Bonus Lessons

Copyright © 1998-2001 by Not Another Writer, Inc.

All rights reserved
Main
About
News Just for being nice, I'm presenting you with some
FAQ Supplemental Lessons designed to help further your
conquest of the C language. Some of these were promised in
the book, others are done for the heck of it, still others are
prompted as solutions to reader's questions (the Bonus
lessons).
Supplemental Lessons

Other Information
Source Code Files
On-line Book Ordering
Related Web Pages 1. If you're looking for the Lessons promised in Chapter
11 (stuff on linked lists), see Chapter 17 below.

2. If you're looking for Chapter 13, see Chapter 13

below. (It's still under construction.)

This information here is updated at least once a month.

Check back often!

The Rules

You are granted the right to print one (1) copy of each
Lesson for your personal use. You cannot duplicate or
mass reproduce the Lesson(s) or distribute any
material from this Web site in any way. This material is
copyrighted. I'm giving you the right to make one copy
because you own C for Dummies and have, in a sense,
already paid me. If you do not own a copy of C for
Dummies, then buy one!

To view a new Lesson, click its link. Then use your

browser's print gizmo to print the Lesson out on your
printer. Staple the pages together, or use a three-ring binder
to create your own Supplemental Lesson Folder. I believe
you'll find that printing things out works better than trying to
read and learn from a computer screen.

The Supplemental Lessons

Chapter 13 - "The Missing Chapter"

Lesson 13.1 - Does Anyone Have the Time?

Lesson 13.2 - Still missing!

Lesson 13.3 - Say Hello to Mr. Bit

Lesson 13.4 - Still missing!

Lesson 13.5 - Color Text

Lesson 13.6 - Introduction to Recursion

Chapter 15 - "See C: C File Manipulation"

Lesson 15.1 - What Lurks on Disk

More to come!

Chapter 16 - "The Wonders of In-Line Assembly"

Lesson 16.1 - Introduction to In-Line Assembly

More lessons on the way!

Chapter 17 - "More Structures (Linked Lists)"

 Lesson 17.1 - The Birth of Linked Lists

Lesson 17.2 - The Adolescence of Linked Lists

Lesson 17.3 - Dawn of the Database

Lesson 17.4 - Decimating the Linked List

Lesson 17.5 - Free At Last!

Chapter 18 - "At Last, A Graphics Chapter"

Lesson 18.1 - Setting the Right Mode

Lesson 18.2 - Hello, Pixel Fairy!

Lesson 18.3 - Loverly Lines

Chapter 19 - "Beginning Windows Programming"

Don't hold your breath for this one!

Linux Supplement!

Linux #1 - The NOVOWELS.C delimma

Linux #2 - Debugging Tips with GDP

Linux #3 - Compiler options to know

Linux #4 - Doing the getch() thing in Linux

Bonus Lessons!

Bonus#1 - Restricting Input

Bonus#2 - An Elegant Float-to-String Conversion

Kludge

Bonus#3 - The Real Answer to the

while(*string++) Puzzle

Bonus#4 - Searching a File for a String

Bonus#5 - Running Another Program

Bonus#6 - Mousing Around

Bonus#7 - Reading Your Keyboard

Bonus#8 - Trouble Header

Bonus#9 - DOS Text Screen Border

Bonus#10 - Command Line Parsing with strtok

Bonus#11 - Multitasking in DOS

Bonus#12 - Putting to Screen Memory

Bonus#13 - Binary to Decimal, Anyone?

Bonus#14 - Sorting Strings

Bonus#15 - DROPNUKE.C workaround

Bonus#16 - Socket Programming Info

Bonus#17 - clrscr() in MSVC++

 Copyright © 1998-2002 by Not Another Writer, Inc.
All rights reserved

Bonus C For Dummies Lesson 13.1Lesson 13.1 – Does Anyone Have the Time?

C has a host of time-related routines, none of which I ever talk about in

the book. This stinks because getting the time or knowing the time or even

displaying the current time is often an important part of most programs.

I've gone by for too long!

TIMER.H

The time functions in C are defined in the TIMER.H header file for the

most part and — stand back! — they're UNIX time functions. Yech! You would

think that a programming language as nice as C would have it better, but

no. (Compiler and operating system specific time functions are available,

however.)

TIMER.H contains many functions, but the one I want to show you is time.

You might guess that time displays the current time. But no. Or that it

displays perhaps the date and time. But no. No! No! No!

The time function returns the number of seconds that have elapsed since

midnight, January 1, 1970. GMT. Uh-huh.

Further, the value is returned in a special time_t type of pointer, which

you must declare in your program:

time_t *timepointer;
 Even so, the number of seconds that have passed since you were 12 (or

maybe not even yet born!) is useless. I mean, can you imagine all the math

required to divvy that up into years, months, dates, hours and seconds?

Egads!

Fortunately, there is a companion TIME.H function called ctime, which

converts the time_t value into handy and veryprintable string. Time for a

program!

Name: TODAY.C

#include <stdio.h>

#include <time.h>

int main()

time_t now;

time(&now);

printf("It's now %s\n",ctime(&now));

return 0;

Shift+Click here to download a copy of the TODAY.C source code. This

program is almost utterly naked C, so it runs anywhere. I just re-compiled

it under gcc in Linux and it worked, so everyone should be happy here.

Compile. Link. Run!

It's now Sat Sep 02 17:05:15 2000

Here's what's going on:

The time_t now; statement creates the time_t pointer variable, into which

that huge number-of-seconds variable is stored. The variable is used by

the time function, time(&now) to create and store the current time — I

mean, number of seconds since Nixon was in the Whitehouse.

The killer is the ctime function inside the printf statement. That's what

converts the number of seconds into a string you can read on the display.

There. Nothing to it.

Well, unless you just want to display the time. Or maybe you just want to

display the date. If so, you have to look elsewhere for your time or date

functions. Alas.

Better DOS Functions

Now the rest of the programs in this lesson require use of the DOS.H

header file, which accesses special DOS routines to display the date and

time. If you're using Microsoft Visual C++ versions 4.0 or later, you

cannot compile and run these programs. Sorry.

These programs do compile under DJGPP as well as Borland C++, providing

you set the target as a 16-bit DOS project.

The DOS.H header defines two functions, getdate and gettime, which fill

special structures with values representing the current date or time.

Nifty.
 getdate requires you to create a date structure, into which it puts values

as follows:

struct date

int da_year; /* current year from 1980 to 2099 */

char da_day; /* day of the month, 1 through 31 */

char da_mon; /* month, 1 through 12 */

};

gettime had you set up a time structure into which it places values in

this manner:

struct time

unsigned char ti_min; /* minutes, 0 to 59 */

unsigned char ti_hour; /* hours, 0 to 23 */

unsigned char ti_hund; /* hunrdredths of seconds, 0 to 99 */

unsigned char ti_sec; /* seconds, 0 to 59 */

};

The following program, NOW.C, demonstrates how to put these functions

together:

Name: NOW.C
#include <stdio.h>

#include <dos.h>

int main()

struct date date;

struct time time;

getdate(&date);

gettime(&time);

printf("Today is %d/%d/%d, it's %d:%02d",

date.da_mon,

date.da_day,

date.da_year,

time.ti_hour,

time.ti_min);

return 0;

Type in the above program, or just relent and shift-click here to download

yourself a copy. Compile it. Run it.

Today is 9/2/2000, it's 17:06

What you see on your screen will, of course, reflect the current date and
 time (according to the computer, at least). A few things to point out:

struct date date;

struct time time;

These two statements create the date and time structures into which

getdate and gettime place the current date and time values. I used the

names date and time for the variables, which could be confusing to some,

but isn't to me!

The printf statement is pretty straightforward. Remember that it's the

backslash, \, that needs to be specified twice if used in a printf

formatting string. Also, see the %02d placeholder? That ensures that the

seconds value always displays with a leading zero. Otherwise a time of

12:5 looks odd, when you're used to seeing 12:05 instead.

I split the variables in the printf statement onto separate lines so you

can better see them. Those are merely the date and time structure values,

though I neglected to put in the seconds and hundredths of seconds values.

Now . . . room for improvement. You have a homework assignment!

I want you to modify the NOW.C program. I would like it to display the

time in a 12-hour format, from 12:00 a.m. on through 12:00 p.m. (noon),

and then starting with 1:00 p.m. for one o'clock in the afternoon on up to

11:00 at night. So all you're doing is applying some logic and programming

to get the display to read "right."

Please work on the modifications on your own. There is no right or wrong

way to do it, though there are better and worse ways! When you're done, or
 if you're stuck, you can click here to see my solution, which is only one

of many potential ways to do it. Good luck!

Bonus C For Dummies Lesson 13.3Lesson 13.3 – Say Hello to Mr. Bit

Remember this guy?

Well, I do. He was the Little Man with the Stupid Hat who taught me how

the decimal system worked.

The Little Man with the Stupid Hat (LMSH) had ten fingers – just like you

do, boys and girls! Easy enough. But the point behind LMSH was to get

everyone to understand the "10s position" and "100s position" and so on,

the way a big decimal number stacks up. Such as:

That's 3 hundreds, 7 tens and 9 ones, which translates into the number

379. Remember how that works? Of course you do. (And if you don't, then at

least you're nodding your head.) Well now I'd like to introduce you to

Binary Man:

Poor Binary Man has only one finger. So he can only count to one. Well,

one and zero. LMSH can use zero fingers to show a zero, and Binary Man can

do so too. Not only that, Binary Man can count to larger numbers just as

LMSH can: you just need more than one of him. Thus:
Because Binary Man has one finger, he counts by twos and not by tens. So

the first place is the 1's, but the second place is the 2's, then 4's,

8's, 16's, 32's and on up, each time double the amount of the previous

Binary Man.

Yes, this is weird.

And it's weird primarily because it's not the way we count. We count in

tens, probably because we have ten fingures. But computers have no

fingers. They have only themselves, so they count by twos. "Base two," is

what it's called. Also known as binary.

Some Binary Numbers

(But not so many as to bore you)

All numbers are really symbols. For example:

That's not ten at all. It's the symbol "1" and "0," which people using

such numbering systems refer to as "ten." There really isn't ten of

anything up there; just symbols. In fact, not all humans use "10" to mean

ten. The Chinese use the following symbol:

Again, that ain't ten of anything. So why not the following:

Your decimal-influenced mind will believe that to be the number 1,010 at

first (and even if it were, it's not one thousand ten of anything). The

number in binary, however, is the value 10. Here's Binary Man again:
So, just like you learned when you were young, you have 1 in the 8s place

and 1 in the 2s place. Add 8 and 2 and you get . . what? Anyone? Anyone .

..?

Of course, you get ten. That's how binary represents numbers. yes, it's

weird. Here is your boring example:

Above you see the value 86. You have:

1 x 64

0 x 32

1 x 16

0x8

1x4

1x2

0x1

That's 64 + 16 + 4 + 2. Add it up and you get 86.

Now isn't this a pain? Sure it is. But you shouldn't worry about it since

it's the computer that figures things in binary. With your programming

skills, you can display that value as decimal or even hexadecimal. So the

binary part is really pointless, though it helps to understand what's

going on. (More on this in a few paragraphs.)

