later easily
test whether the pointer is available or not. Disposing of a pointer
within your program is not necessary unless the amount of memory is
a critical consideration in your program. The heap will be released
when your program terminates.
  To recount. What yPtr actually contains is the memory address of
the value on the heap. When you write yPtr^, the caret indicates
that you do not mean the pointer itself, but the pointed memory
location in the heap. In this example that memory location in the
heap was made to contain 3.14159.
   You can also define the pointer types. Our second example
illustrates. It displays the squares from one to ten.
  {$M 16384,0,80000}
  type arrayType = array [1..10] of real;
  type arrayPtrType = ^arrayType;
  var A : arrayPtrType;
      i : integer;
  begin
    if SizeOf(arrayType) > MaxAvail then begin
      writeln ('Out of memory');
      halt;
    end;
    New(A);
    for i := 1 to 10 do A^[i] := i*i;
    writeln (A^[9]);
  end.
For an actual application using pointers, see the item "How can I
copy a file in a Turbo Pascal program?"
--------------------------------------------------------------------

From [e-mail] Sun Apr 28 00:01:15 1996
Subject: Reading errorlevel

75. *****
 Q: How can I read another program's errorlevel value in TP?

 A: This question is best answered by an example. Here is a very
elementary program that returns errorlevel 14 on exiting.
  program faq2;
  begin
    writeln ('Hello world...');
    halt(14);
  end.
Below is the program that calls FAQ2.EXE and detects its errorlevel.
  {$M 2000,0,0}
  uses Dos;
  begin
    SwapVectors;
    Exec ('r:\faq2.exe', '');  (* Execution *)
    SwapVectors;
    WriteLn('...back from Exec');
    if DosError <> 0 then
      WriteLn('Dos error #', DosError)
      else
      WriteLn('Success; child process errorlevel = ', lo(DosExitCode));
  end.
The output should be
  Hello world...
  ...back from Exec
  Success; child process errorlevel = 14
--------------------------------------------------------------------