Hexadecimal is actually a shortcut for binary. Most programmers keep a

table like this one handy, so they can easly convert between hex and

binary.
 Remember that binary is base 2. Computers are obsessed with base 2.

The printf function lacks a method for displaying binary values. (You

can display Hex and decimal just fine.) But don't panic! There's a

binary value display function at the end of this lesson.

Bit Twiddling

C has a few operators that let you manipulate integer values at the bit

level, what's known in programming circles as bit-twiddling. Here's the

mysterious list for you:

<< Shift bits left

>> Shift bits right

& Boolean AND operation

| Boolean OR operation

^ Boolean XOR operation

~ One's compliment (unary)

Okay. You've seen them. I'll save the details for the next lesson. But I

can't let you down here without giving you at least one program. The

following program contains the infamous binString function, which returns

a string representing a binary value. It actually uses the & and <<

operators shown above, so you'll have to wait to find out exactly how it

works.
Name: BINBIN.C

#include <stdio.h>

#include <stdlib.h>

char *binString(int value);

void main()

char value[8];

int v;

printf("Enter a value, 0 to 32000:");

gets(value);

v = atoi(value);

printf("Decimal value is %i\n",v);

printf("Hexadecimal value is %4X\n",v);

printf("Binary value is %s\n",binString(v));

char *binString(int value)

static char bin[17];

 int index;

for(index=0;index<16;index++)

if(value & 0x8000)

bin[index] = '1';

else

bin[index] = '0';

value = value << 1;

bin[16] = 0x00;

return(bin);

Shift+Click here to download yourself a copy of the BINBIN.C source code.

Shove it into your editor. Compile. Link. And . . . run!

Enter a value, 0 to 32000:

Type in your favorite number, such as 12345 (no commas). Press the Enter

key and voila:

Decimal value is 12345

Hexadecimal value is 3039

Binary value is 0011000000111001

The program displays the value you input three different ways: decimal,

hexadecimal and in binary. If you add up the binary digits, you'll get the

proper decimal value (which you can verify using Windows' Calculator

program in the Scientific mode).

A description of how BINBIN.C works, specifically the binString function,

will be offered in the next lesson. For now, if you like, feel free to use

binString in any program that requires a binary number string. The

function works with any integer value.

Try running the program a few more times with some additional numbers.

Try some negative numbers (-1 through -32000) to see what happens. (I'll

explain it in the next lesson).

Try entering some "holy" computer numbers: 255, 16384, 1024, 4096,

32767, etc.

The printf function automatically displays decimal values. This is okay

since you, as a human, expect that.

You can use the %X (or %x) placeholder in printf to display a

hexadecimal value. No sweat.

Binary values require their own function to display, which is what

binString does in the program. Even so, some comiplers may have a "bin"

placeholder character in their printf function.

Note how the binary and hexadecimal values relate. See how hex is a

convenient shortcut for binary?

Bonus C For Dummies Lesson 13.5Lesson 13.5 – Color Text

Windows may have it all, but DOS could always print in colored text. The

only PC that can’t print in colored text was the old monochrome system.

Since then, and since the color displays have gotten better and better,

printing DOS stuff in color has been a snap.

The bad news is that there are no native C language routines or functions

for writing colored text on the screen. (Well, they have ‘em in Borland

C++, which I’ll get into later.) The good news is that, thanks to the

versatility of the C language, you can write your own routines. This

shouldn’t be hard, providing you know your BIOS calls as described in

Volume I.

But there’s more good news and bad news!

The bad news is that writing colored text to the screen requires that you

understand the color codes used by the PC. Even so, the good news is that

it’s not that hard and there’s a table shown later in this Lesson that

describes everything you need.

Enough news!

Writing colored text

The putchar and printf functions in STDIO.H are actually shortcuts to the

secret, inner BIOS routines in the PC’s guts. They use Interrupt 0x10 (the

Video interrupt) function 0x0A, "Write attribute and character at cursor."

This function has the following parameters, which are sent to the BIOS in

various microprocessor registers:

Parameter Value Register

Function number 0x09 AH

Character code ? AL

Display page 0x00 BH

Text color ? BL

Character count ? CX

The Character code is the ASCII value of the character you’re displaying,

which is just like the character you’d specify in a putchar function call.

The Display page is usually zero. (The PC has several text display pages,

though I’ve rarely seen any program use anything other than page 0. (If

you wanted to, you could write a program to display text to different

display pages and then "flip" the pages to rapidly display information –

another waste of time!) Click here for the waste-of-time info.)

The text color tells the BIOS which foreground and background colors to

set for the text, whether the text is blinking and whether or not the

"high intensity" colors are used. More on that in a second.

Finally, the rarely used Character count value tells the BIOS how many

characters to write to the screen. Normally you write only one, but you

could write up to 65,000 characters providing you send the proper value to

the Video interrupt. It boggles the mind!

Always use display page zero. That’s the one your PC is probably using

right now (in the text mode, anyway).

 You must manually move the cursor after making this call. It does not

move the cursor for you.

To move the cursor, use the locate function introduced in Lesson 5-3.

You’ll also need to use another function to read the cursor so that

you’re sure you move it properly.

(It’s a bother, but you wanted to write in color!)

The color numbers are listed later in this Lesson.

And those color numbers should be unsigned char types. Anything else and

you’ll peeve the compiler.

Feeling blue?

When Mr. Norton introduced his Norton Utilities version 2, he added a tool

that changed the color of the DOS prompt and all text displayed at the DOS

prompt. I have no idea how he did that, but you might get a hint of what’s

going on by trying the following program.

Name: BLUE.C

#include <stdio.h>

#include <dos.h>

#define VIDEO 0x10

#define BLUE 0x1F

void dcolor(char ch,unsigned char color);

void main()

char *text = "Am I blue?";

unsigned char x;

while(*text)

dcolor(*text,BLUE);

text++;

putchar('\n');

void dcolor(char ch,unsigned char color)

union REGS regs;

int x,y;

/* First, read the cursor */

regs.h.ah = 0x03; //Read cursor

 regs.h.bh = 0x00; //"page"

int86(VIDEO,&regs,&regs);

y = regs.h.dl; //save Y pos.

x = regs.h.dh; //save X pos.

/* Now, write the color char */

regs.h.ah = 0x09; //Write color

regs.h.al = ch; //character

regs.h.bh = 0x00; //"page"

regs.h.bl = color; //color

regs.x.cx = 1; //count

int86(VIDEO,&regs,&regs);

/* Move the cursor forward one notch */

y++;

/* Reset the cursor's position (locate()) */

regs.h.ah=0x02; //Move cursor

regs.h.bh=0x00; //"page"

regs.h.dh=x; //row
 regs.h.dl=y; //column

int86(VIDEO,&regs,&regs);

Carefully type the above source code into your editor. Actually, since

this is on a Web page, you can copy and paste the source code into your

editor, or Shift+click on this link to download the source code directly.

Compile and run. You’ll see the following text displayed:

Am I blue?

The background color is blue and the text is bright white. I’ll explain

how that worked in the next section.

This program was created on the Borland C++ compiler.

Older versions of Microsoft C can compile this program just fine. In

some cases you may need to specify _REGS instead of REGS in the dcolor

function. (Just put an underline before REGS.)

The while(*text,BLUE) construction is explained in Lesson 10-7.

The cursor position must be read before the character is displayed. (You

could read it afterward, it doesn’t matter.) The row and column

positions are saved in variables y and x. Then the cursor’s position is

reset by incrementing the y (column) value. This is necessary because

function 0x09 does not move the cursor by itself.

How Color Text works on your PC

There is room for up to 2000 characters on the standard PC text screen.

That’s 25 rows of 80 columns of characters. Wow.

Knowing that a character is a byte, you may think that the screen’s

"memory" can hold 2000 bytes. But that’s not so. The screen is actually

4048 bytes big. That’s because each character on the screen has a

companion byte – an attribute byte that tells the monitor which color to

display the character, both background and foreground.

The character and attribute bytes go hand in hand. Normally you never mess

with the attribute bytes. You merely tell your program to display

characters and it does so, placing them on the screen one after the other

or in a specific spot. But Video interrupt function 0x09 allows you to

also set the attribute byte and thereby control the foreground and

background text color.

The PC produces color text using three colors: Red, Green and Blue. If

you’ve been around long enough, you may remember the old RGB monitors.

That’s Red, Green and Blue all over again; the three primary colors used

to create just about any other color you see on your screen.

In the attribute byte, there are three bits for the foreground color: one

bit for Red, another for Green and another for Blue. Here’s how that looks

in the handy eight-bits-in-a-byte graphic thing:

RGB

The byte above is shown with eight bits, 7 through 0 reading left to
right. Bit 0 is the Blue bit; bit 1 is the Green bit and bit 2 is the Red

bit.

Using binary math you haven’t yet been properly introduced to (which is

coming in one of those "missing Lesson 13 lessons"), you can substitute

values for R, G and B above:

0000-0001 is 1, or 0x01, for Blue colored text

0000-0010 is 2, or 0x02, for Green colored text

0000-0100 is 4, or 0x04, for Red colored text

The numbers shown above (0000-0001) are binary; not decimal. This is base

2 counting stuff; so 100 is not "one hundred," is the binary number 100,

which is four. (This all makes total sense after I get off my butt and

write Lesson 13-3.) Anyway, the values aren’t important.

Instead of quibbling about binary here, change the BLUE.C program. Change

line 5 to read:

#define BLUE 0x01

This sets the attribute to 0x01, which is the Blue bit for blue text.

Save to disk! Compile and run! You should see the text as blue with a

black background.

Now go back to the source code and change the same line to the values 0x02

and then 0x04. Compile and run after making each change. You’ll see Green

text and then Red text when you run each of the programs.

Musical interlude . . .

If you’ve been using a PC since the dark days, then you might recall that
 there are actually seven colors you can make text on the screen. In

addition to Red, Green and Blue, there is Cyan, Magenta, Brown and Gray.

These colors are achieved by combining the primary three. On a binary,

bit-by-bit level, it looks like this:

0000-0011 is 3, or 0x03, for Green+Blue colored text, Cyan

0000-0101 is 5, or 0x05, for Red+Blue colored text, Magenta

0000-0110 is 6, or 0x06, for Red+Green colored text, Brown

0000-0111 is 7, or 0x07, for Red+Green+Blue colored text, Gray

Finally, there is an eighth color available: Black:

0000-0000 is 0, or 0x00, for black

Black simply means that none of the colors are used. The PC really does

display the text, but since it’s black-on-black, you can’t see it. (A

black foreground is used primarily with a colored background.)

Rather than have you re-edit the BLUE.C source code to try all of the

color values shown above, just modify BLUE.C as follows:

First, remove the #define BLUE statement. Then edit the main function to

read as follows:

void main()

char *text = "Sample Text";

char *t;

unsigned char fc;

 for(fc=0x01;fc<0x08;fc++)

printf("%2X - ",fc);

t = text; //reset pointer

while(*t)

dcolor(*t,fc);

t++;

putchar('\n');

Save the changed file to disk as CLRTEXT.C and compile and run. You’ll see

something like this displayed:

1 – Sample text

2 – Sample text

3 – Sample text

4 – Sample text
5 – Sample text

6 – Sample text

7 – Sample text

Each of the "Sample text" lines appears in a different foreground color.

The number and hyphen part of the output (displayed by printf) is not

affected by your color changes. Why? Because printf uses a different

display routine. That text shows up as gray on white (like number 7).

The %2X part of the printf statement displays the color number as a

hexadecimal value on the screen. This comes in handy later when you’re

trying to pick a specific color on your screen.

The pointer needs to be reset to allow the while loop to scan the same

string of text for each iteration of the for loop. This is advanced

pointer stuff mulled over in Chapter 10 of Volume II.

Turn on your brights!

The fourth bit in the character attribute byte is for intensity. It’s not

a color. Instead, the intensity byte tells the display whether or not to

display the foreground color as bright or dim.

IRGB

The colors you’ve seen so far as dim, actually. To see the whole 16 course

meal in both dim and bright modes, modify the CLRTEXT.C source code so

that the for loop reads as follows:

for(fc=0x00;fc<0x10;fc++)
Basically, change the first value to 0x00 and then the 0x08 into an 0x10.

That tells the program to cycle through all 16 color combinations, from

everything off to everything on (bright white!).

Save the changes to disk. Compile and Run.

You’ll see the sixteen possible foreground colors displayed on your

screen.

Color code 0x00 is black – no colors at all.

Color code 0x0F is all white: 0000-1111 in binary. Looking at the above

binary table, you see that the Intensity, Red, Green and Blue bits are

all on. Everything’s on!

The for loop cycles color codes from 0x00 through 0x0F. The code 0x10,

specified in the for loop, isn’t used as a text color value since that’s

what tells the loop when to stop. (Review your for loop info if this

confuses you.)

Moving on to background colors

If things weren’t confusing enough, the PC also allows you to choose up to

8 different colors for the text background. Yup, these are the same 8

colors used for the foreground text, codes 0 through 7 (see above). There

is no "intensity" bit for the background color attribute, just Red, Green

and Blue bits. And they live in the attribute byte thusly:

RGBIRGB

So for, say, a red background you would switch on the Red background bit:

0100-0000, or 0x40 for a Red background

 Actually, the above attribute byte also sets a black foreground as well

(since none of the Red, Green or Blue foreground bits are set). So this is

where things get a bit tricky because you must set both foreground and

background attributes at once to get the text color you want.

But don’t panic! Fortunately, the nature of hexadecimal numbers is such

that you can easily specify foreground and background colors without

having to dig through a binary reference chart. Consider this:

0 = Black 4 = Red

1 = Blue 5 = Magenta

2 = Green 6 = Brown

3 = Cyan 7 = Gray

The hexadecimal digits above work for both foreground and background

colors. So if you want Blue text on a Red background, you would use: 0x41

with 4 as the Red background color and 1 as the Blue foreground color.

(The table at the end of this Lesson contains more detail, but the above

information should sate you for now.)

Time for another sample program!

Modify CLRTEXT.C so that the main function now looks like this:

void main()

char *text = "Hi!";

 char *t;

unsigned char fc,bc;

for(bc=0x00;bc<0x80;bc+=0x10)

for(fc=0x00;fc<0x10;fc++)

printf("%2X",fc+bc);

t = text; //reset pointer

while(*t)

dcolor(*t,fc+bc);

t++;

putchar('\n');

Essentially there is now a "nested loop" in the function. It loops through

all the foreground and background colors one at a time, displaying each
along with its hexadecimal values.

Save! Compile and Run!

You’ll see a grid displayed, detailing all the colors possible for the

PC’s screen.

Note the differences between the high-intensity and regular foreground

colors.

All the colors with the same foreground and background values (0x11,

0x22, 0x33 and so on) show up as solid on the screen. Better not use

those.

The foreground and background values are added in the program to get the

final text attribute (fc+bc in the program). This works since the text

color values don’t overlap each other in a byte – again, this is

advanced binary stuff you’ll have to read about in Chapter 13.

Eventually.

By the way, Chapter 13 uses text attributes in DOS to explain binary

math and logic – really weird ass stuff.

The PC’s text screen normally uses the 0x07 attribute for all

characters. That is, 0 for black background and 7 for a gray foreground.

Finally, the annoying blink bit

There are eight bits in a byte (on the PC at least), so there is one more

bit left to explain in the PC’s text attribute byte: the blinking blinking

attribute. By far the most annoying attribute available, which is why I

saved it for last:

BLRGBIRGB

When you set the blinking attribute on, the foreground text blinks. Very

annoying. To see this in action, change the source code for CRLTEXT. You

only need to change on line. Modify the call to the dcolor function so

that it looks like this:

dcolor(*t,fc+bc|0x80);

You’re adding the pipe character (Shift+\) and then 0x80 to the math

function that sets the color attribute. This is a logical binary operation

that sets the 7th bit in the byte without affecting the other bits. So

it’s fc plus bc OR 0x80.

Save the file to disk. Compile and Run!

You should see the same grid appear on the screen, but with all blinking

foreground text.

Now there is an off-chance you may see the grid displayed with bright

background text. For example, if you’re using a DOS window in Windows,

then the video driver may be set to prevent blinking text. If so, press

Alt+Enter to switch to the full screen mode. Then type the following DOS

command:

mode 80

This switches to the regular text screen. Now run the CLRTEXT.C program

again. The foreground text should blink in an annoying manner. (Press

Alt+Enter to return to Windows.)

And now, finally after all this work, you have the Massive Text Attribute

Table. Simply combine the foreground and background hex values to get the

type of colored text you want:

High bitsLow bits

CodeBackground colorCodeForeground color

00Black00Black

10Blue01Blue

20Green02Green

30Cyan03Cyan

40Red04Red

50Magenta05Magenta

60Brown06Brown

70Gray07Gray

80Black (blinking)08Dark gray

90Blue (blinking)09Bright Blue

A0Green (blinking)0ABright Green

B0Cyan (blinking)0BBright Cyan

C0Red (blinking)0CBright Red

D0Magenta (blinking)0DBright Magenta

E0Brown (blinking)0EYellow

F0White (blinking)0FBright White

The display color string routine

The dcolor function used throughout this lesson presents a handy way to
 display one text character in color on the screen. A nice companion

routine would be something that displays a whole line of text in a

specific color. That would be the dscolor function, which is shown below

in the final, final rendition of the CLRTEXT.C program, which you can

download by Shift+clinking this link.

Name: CLRTEXT.C

#include <stdio.h>

#include <dos.h>

#define VIDEO 0x10

void dscolor(char *string,unsigned char color);

void dcolor(char ch,unsigned char color);

void main()

char *text = "Hi!";

char *t;

unsigned char fc,bc;

for(bc=0x00;bc<0x80;bc+=0x10)
 {

for(fc=0x00;fc<0x10;fc++)

printf("%2X",fc+bc);

dscolor(text,fc+bc);

putchar('\n');

void dscolor(char *string,unsigned char color)

while(*string)

dcolor(*string,color);

string++;

void dcolor(char ch,unsigned char color)

union REGS regs;

int x,y;
/* First, read the cursor */

regs.h.ah = 0x03; //Read cursor

regs.h.bh = 0x00; //"page"

int86(VIDEO,&regs,&regs);

y = regs.h.dl; //save Y pos.

x = regs.h.dh; //save X pos.

/* Now, write the color char */

regs.h.ah = 0x09; //Write color

regs.h.al = ch; //character

regs.h.bh = 0x00; //"page"

regs.h.bl = color; //color

regs.x.cx = 1; //count

int86(VIDEO,&regs,&regs);

/* Move the cursor forward one notch */

y++;

/* Reset the cursor's position (locate()) */

 regs.h.ah=0x02; //Move cursor

regs.h.bh=0x00; //"page"

regs.h.dh=x; //row

regs.h.dl=y; //column

int86(VIDEO,&regs,&regs);

If you get into this color stuff, you may consider creating your own set

of #define routines to set the color values. For example:

#define FG_Black 0x00

#define FG_Blue 0x01

#define FG_Green 0x02

#define FG_Cyan 0x03

#define FG_Red 0x04

#define FG_Magenta 0x05

#define FG_Brown 0x06

#define FG_Gray 0x07

#define BG_Black 0x00

#define BG_Blue 0x10

#define BG_Green 0x20

#define BG_Cyan 0x30

#define BG_Red 0x40

#define BG_Magenta 0x50

#define BG_Brown 0x60

#define BG_Gray 0x70

#define Bright 0x08

#define Blinking 0x80

Here is how the defines work: To create a certain color combination,

specify the proper keyword above and separate it by the | (pipe), which is

the logical OR. So to have bright white text on a blue background, you

would use the following:

dscolor("Bright White on Blue",FG_Gray|Bright|BG Blue);

That’s FG_Gray (foreground text color gray) OR Bright (for high intensity)

OR BG_Blue (background text color blue). Again, there will be more of this

binary logic junk in Chapter 13 – when I get around to writing it . . .

The Borland Solution

The Borland C compiler has a whole slew of color text commands, plus

versions of common C routines that display text in color. This is a

veritable treasure of fun things, which is never really highlighted

anywhere since very few people bother with DOS programs any more.

Two different commands are used to set the text and background color. Text

color is set with the textcolor function:

textcolor(color)
 And background color is set with the textbackground function:

textbackground(color)

Each command requires an integer value, equal to the values described

earlier in this Lesson for various colors to display on the screen. Or you

can use the color macros predefined in the CONIO.H header file:

BLACK DARKGRAY

BLUE LIGHTBLUE

GREEN LIGHTGREEN

CYAN LIGHTCYAN

RED LIGHTRED

MAGENTA LIGHTMAGENTA

BROWN YELLOW

LIGHTGRAY WHITE

The colors you set are then used by any subsequent calls to the cputs or

cprintf functions. Both of those functions work just like the regular C

counterparts, puts and printf, though the output is in the color specified

by the most recent textcolor and textbackground functions.

You must specify the CONIO.H header file in your source code for these

functions to work.

Bonus Information on Display Pages

Bonus C For Dummies Lesson 13.6Lesson 13.6 – Introduction to Recursion

It's madness! Madness!

 Recursion is one of the most frustrating topics in C. No one gets, at

least not right away. And if you do get it, then you'll never understand

it – at least until you need it.

I wasn't originally going to discuss recursion anywhere. I feel it's

beyond the scope of the books and this Web page. But I've gotten some

questions and I feel it's a neat trick. So I thought I'd toss it up here

in the Missing Chapter since recursion is a subject often missing from

many C language tomes.

Brace yourself

I am Not An Endless Loop

I am Not An Endless Loop

I am Not An Endless Loop

...

Recursion is the art of calling a function within that same function. Now

the guy who thought this up had to be totally nuts. Or a genius. Because,

honestly, why would you call a function from within a function? Let me

drag you through the obvious here.

Name: RECURSE.C

#include <stdio.h>

void repeat(void)

{
 puts("Recursion can drive you mad!");

repeat();

void main()

repeat();

The above program, RECURSE.C, is stupid. I admit it. But it shows you

recursion on the very basic level (albeit stupid recursion). The function

repeat is called from within itself. Anyone who's reading this and who's

been through Volume I of my C book will immediate know what's going to

happen when this program is compiled and run.

Download the source code for RECURSE.C by Shift+clicking here, or you can

type the thing in yourself. By the way, notice that I wrote it "upside

down" so I didn't have to prototype the repeat function.

Compile. Yes it will compile. Remember that C compiles stupid things all

the time.

Run.

Recursion can drive you mad!

Recursion can drive you mad!

Recursion can drive you mad!

Recursion can drive you mad!

Recursion can drive you mad!

Press Ctrl+C to stop it. (And please so stop it before you computer

crashes, which it will do if you let the program run too long.)

The program calls the repeat function, which displays the string

Recursion can drive you mad!.

But the repeat function then calls itself again, which displays the same

string.

Over and over.

The reason your computer crashes if the program runs too long is that

you run out of stack space, the infamous "Stack Overflow" error message

that drives DOS users nuts. This is explained in the next section.

Recursing Yourself to Death

Now, obviously, recursion is a useful tool. The program RECURSE.C does not

show that. It merely illustrates, on a stupid level, how recursion works.

But the program doesn't work because there is no exit from the repeat

function. The program is really an endless loop since repeat offers no

chance to return to the main function. A program that really uses

recursion would have that return.

So why does it crash? Because eventually the microprocessor runs out of

stack storage space. Calling the function over and over without ever

returning builds the stack to massive proportions, which eventually

crashes the computer.

The stack itself is a storage place in memory, typically at the top of

free memory. A special microprocessor register, the Stack Pointer, holds

the stack's address. Values are pushed onto the stack and can then be

popped off the stack.

Every time you call a function in C, the stack pointer holds the return

address. For example.

In the figure above, the intro function is being called. The memory

address where that function was called, 0x820 in the figure, is saved

(pushed) onto the stack for storage. The stack pointer then moves down to

the next available location on the stack:

So SP, the Stack Pointer in the microprocessor, now points to another

address while the intro function runs.

After intro is done, control returns to the main function. The

microprocessor finds that exactly location by popping the address stored

on the stack pointer:

What happens with the RECURSE.C program is that the return address keeps

getting pushed onto the stack. The stack keeps moving to a lower address

in memory until the stack eventually destroys other data. The computer

crashes.

To see how this works you can use the _SP macro in Turbo C++/Borland C++
 (sorry Microsoft users!). Modify the RECURSE.C program by sticking the

following statement into the repeat function:

printf("%04X - ",_SP);

The final program should look like this:

Name: RECURSE.C

#include <stdio.h>

void repeat(void)

printf("%04X - ",_SP);

puts("Recursion can drive you mad!");

repeat();

void main()

repeat();

Compile and run. This time you'll see the stack pointer's address

displayed in the output. Something like this:

FBAA - Recursion can drive you mad!

FBA6 - Recursion can drive you mad!

FBA2 - Recursion can drive you mad!

FB9E - Recursion can drive you mad!

FB9A - Recursion can drive you mad!

FB96 - Recursion can drive you mad!

FB92 - Recursion can drive you mad!

FB8E - Recursion can ^C

Press Ctrl+C to stop so you can get a better view. That four-digit hex

number gets smaller and smaller as the stack keeps pushing down into low

memory. Eventually, as it reaches zero, the computer will crash.

To avoid crashing, your recursion routine must have an exit. Otherwise,

it's just not recursion (and it messes up the stack).

A Way Out

When I first read about recursion I thought, "There blows the stack!" As

an old-line assembly language programmer, I knew that calling a routine

from within itself ran the risk of pushing the stack out the back of the

computer. Fortunately, my friend Wally Wang showed me how to write a

recursive routine that eventually bails itself out.

The following program is X5.C, which contains a modification of the repeat

function in the RECURSE.C program. This time, the repeat function has a

way out, but notice how recursion is still used to call the same routine

from within itself.

Name: X5.C

#include <stdio.h>

void repeat(int x)

if(x)

printf("%04X - ",_SP);

puts("Recursion can drive you mad!");

x--;

repeat(x);

void main()

repeat(5);

Download this source code by Shift+clicking here, or you can just modify

the existing source for RECURSE.C in your editor. Remember to save the

file as X5.C.

Note how the repeat function has been re-typed. It now requires an integer
as input. Also, an if test is used, which is the key to making the

recursion work.

Microsoft Compiler People: Do not include the printf statement that

displays the stack pointer (_SP) value!

Compile and run.

No need to press Ctrl+C to stop it this time. You should see something

similar to this as output:

FFEE - Recursion can drive you mad!

FFE6 - Recursion can drive you mad!

FFDE - Recursion can drive you mad!

FFD6 - Recursion can drive you mad!

FFCE - Recursion can drive you mad!

The repeat function worked through only five times, which is the value

originally passed to the function. Recursion works! Here's how:

First, the repeat function requires an integer value. That value tells the

function whether or not to repeat itself:

if(x)

Any positive value for x causes the loop to repeat. If x is zero, then the

repeat function doesn't do anything but return. (Ah-ha! It can return!)

If x is greater than zero, then the if statements run:

printf("%04X - ",_SP);

puts("Recursion can drive you mad!");

x--;

repeat(x);

The stack pointer address is printed, along with the catchy phrase. But

then x is decremented. The repeat function is called again, but this time

with a new, smaller value for x. Then the whole thing happens all over.

Eventually, X will equal zero. When it does, the if statements don't

execute and the repeat function merely returns. At that point, the return

addresses start popping off the stack: Pop! Pop! Pop! Pop! The repeat

function just keeps returning to itself until eventually the big knot

totally unwinds and control returns to the original calling function,

main.

Amen!

Now you can sit and stare and try to figure it out.

Borland Compiler People: Before you ruminate, you can add one more line of

code to the program to visually inspect the popping off of the return

addresses from the stack. Modify the if statements to read as follows:

printf("%04X - ",_SP);

puts("Recursion can drive you mad!");

x--;

repeat(x);

printf("%04X popped off!\n",_SP);

The last statement is executed as the repeat function returns (when the
recursion starts unwinding). Now, when x equals zero, and the repeat

function returns, you'll see the return addresses as they're popped off

the stack.

FFEE - Recursion can drive you mad!

FFE6 - Recursion can drive you mad!

FFDE - Recursion can drive you mad!

FFD6 - Recursion can drive you mad!

FFCE - Recursion can drive you mad!

FFCE popped off!

FFD6 popped off!

FFDE popped off!

FFE6 popped off!

FFEE popped off!

Cute.

Something Remotely Useful

There was a great article on recursion in the recent Dr. Dobb's Journal,

though most of it was above my head. But that, and a flurry of recent

letters from readers, got me on the subject. I'll show you a really nifty

example of recursion in the next lesson, but before then I thought I'd

pass the following gem on to you.

I've been messing a lot recently with strings and arrays and needed some

way to count the number of characters in a string. There is an strlen

 function, which is fine, but like most people who teach C programming I

just wouldn't settle for that. No, I wanted to count the number of

characters in a string without using strlen or subtraction or even a while

loop. No, no, no. I wanted recursion.

Name: CCC.C

#include <stdio.h>

int string_count(char *s)

int count = 1;

if(*s++)

count += string_count(s);

else

count--;

return(count);

void main()

char *string="A sailor went to cc, see?";

 int x = 0;

x = string_count(string);

printf("The string \"%s\"\nis %i characters long\n",string,x);

Type this monster into your editor, or Shift+click here to download a

copy.

Compile and run.

The string "A sailor went to cc, see?"

is 25 characters long

Now this one will probably baffle you. You can see how the count variable

is incremented (actually added to itself with the += operator), but

doesn't count get reset to 1 every time the string_count function is

called?

Yes, it does. But who cares! The string_count function does not count the

length of the string each time it's called. No, it counts the number of

times the function returns.

You see, string_count returns an integer value that is added to the count

variable, but that doesn't happen until the function returns.

For example, for every character in the string, the function is called and

only the following chunk is executed:

if(*s++)

count += string_count(s);
 The if statement checks to see if the NULL (end of string) character has

been encountered. If not, then the pointer s is incremented and

string_count is called again. Nothing happens to the count variable

because the string_count function hasn't yet returned from itself. (Think

about that one for a while.)

When the NULL character is encountered, then string_count starts to

return. Only then is the count variable added to itself. The value of

count grows by one every time string_count returns.

The else part of the if test decrements the value of count, which happens

only once to account for the NULL at the end of the string. (Otherwise the

size returned would always be one character too long.)

So, to wrap it all up, the string_count function uses recursion to call

itself once for every character in the string. Then, it counts the number

of returns, incrementing the value each time to get the size of the

string.

Isn't this utterly insane?

Copyright © 1998-1999 by Not Another Writer, Inc.

All rights reserved

Bonus C For Dummies Lesson 15-1Lesson 15-1 — What Files Lurk on Disk

Programmers do not live by simple file access alone. No, there must be

directory access as well!

I get these questions all the time: So how can I find a file on disk in C?

How can I change directories? How can I find out which disk or directory a

file is using? They can all be answered when you unwrap the mystery of

directory access in C.

Of course, there really is not directory access in C. It's all a myth! The

actual routines used to access directory information are found in

customized libraries for whichever operating system and compiler you're

using. (Can you tell where this is leading?) For DOS, the library is

DIR.H, which means these programs compile well in Borland C++ and DJGPP,

but not at all in Microsoft Visual C++ 4.2 or later, or in Linux. Sorry.

Finding Files on Disk

There are two companion functions used to find files on disk, both of

which are defined in the DIR.H header: findfirst and findnext.

The findfirst function is used first. It sets DOS up to look for files

that match a specific pattern, which would be a matching scheme using

wildcards or a specific filename itself. Esentially, the pattern is the

option you would use with the DIR command:

DIR *.*

DIR *.C

DIR HELLO.EXE

*.*, *.C and HELLO.EXE are all patterns findfirst would use.

Next findfirst uses a special file block structure, also defined in the

DIR.H header. This structure is filled with information about the files
 found: the file's name, size in bytes, date and time, attributes and other

stuff they don't tell you about.

Finally, findfirst is passed the attributes of files to find. That way you

can search for archive, read-only or even directory files. Searching for

directory files, by the way, is the method by which you can scour the

entire hard drive for files. (More on that in a later lesson.)

The following program shows you how findfirst can be set up to locate the

first file in the current directory.

Name: D.C

#include <stdio.h>

#include <dir.h>

/* File attribute definitions */

#define NORMAL 0x00

#define READONLY 0x01

#define HIDDEN 0x02

#define SYSTEM 0x04

#define VOLLABEL 0x08

#define DIR 0x10

#define ARCHIVE 0x20

int main()

struct ffblk fblock; /* Create file info block */

if( findfirst("*.*",&fblock,NORMAL) != 0 )

puts("Oops! Some kind of error!");

return 1;

printf("I have found the file %s\n",fblock.ff_name);

printf("It is %d bytes in size.\n",fblock.ff_fsize);

return 0;

Hunt and peck this source code, or just click here to download. Compile

it. Run.

You might see something like this as the output:

I have found the file D.EXE

It is 31792 bytes in size.

The findfirst function ventures out to disk and locates the first file it

finds, which above was D.EXE (the program file itself). That file rang in

at 31,692 bytes on my computer (compiled with BCC 5.0).

 Here's what went on:

The first struct statement defines the file block structure used by the

findfirst (and findnext) functions. The ffblk structure is defined in

DIR.H. Here's what it looks like:

struct ffblk

char ff_reserved[22]; /* Internal use and secret stuff */

char ff_attrib; /* 8-bit file attribute */

unsigned ff_ftime; /* Modification time (encoded) */

unsigned ff_fdate; /* Modification date */

long ff_fsize; /* file size in bytes */

char ff_fname[13] /* filename string */

This is all standard information, all of which is stored in an MS-DOS

directory entry. And this is old DOS stuff too, since the creation

date/time, access date/time as well as the long filename information is

not present. But so much for that.

The structure is declared in D.C using struct ffblk fblock. Then the

findfirst statement can do its magic:

findfirst("*.*",&fblock,NORMAL);

The first argument is the pattern to match. In this case, it's *.*

(star-dot-star), to match all files in the directory. The second argument

is the address of the file block, where information about the found file
 is placed by findfirst. Then comes the attribute to match, which is

defined as NORMAL in the program.

The whole findfirst function is placed inside an if statement to test for

an error. If the function returns a non-zero value, then an error has

occurred and the program bails out. Otherwise, the information stored in

the structure is displayed using two printf statements.

Finding the Rest of the Files

You find one, you can find them all! After the initial findfirst function,

subsequent calls to the findnext function (using the same fileblock

structure), yields you all the files in the directory. Well, eventually.

And one at a time, as the following program demonstrates:

Name: DR.C

#include <stdio.h>

#include <dir.h>

/* File attribute definitions */

#define NORMAL 0x00

#define READONLY 0x01

#define HIDDEN 0x02

#define SYSTEM 0x04

#define VOLLABEL 0x08

#define DIR 0x10

#define ARCHIVE 0x20

int main()

struct ffblk fblock; /* Create file info block */

if( findfirst("*.*",&fblock,NORMAL) != 0 )

puts("Oops! Some kind of error!");

return 1;

printf("%s\t%d\n",fblock.ff_name,fblock.ff_fsize);

while( findnext(&fblock) == 0)

printf("%s\t%d\n",fblock.ff_name,fblock.ff_fsize);

return 0;

Enter this program into your editor, or shift-click here to download it.

You can also create the program by simply modifying D.C as shown above.

Essentially there is just one extra line and the modified printf statement

is used twice.
 Compile and run.

My output looked something like this:

D.EXE 31792

D.C 605

D.OBJ 732

DR.C 643

DR.EXE 31770

DR.OBJ 737

You see filenames and file sizes. Interesting. A few more moves and you

could recreate the entire DIR command!

Nothing much changes between DR.C and D.C. The printf statement is

modified to display only the filename and size. And it's used twice.

The second time printf is used is inside a while loop. The loop uses the

findnext function as its test:

findnext(&fblock)

Using the findnext function with the same file block used by a previous

findfirst function yields the next matching file from the directory.

That's it! The file block structure is refilled with information about the

new file. And as long as the value returned is not zero, the while loop

keeps searching for new files in the directory.

The while loop ends when the last file from the directory has been read.
 You can change the wildcard specification to locate specific files, but

you cannot (usually) specify a full pathname. In C you need to use the

chdir function to change to another directory and search there. (More on

that in a later lesson.)

If you need to build a database of files in the directory, then you'll

need to create a linked list using structures similar to the file block,

but with the linked list pointer added. (See Chapter 17.)

Changing the file attribute locates only files of that specific type.

For example, before Windows came along I wrote a DOS utility to locate

hidden files in a directory, which was basically the same routine shown

above but with the HIDDEN attribute specified.

Bonus C For Dummies Lesson 16-1Lesson 16.1 – Welcome to the Lower Level

The C language is considered a mid-level programming language. Mid-level

because it has some parts that are considered high-level — the

English-like words used for file access for example — and it has the

grunts and squawks of low-level languages, which includes its abilities to

twiddle bits and tweak memory with pointers and all that. But C also has

another leg up because most C compilers allow you to embed the one of the

lowest form of computer languages possible right there into your source

code. It's called in-line assembly.

Both Borland C++ and Microsoft C++ have the built-in assembly language
 feature. Even so, your compiler may implement it differently from the

way documented here. Check your Help file for the details.

A keen advantage of in-line assembly is that it allows all compilers to

access the computer's BIOS and DOS functions.

If you're using DJGPP, then you'll need to use the ATT nottaion, not the

Intel Assembly notation shown here. Click here for more information.

(Sorry!)

Need I mention that this information here is specific to a PC. You can

use in-line assembly on non-PC computers, but the results will be

different.

A Program Before You Get Too Bored

The following is a C program in name only. It's actually an Assembly

language program placed inside a C program. Normally the whole dang doodle

program wouldn't be assembly, only a small part would. But this is one of

those demos, you know.

Name: HELLOW.C

void main()

char *string = "Hello, Weirdo!$";

_asm mov dx,string

_asm mov ah,9

 _asm int 21h

Type the above program into your editor, or Shift+click on this link to

download it.

Compile and run! You'll see:

Hello, Weirdo!

Honestly, there is nothing there to indicate that any string would

display. What you did was to use in-line assembly language — a second

programming language embedded in your source code — to cause the computer

to do something. In this case, you used a DOS function called an

"interrupt" to display text on the screen.

All the lines starting with _asm are assembly language directives.

Some versions of Microsoft C++ may require you to use two underlines

before asm.

Borland C++ may require the presence of the TASM assembler to compile

this program.

More detail on how the program works

The Miracle of Assembly Language

The microprocessor speaks only one language. It's called machine language.

Like the name says, it's a language spoken primarily by machines, not by

soft hairy beings like you or I.

Machine language consists of codes, often one- or two-byte instructions

that tell the microprocessor to do something: access memory, compare

values, jump to another piece of code, and so on. It's all basic (very

basic) programming.

As an example of machine language, consider this:

B4 09

The B4 is an 80x86 instruction to place a value into the AH microprocessor

register — a storage place in the microprocessor for values. The value

placed into the AH register is 9. From this you can deduct the following:

You would have to be nuts to write a program in machine language!

The truth is, no one writes programs in machine language. Instead, they

use a programming language called Assembly. The Assembly language is

merely a shortcut way to program a microprocessor in machine language.

Though, instead of hexadecimal values, short mnemonics are used. Instead

of B4 09 you would write:

mov ah,09h

The "mov ah" is assembly code for "put the following value into the

microprocessor's AH register." And the 09h is assembly notation for the

value 9 hexadecimal, what you would write a 0x09 in C. This is much easier

than remembering B409. But to the computer, it's the same deal.

So the question begs, why bother?

Machine language is fast. Very fast. If Microsoft re-wrote all of Excel

directly in machine language instead of C, the program would smoke.

Unfortunately, it takes a lot longer to write programs in assembly than it

does C. In C you have functions already defined for you. In assembly, you
 have to create them all using the grunts and snorts of machine language.

That takes a l-o-n-g time.

The solution is then to use C where C is most efficient: functions and

other high-level language stuff. Then use assembly language for the stuff

you need to run veryfast. Hence, in-line Assembly was born.

For this Lesson and the rest of this Chapter, I'll use in-line assembly

merely to show you some tricks. In real life you'll probably use it only

when you have some clumsy chunk of code you want to run veryfast.

More good news: Since in-line assembly works the same in both Borland

and Microsoft C, I can show you some common routines between the two. No

more "Borland-only" stuff.

Future lessons explain in detail about the microprocessor registers and

how they work.

Most of this stuff is basic 8086 Assembly language, which still works on

today's monster 32-bit Pentium systems. In fact, I would say very little

assembly coding is done in the full 32-bit mode. (Mostly because your C

compiler creates the 32-bit code for you much more efficiently.)

An Example From the Past

The following jewel comes from Volume I of C for Dummies, back when this

whole BIOS and DOS call stuff was alien. Hopefully this will make it even

more alien!

Name: DOSVER2.C
#include <stdio.h>

void main()

char major,minor;

_asm mov ah,30h

_asm int 21h

_asm mov major,al

_asm mov minor,ah

printf("This is DOS version %i, release %i\n",major,minor);

Type the above source code into your editor, or Shift+click this link to

download the source code. This is a subtle modification of the original

DOSVER.C program (which you can download if you Shift+Click here). Gone

are the DOS.H header file, REGS declarations and the int86() function.

Those aren't needed since you're accessing the microprocessor directly

through in-line Assembly language.

Compile and Run.

The results you see on the screen may vary, depending on which version of

DOS or Windows you're using. For most versions of Windows 95 you'll see

either one of the following:

 This is DOS version 7, release 0

Or, for later versions of Windows 95:

This is DOS version 7, release 10

This is all done using Assembly language right inside the C source code.

And you'll notice how easily the variables are echanged between the two

formats. (Which is something I'll get into later.)

The DOS.H header file is no longer needed in DOSVER2.C because you're

using Assembly language and not the int86() function.

The major and minor variables are declared as chars here instead of

ints. The reason is that the AL and AH registers are byte-sized

(8-bits). If the variables were integers, then the 8-bit values would

not be stored properly.

This program may crash with some later versions of Microsoft C++ on

Windows 95. There is a way to compile the program, but you must convince

the Developer Studio how to create DOS programs. (If you know how to do

this, please email me the solution. Thanks!)

Obligatory blow-by-blow of DOSVER2.C.

The next lesson covers the microprocessor itself, its registers and how

they relate to the REGS & REGS stuff in C.

Bonus C For Dummies Lesson 17-1Lesson 17-1 – The Birth of Linked Lists

Structures are just about the best way to store database-like

information. Well, short of buying a database (which is what I would

recommend to a beginner). You could probably prepare a heck of a little

database using the simple structure programs demonstrated throughout

Chapter 11 in Volume II. Unfortunately, there are some dreadful

limitations.

For example, the program BDAYS.C attempted to create a database of

birthdays. Declaring an array of structures did this:

struct family myfamily[10];

At first glance that seems perfectly fine. In fact, if you set out to

build a database program, you could just expand upon the rudiments of the

BDAYS.C program: just change the size of the array to match the maximum

possible input you could imagine. No one would ever suspect a thing! But

you would be wasting memory. And you could potentially run out of memory,

for example, when the Osmond family started using your birthday database

program.

The solution (or, more properly, the elegant solution) is to employ a

technique known as a linked list. That's the subject of this Lesson.

A linked list is actually considered an advanced programming concept,

which is why it's not covered in the book. (Well, that and I was running

out of space.) It's advanced because it deals with pointers in a way

most people wouldn't begin to guess at.

Just like most advanced concepts, someone smarter than you and I has

already worked out the details for doing linked lists. You merely copy
 the way that superbrain at MIT developed linked lists in the '70s, just

like everyone else does. This isn't called cheating. No, it's "using an

algorithm."

There is no requirement that you make a linked list when working with

structures. Arrays are just fine most of the time. In fact, linked lists

drive many programmers bonkers.

Feel free to skip this stuff if you like.

Before moving on, you must understand structures through Lesson 11-5 in

the book. Further, a good grasp of pointers and the malloc() function is

necessary here. You've been warned!

The Whole Big Fraggin' Idea Behind Linked Lists

Know these tidbits and you'll begin to understand linked lists:

A linked list is a bunch of structures containing pointers.

The pointers are of the structure type -- a pointer to the same

structure.

The pointer contains the address of the next structure in the list.

It all works like following clues to get to a hidden treasure; one map

takes you to another map, which takes you to another map and so on. In a

linked list, each structure contains an extra item: a pointer (of the

structure's type) which contains the address of the next structure in the

list.

The pointer is the link.

The group of structures is the list.

 The following figure attempts to illustrate this.

Each structure contains the address (pointer) of the next structure in the

list. The first structure contains "George Washington" and then the

address of the next structure in the list. This keeps going down to the

last item in the list, which contains 000 or the NULL pointer.

A Program to Make You Feel Better

The problem with other C books that try to teach linked lists is that they

give you the whole friggin' linked list program at once. I'll be

different. Below is the silly source code for PREZEZ1.C, which starts to

create a linked list-like structure as shown in the above illustration.

Name: PREZEZ1.C

#include <stdio.h>

#include <stdlib.h>

#include <malloc.h>

#include <string.h>

void main()

struct pres {

char name[25];

struct pres *next;

 };

struct pres *president;

/* Create the first item in the list */

president = (struct pres *)malloc(sizeof(struct pres));

strcpy(president->name,"George Washington");

president->next = (struct pres *)malloc(sizeof(struct pres));

/* Stop here and display the results */

printf("The first structure has been created:\n");

printf("president->name = %s\n",president->name);

printf("next structure address = %i\n",president->next);

Type in the above source code, or Shift+click on this link to download a

copy.

Compile and Run.

Here is what the output should look like:

The first structure has been created:

president-> = George Washington

next structure address = 3672

The structure was created and George Washington was stored in its string

variable. Then the address of the next structure for which malloc() has

allocated space is also included (though normally that value wouldn't be

displayed). This sample holds the rudiments for all linked lists, but it's

only showing you part of the Big Picture.

Refer back to Chapter 11 to discover what the -> symbol does.

Of course, the value you see on your screen may or may not be 3672 for

the address of the next structure.

This statement may toss you for a loop:

president->next = (struct pres *)malloc(sizeof(struct pres));

This should make sense to you: the next variable requires a pointer

value -- an address. To get that address, you simply use the malloc()

function as shown above. Malloc returns the address of a chunk of memory

equal to the size of the pres structure. (This was all explained in

Chapter 11.)

Adding the Next Structure to the List (First Attempt)

The PREZEZ1.C program makes room for another structure, but it doesn't do

anything with that room. This doesn't displease the C gods any; it's just

that the program takes only half a step and never completes the process,

which is to fill in the new structure created. This is really cinchy to

do, but requires a bit of modification to the program.

First, a new structure variable should be created:

struct pres *new;

Second, that variable should be used to hold the value returned from the

malloc() function. So that line should read:

new = (struct pres *)malloc(sizeof(struct pres));

The value returned from the malloc() function -- the address of the new

structure in memory -- is saved in the new pointer variable. That value

also needs to be stored in the first structure, so the following line is

needed:

president->next = new;

Now the program still runs the same, but the new variable holds a value

"that is never used" blah-blah-blah. So code needs to be added to fill in

the new structure. Something like:

strcpy(new->name,"John Adams");

That fills in the name part of the new structure, but to fill in the

pointer part you'll need something like this:

new = (struct pres *)malloc(sizeof(struct pres));

Oops! That changes the value of the new variable. Which means you can't

use the following code:

new->next = new;

Ugh! Looks like another structure pointer variable is needed. You must

keep the value of the current structure's address but also have a variable

to store the next structure's address. Will the madness end?

 The president variable holds the address of the first structure in the

list

The new variable holds the value of the second structure in the list.

But . . .

You cannot use the new variable to hold the address of the third

structure in the list unless you first save that address in yet another

pointer variable.

Yes, there is a solution.

No, the solution does not involve an array of pointer variables, though

you should pat yourself on the back for thinking of it.

Adding the Next Structure to the List (Second Attempt)

What you need to work all the structures and their addresses are only

three pointer variables:

One pointer variable holds the address of the first item in the list,

*first

One pointer variable holds the address of the current item in the list,

*current

One pointer variable is used to hold the address of the next item in the

list, *new

Of course there are problems with this, but don't rush ahead of yourself.

You still need to fix up the PREZEZ.C program so that you can add another

president's name to the list.

Name: PREZEZ2.C

#include <stdio.h>

#include <stdlib.h>

#include <malloc.h>

#include <string.h>

void main()

struct pres {

char name[25];

struct pres *next;

};

struct pres *first;

struct pres *current;

struct pres *new;

/* Create the first item in the list */

first = (struct pres *)malloc(sizeof(struct pres));

strcpy(first->name,"George Washington");

new = (struct pres *)malloc(sizeof(struct pres));

 first->next = new;

/* Fill in the next item: */

current = new;

strcpy(current->name,"John Adams");

new = (struct pres *)malloc(sizeof(struct pres));

current->next = new;

/* Stop here and display the results */

printf("The first structure has been created:\n");

printf("first->name = %s\n",first->name);

printf("next structure address = %i\n",first->next);

printf("The second structure has been created:\n");

printf("current->name = %s\n",current->name);

printf("next structure address = %i\n",current->next);

Type the above source code into your editor, or Shift+click here to

download the PREZEZ2.C file. Some things have changed: The old president
 variable has been replaced by the first variable, and the new and current

pointers have been added.

Compile and Run!

Here's a sample of the output you might see:

The first structure has been created:

first->name = George Washington

next structure address = 3768

The second structure has been created:

current->name = John Adams

next structure address = 3800

Of course, this program is still limited in its scope. However, it

provides a good foundation for linked lists.

There are some limitations: For example, a loop would be better suited to

add new items to the list. Also, there must be some way to tell when

you've reached the end of the list. These issues will be addressed in the

next lesson.

Bonus C For Dummies Lesson 17-2Lesson 17-2 – The Adolescence of Linked Lists

I think I could rattle off about a dozen reasons why so many people find

linked lists confusing. Above all, it has to be that most programming

books present you with a Linked List Program that's all complete. There's

nothing to really learn since you don't ever discover the painful process
-- the journey. You're merely presented with the reward and implored to

understand it. This is why linked lists come in second only to pointers on

the Eyeballs Bugging Out of Your Head meter.

Meanwhile, back at the White House . . .

From Lesson 17.1 you've seen a vague attempt at creating a linked list

using the PREZEZ.C series of programs. Generally, it's rather cinchy to go

ahead and create a structure with pointers (this was covered in Lesson

11.5):

Define the structure.

Define pointer-structure variables.

Use malloc() to create space for the pointers.

Fill the space for the new structure using the -> thing.

In fact, aside from using malloc() and the -> thing, structure-pointers

are no different than structures themselves. You just create them on the

fly, as you need them.

With a linked list, you not only create the structures, but you store the

address of each structure in the list; the current structure holds the

address of the next structure in the list. That way you can find things;

the list is "linked" together by the address in memory of the next

structure. Nothing gets lost!

So far, the PREZEZ2.C program has it almost right. The bottom line is that

you don't need a whole basket full of structure pointer variables to make

a linked list. You need only three: first, current and new. Working those
into a loop is the problem. (And then there's the problem of starting and

stopping the linked list, but don't rush yourself!)

First Detour: Filling the linked list

A linked list has to be a database of sorts. Somewhere low in memory in

your PC is a linked list describing all the disk drives in your system.

The operating system creates it when it boots: the BIOS tells the

operating system how many drives are physically present and the operating

system creates a linked list describing each drive. If any other drives

are added during the boot process (CD-ROMs, removable drives, network

drives), they're added to the linked list. When the thing is done, the

operating system gives each drive a letter depending on its position in

the list. Oh, but that's all trivial.

I've set up the PREZEZ.C series of programs to use static data -- strings

in double quotes -- instead of input from the keyboard:

strcpy(first->name,"George Washington");

and

strcpy(current->name,"John Adams");

This is necessary partly because of the database-like nature of linked

lists but also because writing an input routine from the keyboard would

make the program bigger. (That type of linked list program is coming, by

the way.) So the first problem to tackle with spiffing up the PREZEZ.C

program is to handle inputting the strings into the list. And the best way

to do that is with an array and a loop.

 The following program is POTUS.C. Potus is the code word used for the

President Of The United States. (Kind of James Bondish, huh?) It shows how

to display an array of strings using a while loop. A null string ends the

loop, as opposed to some counter or number.

Name: POTUS.C

#include <stdio.h>

void main()

char *presidents[] = {

"George Washington",

"John Adams",

"Thomas Jefferson",

"James Madison",

"James Monroe",

""

};

int index = 0;

while(*presidents[index])
 {

printf("%s\n",presidents[index]);

index++;

Type in the above source code, or Shift+click on this link to download a

copy.

Compile and Run -- you know the drill. When you run it, you should see the

following:

George Washington

John Adams

Thomas Jefferson

James Madison

James Monroe

The array of Presidents works like any string array: It's an array of

pointers, but in this case the pointers contain the address of a set of

strings in memory. (Don't let that "array of pointers" thing get to you;

it's an array o' strings.)

The last item in the array is a NULL string. This is similar to making the

last item in a list of integers zero; it marks the end of a long list. The

array isn't even defined with a number in the brackets. That way you can

add new names without having to re-count anything.

 To count the items in the list, an index variable is used. It's

initiazlied to zero, the first element in the array.

A while loop ticks through each string in the list, displaying each string

and incrementing the index variable. Only when the NULL string is

encountered does the loop stop.

And that, ladies and gentlemen, is exactly how you will fill the names in

your linked list of presidents. End of side trip.

See Chapter 9 for more information on arrays and strings.

Lesson 10.7 discusses the relationship between pointers and strings.

If you dare to look at it, Lesson 10.10 covers the thorny ground of

arrays of pointers.

Second Detour: Looping the Linked List

To create the linked list you need a loop, a loop with a single if test in

it to be sure you handle the first structure in the linked list properly.

After that's done, filling in the remainder of the structures works

similarly to the way the POTUS.C program displayed the presidents' names.

The following program turns the trick:

Name: PREZEZ3.C

#include <stdio.h>

#include <stdlib.h>

#include <malloc.h>

#include <string.h>
void main()

struct pres {

char name[25];

struct pres *next;

};

char *presidents[] = {

"George Washington",

"John Adams",

"Thomas Jefferson",

"James Madison",

"James Monroe",

""

};

struct pres *first;

struct pres *current;

struct pres *new;

int index = 0;

/* Initialize all the pointers */

 first = (struct pres *)NULL;

current = (struct pres *)NULL;

new = (struct pres *)NULL;

/* Create the first item in the list */

first = (struct pres *)malloc(sizeof(struct pres));

current = first;

/* Create each item in the list */

while(*presidents[index])

strcpy(current->name,presidents[index]);

new = (struct pres *)malloc(sizeof(struct pres));

current->next = new;

current = new;

index++;

/* display the results...

* (this section to come)

* Just display first item in the list

*/

printf("The first structure has been created:\n");

printf("first->name = %s\n",first->name);

printf("next structure address = %i\n",first->next);

Type in the above source code or Shift+click here to download a copy. The

program is explained in detail below.

Compile and run!

The program's output is stunted, displaying only the first structure in

the list:

The first structure has been created:

first->name = George Washington

next structure address = 3736

I limited output because displaying the structures opens a whole new can

of worms that this program fails to address. (More on that later.)

How it Works

Here are the details -- the blow-by-blow, though I lack rights to use that

icon here:

The program starts by creating the first structure in the linked list.

That address marks the start of the linked list, so the *first variable
must never change:

first = (struct pres *)malloc(sizeof(struct pres));

current = first;

After the *first variable is created, it's stored in the *current variable

for use inside the loop. At this point, the first structure can be filled

in the loop just like every other structure.

The guts of the program consists of a while loop, identical to the one in

POTUS.C:

while(*presidents[index])

The loop continues to spin as long as there are strings in the presidents

array. The index variable is incremented at the end of the loop to keep

track of the position in the array:

index++;

Inside the loop, two things happen: The string is saved in the structure

and a new structure and pointer are created and stored:

strcpy(current->name,presidents[index]);

new = (struct pres *)malloc(sizeof(struct pres));

current->next = new;

This works the same as for the PREZEZ2.C program, however in this program

you've eliminated the extra pointer variables and replaced them with

*current and *new. The *current pointer references the current structure
 and *new is used only to get the address of the next structure in the

list.

The following statement ensures that the next structure is properly

filled-in:

current = new;

This means that the next time the loop spins, it will be filling in the

next structure in the list. After that, the index variable is incremented

and the process continues, creating structures to fill.

And Now, the Problems . . .

The PREZEZ3.C program isn't without its problems. Before that, note what

works:

The program creates a linked list, with each structure containing the

address of the next structure in memory.

The loop fills each structure using a minimum of pointer variables

The first structure's address is stored in the *first variable, allowing

the rest of the program to locate that structure and then all the other

structures in the linked list.

And now a few things that don't work:

The program doesn't display all the items in the list. No biggie; the

next program in this Lesson shows you how. (I just didn't want the thing

overly bulked up.)

The program works only in one fell swoop: it creates the linked list all

at once. If you wanted to write a program that added items to the linked
 list one at a time, this code wouldn't cut it. (More on that in the next

lesson).

Okay. Minor gripes, but legit. The first item in the gripe list is most

important. How would you display all the items in the linked list? It's

possible, of course. But how?

Think about it! Go back and look at the source code. I'll wait here.

Do-be-do-be-do Dee-do-be-do-be . . .

Right. You need some kind of loop. A for loop would work, but you'd be

cheating; you already know how many linked lists there are so you could

hard-code the thing:

/* Display the results */

current = first;

for(index=0;index<5;index++) { printf("structure %i: ",index+1);

printf("%s\n",current->name);

current = current->next;

Use the *current pointer to stomp through the linked list. Initialize it

to the start of the list, as shown above with current = first;. Then,

because you know there are only five structures in the list, re-use the

index variable in a for loop to march through the list only five times.

Inside the loop, use printf to display each structure's data. Then the

*current pointer is reset to the address of the next structure using the
current = current->next; statement. Nifty. Tidy. And it works.

BUT IT'S NOT GOOD PROGRAMMING!

You can Shift+click here to download the PREZEZ4.C program, which contains

the above for loop to display the structures. Go ahead, even though you

know it's not the real solution to the problem.

Compile the code and run the program. Output looks like this:

Structure 1: George Washington

Structure 2: John Adams

Structure 3: Thomas Jefferson

Structure 4: James Madison

Structure 5: James Monroe

If you're from the school of "If it works, it's done," then you can stop

here. Otherwise, you'll need to devise a more elegant solution.

Right: it still needs a loop. But why not a while loop? After all, a while

loop was used to create the structures. Likewise, a while loop can be used

to display them. The drawback to that approach is that a while loop needs

a condition on which to stop. Right now, there is no such condition, which

is yet another problem with the PREZEZ.C series of programs: how can you

tell when the linked list is done.

Now you could create a *last pointer. But that's overkill. (If you thought

of that, then you're using your brain too much. This is C programming, not

Pascal!) Think instead of how the first while loop works. How does it know
 when to stop?

NULL...

Right. To end the linked list, you need a NULL pointer. The final

structure in the list should have a pointer to NULL instead of a pointer

to the next structure. That makes so much sense you'd expect it to be

taught in Sunday School: The last structure in the list, logically, has

nothing following it. It demands a NULL pointer!

Here's another look at that linked list illustration to drive the point

home:

Look at the last structure in the list. What does it end with? 000? Isn't

that a NULL? Absolutely!

Now two things must be done: The loop that creates the linked list must

cap it off with a NULL pointer in the last structure. After that, the loop

that displays the list uses the same NULL to determine when the list is

done. You could ruminate on this for a time if you like, or just peek

ahead to the following source code, the last of the PREZEZ.C series.

Name: PREZEZ5.C

#include <stdio.h>

#include <stdlib.h>

#include <malloc.h>

#include <string.h>
void main()

struct pres {

char name[25];

struct pres *next;

};

char *presidents[] = {

"George Washington",

"John Adams",

"Thomas Jefferson",

"James Madison",

"James Monroe",

""

};

struct pres *first;

struct pres *current;

struct pres *new;

int index = 0;

/* Initialize all the pointers */

 first = (struct pres *)NULL;

current = (struct pres *)NULL;

new = (struct pres *)NULL;

/* Create the first item in the list */

first = (struct pres *)malloc(sizeof(struct pres));

current = first;

/* Create each item in the list */

while(1)

strcpy(current->name,presidents[index]);

index++;

if(!*presidents[index])

current->next = (struct pres *)NULL;

break;

else

{
 new = (struct pres *)malloc(sizeof(struct pres));

current->next = new;

current = new;

/* Display the results */

current = first;

index = 1;

while(current)

printf("Structure %i: ",index++);

printf("%s\n",current->name);

current = current->next;

Type in the above source code or Shift+click here to download a copy.

Compile and Run. The output will be the same as for PREZEZ4.C (above) but

the method is better, more flexible. Here's what's going on:

First, a while(1) ("while true") loop is used to spin indefinitely. That's

because an if-else structure is used inside the loop to determine when the

last item from the array has been read. (There are other ways to do this
as well.)

Inside the loop, the structure is filled in two steps. First, the

president's name is fetched from the array and stuffed into the structure:

strcpy(current->name,presidents[index]);

This works the same for all the structures, whether they're the first,

middle or last.

Next, the index variable is incremented:

index++;

You need to determine whether or not this is the last structure and the

way to know that is to see if there is another item lurking in the

presidents array. An if statement checks for you:

if(!*presidents[index])

This is a peek ahead. The value of the string there isn't used now; it

will be used the next time the loop repeats (if it repeats). Otherwise, if

tests to see whether anything is there at all. If not (which is the ! part

of the test), then the next pointer in the structure is assigned a NULL

structure-pointer value and the loop breaks:

current->next = (struct pres *)NULL;

break;

Otherwise, malloc() grabs another chunk of memory. The new address is

saved, the current pointer is reset to that address so that the loop can

repeat and the new structure be filled:

new = (struct pres *)malloc(sizeof(struct pres));

 current->next = new;

current = new;

To display the results a simpler construction can be used. After all, the

next pointer in each structure variable holds an address for every item in

the list but the last one. In that case, it holds the NULL value. And all

of the C language looping statements interpret NULL as a FALSE.

Before the loop starts, the current pointer is initialized to the address

of the first item in the list, and the index variable is set to one (for

counting purposes):

current = first;

index = 1;

After that, the loop can spin until the value of the next pointer is zero

(index is not used for the loop!):

while(current)

printf("Structure %i: ",index++);

printf("%s\n",current->name);

current = current->next;

Only the name part of each structure is displayed with a printf. In fact,

it could all go on one line:

printf("Structure %i: %s\n",index+1,current->name);

Then, the current variable is set equal to the address of the next
 structure in the linked list:

current = current->next;

When the last item in the list is encountered, the current variable will

be set equal to NULL. That's the condition the while loop interprets as

the end of the loop.

The program may not be utterly elegant, and it's certainly not the only

way to accomplish the task, but it works and is flexible: You can add

the rest of the presidents' names to the array and the sucker will still

work as advertised -- which is the bottom line.

The next lesson shows a different way to add and display elements in a

linked list -- more of a database-like thing than the PREZEZ.C series of

programs.

Bonus C For Dummies Lesson 17-3Lesson 17-3 – The Dawn of the Database

Linked lists are really all about databases. The structure itself is like

a record in a database, and the variables in the structure are fields. I

can't see how it could get any more obnoxiously obvious. And if that's all

true, then the linked list is really just another term for a database. So

starting with this Lesson and continuing for a few more Lessons, you're

about to take a bath into the world of databases.

The Ubiquitous Bank Account Program

It seems like the last few times I've gotten questions from readers about

linked lists it had to do with banking programs. You know the type: They
 lose track of you and your money – unless you owe them money, in which

case they're on you like white on rice.

The following program is BANK1.C, which should tell you immediately that

it's only the first in what will probably be several more BANK programs.

This program creates a linked list one record at a time. Unlike the

previous linked list examples, this one take a giant step into the area of

databases.

Name: BANK1.C

#include <stdio.h>

#include <stdlib.h>

#include <malloc.h>

#include <string.h>

#include <ctype.h>

#include <conio.h>

void addNewAccount(void);

struct account {

int number;

char lastname[15];

char firstname[15];

float balance;
 struct account *next;

};

struct account *first,*current,*new;

int anum = 0;

void main()

char ch;

/* Initialize all the pointers */

first = (struct account *)NULL;

current = (struct account *)NULL;

new = (struct account *)NULL;

do

puts("\nA - Add a new account");

puts("Q - Quit this program\n");

printf("\tYour choice:");

ch = toupper(getch());

switch(ch)

{
 case 'A':

puts("Add new account\n");

addNewAccount();

break;

case 'Q':

puts("Quit\n");

default:

break;

while(ch != 'Q');

void addNewAccount(void)

char buffer[64];

new = (struct account *)malloc(sizeof(struct account));

/* Check to see if this is the first record

* If so, then initialize all the pointers to this,

* first structure in the database

*/
 if(first==(struct account *)NULL)

first = current = new;

/* Otherwise, you must find the end of the structure list

* (Easily spotted by the NULL pointer) and add on the

* new structure you just allocated memory for

*/

else

current = first; //make the first record the current one

//and loop through all records:

while(current->next != (struct account *)NULL)

current = current->next;

//the last record is found

current->next = new; //save the address of the new record

current = new; //and make the current record the new one

/* Now, you just fill in the new structure */

 anum++;

printf(" Account number: %5i\n",anum);

current->number = anum;

printf(" Enter customer's last name: ");

gets(current->lastname);

printf("Enter customer's first name: ");

gets(current->firstname);

printf(" Enter account balance: $");

current->balance = atof(gets(buffer));

/* Finally, cap the new record with a NULL pointer

* so that you know it's the last record:

*/

current->next = (struct account *)NULL;

You'd be nuts to type in that whole program by hand. Instead, Shift+Click

here to download a copy for your files.

Compile the code. Run it.

A - Add a new account

Q - Quit this program

Your choice:

There are only two options at this stage. Type A to add a new record.

Fill in the last name, first name and a bogus account value. After that,

you're returned to the main menu.

Presently there's no way to review the records you're creating; that

function will be added in the second half of this lesson. Before then,

what's going on needs to be dissected.

You can press A again to add another record. In, fact you can keep

adding records all day long.

The program already specifies the account number for you, starting with

1 (one). This value is stored in the anum variable.

Notice that the anum variable is global, as is the definition of the

structure and the three linked-list pointers: first, current and next.

Please review Lessons 17-1 and 17-2 on linked lists if the concept is

new to you.

Chapter 11 in Volume II of C for Dummies discusses structures in

general.

How it All Works

The key to the program is a small snippet of code that's used to find the

last structure in a linked list. This is something new; the previous

 lessons all stuffed the link list in one fell swoop. This time, to add a

new structure (or actually perform just about any operation on the linked

list) you need to know which structure is the last. Here's the simple

while loop that does that:

while(current->next != (struct account *)NULL)

current = current->next;

The pointer variable current contains the address of the current

structure.

The value of current->next is the address of the next structure in the

list. If that address is NULL, it means you've found the last structure in

the list. The while loop stops.

If the value of current->next is not NULL, then the statement

current = current->next

adjusts the current pointer to contain the address of the next structure

in the list. The process repeats.

This while loop elegantly stomps through every structure in the list.

After it's done, the value of the current pointer is the address of the

last structure in the list (the one with NULL as the value of

current->next. The Figure below illustrates this somewhat.

Working the whole BANK1.C program from the top down: The main function

merely displays a menu and calls other functions to do the actual work.

Nothing there should surprise you, other than the do-while loop instead of
 a while(TRUE) loop (which I normally use). Again, this is just Yet Another

Way things can be done in C.

The addNewAccount function is where the fun starts. And the fun starts

with allocating space for the new structure:

new = (struct account *)malloc(sizeof(struct account));

Now, normally there should be some code there comparing new to NULL just

to see if memory was available. But these structures are so small it

should never happen. (In a "real" program you must do that.) (I know I'm

bad for not checking. A later lesson actually does the checking. Promise!)

After creating the new structure, you need to check to see if it's the

first structure in the linked list:

if(first==(struct account *)NULL)

All the structure pointers were initialized to NULL in the main function,

so if first is still a NULL pointer, then you're adding the first record

in the list. (Yet another reason to always initialize pointers!). The

address of this, the first structure, is then stored in the first pointer

variable, as well as the current pointer variable:

first = current = new;

Otherwise, if there are already structures in the list, you need to spin

up to the last structure and add the new structure to the end. That

happens in an else part of the program:

else

{
 current = first; //make the first record the current one

//and loop through all records:

while(current->next != (struct account *)NULL)

current = current->next;

//the last record is found

current->next = new; //save the address of the new record

current = new; //and make the current record the new one

First the current pointer is initialized to the address of the first

structure in the list. Then your friendly while loop is used to spin

through all the structures until the last one is found (the one with NULL

as its current->new value.)

After the while loop spins, the record current points at is the last

record in the list. First, the current->new value is changed from NULL to

the address of the new structure. Then the new structure is made the

current structure, so that it's data may be filled in.

anum++;

printf(" Account number: %5i\n",anum);

current->number = anum;

The value of the global variable anum (account number) is incremented and

assigned to the new account. The extra spaces in the printf string line up

the colon with the next several prompts. And the %5i makes the integer
 display to 5 digits.

The next few questions and answers are pretty easy to figure out

(providing you've been through Volume I).

Finally – and most importantly – the current->next variable is assigned

the NULL value, marking the end of the list:

current->next = (struct account *)NULL;

That ties everything up nicely.

Adding the List All Routine

It's pointless to create data and shove it into memory without any chance

to getting it back out. It's like Garbage-In/Garbage-Out, but without the

Garbage-Out part. Fortunately, listing the contents of the structure is a

snap.

In fact, it's so easy, I would like you to try it on your own. So don't

peek ahead! Sit down and write a routine called listAll and stick it on

the end of your BANK1.C source code file. Then modify the main function so

that pressing the L key displays all the files in the list.

Do it now!

Compile? Run?

Did it work?

If not, keep in mind you can use the NULL byte at the end of the list as a

test for the end of the loop.

How about this. Run the program again, and press L to list all the records
 before you entered any records.

Did it crash?

Okay. Give up? The following is my code for the listAll function. This

isn't the best or only way to do, just another way:

void listAll(void)

if(first==(struct account *)NULL)

puts("There are no records to print!");

else

current=first;

do

printf("%5i: %-15s %-15s $%8.2f\n",\

current->number,\

current->lastname,\

current->firstname,\

current->balance);

while((current=current->next) != (struct account *)NULL);

First, an if test is done to ensure that there are records to list. If the
 value of the first pointer is still NULL, then the linked list is empty.

The program can go home!

The main bulk of the structure listing routine is a do-while loop. This is

one of the smoothest ways to do things without skipping over the first or

last item in the linked list. The loop is driven by the following:

(current=current->next) != (struct account *)NULL

The current=current->next statement moves the current pointer up through

the linked list. But the != stops it when the result of that operation is

the NULL pointer. This is one way to read in the last structure in the

list (other ways skip the last structure). After all,

current=current->:next does assign NULL to the value of current for the

last structure. When that happens, the do-while loop stops cold.

Inside the loop, a printf statement formats the output. This lines up

everything in nice, neat columns and rows. Here is sample of this

routine's output:

1: Lennon John $ 100.00

2: McCartney Paul $ 100.00

3: Harrison George $ 100.00

4: Starr Ringo $ 1000.00

Don't worry about modifying your program to look like mine. Instead, just

Shift+click here to download the complete source code for BANK2.C.

Compile and run BANK2.C if you like. Compare it with your own source code.
 Remember, it's not the best way to do things, just another way.

The backslashes in the printf statement are to separate it onto

different lines. (It is rather long.)

Numbers between the % and the i or s in a printf formatting string tell

printf how many spaces to use when displaying data.

The - (minus sign) in the printf statement left-justifies the output.

The next lesson continues this series with examples of deleting records

in the database. Betcha can't wait!

Bonus C For Dummies Lesson 17-4Lesson 17-4 – Decimating the Linked List

Most of us take the word "decimate" to mean "destroy," which is true but

not the original meaning of the word. When the Roman army decimated a foe,

they just killed every tenth guy. Ten = dec, which is where decimate came

from.

Decimation is what you're about to do in your BANK.C program. But you

won't be destroying the list entirely. Instead, you'll just be killing off

random structures in the linked list. This is entirely workable – in a

sort of insane, twisted way.

Getting rid of deadbeat records

If you have a database that's just a series of 3x5 cards, then removing

old entries is cinchy: pull out the dead card. No one would know!
Now suppose your database is a series of structures in an array. How can

you delete structure account[7]? If you just zero out the entries, then

you have a blank record in the database. No, you'd probably have to copy

all the structures account[8] and above down a notch. Yeah, that might

work . . .

But arrays are boring. The truth is pointers are much easier to work with.

After all, a linked list is nothing but a string of pointers to various

spots in memory. Consider this illustration (not to scale):

The malloc function can really put the new structure it allocates anywhere

in memory. All you get back is the address in a pointer variable. Even so,

providing you link the structures together, it doesn't matter where in

memory each record lives.

To delete a record, you merely remove its pointer: Take the pointer in the

previous record and have it point at the next record. Lookee:

The pointer in the second structure now points to the fourth structure.

The third structure is out of the list. (And, realistically, in memory the

records may not be sequential anyway.) How does it work? Pointers!

Yes, it's true, the deleted structure is not removed from memory.

Well, you could use the free function to remove the structure from

memory. Mostly, however, programs just re-pack or re-index, which

removes the dead records.

 Many programs use this technique. Even DOS/Windows: A delete file is

merely "skipped over" on disk. This is how UNDELETE programs work; they

find the remains of the dead file and then put it back into the list.

The fact that pointers connect these structures will be used in a later

lesson to sort them all.

Adding a delete routine to the BANK.C program

To delete a structure from the BANK.C program you need to know three

things:

Which structure to delete

The address stored in previous structure's *next pointer (which pointers

to the structure you need to delete)

The address of the following structure in memory

So, basically, you take the address stored in the previous structure's

*next pointer and store the address of the next structure in there.

(Review the illustrations earlier in this Lesson to get an idea of how

this works). Only one new variable is needed: a pointer to hold the

address of the previous structure in memory. The following code does it

all:

void deleteAccount(void)

char ch;

struct account *previous;

if(first==(struct account *)NULL)

puts("There are no records to delete!");

else

current=first;

do

printf("%5i: %-15s %-15s $%8.2f\n",\

current->number,\

current->lastname,\

current->firstname,\

current->balance);

printf("DELETE THIS RECORD?");

ch = toupper(getch());

if(ch=='Y')

puts("Yes!");

if(current==first)

first=current->next;

break;

else
 {

previous->next = current->next;

break;

else

puts("No!");

previous=current;

while((current=current->next) != (struct account *)NULL);

Copy and paste the above code into your BANK2.C program. Remember to

prototype! Remember to add a Delete item to the menu! Save your source

code to disk as BANK3.C. Or, just Shift+Click here to download a copy of

BANK3.C

Compile and Run!

Enter about three or four different records, then try the new Delete

command. Delete a record in the middle, then list 'em all. Then try

deleting the first and last records to make sure that works.
 So How Does It Work?

The deleteAccount function is relatively simple, with only a few new

tricks. First, only two variables are required: ch to hold the Y or N key

press response, and *previous to hold the address of the previous

structure in memory. Most everything else is pulled from other parts of

the program.

The entire function works off of a do-while loop exactly the same as the

one found in the listAll function – except for the Y or N prompt:

printf("DELETE THIS RECORD?");

ch = toupper(getch());

If the user press Y, then another if-else structure is used to delete the

record.

First, a test is made to see if the current record is the first record:

if(current==first)

first=current->next;

break;

If the current record is the first record, then all you need to do is

change the value stored in the first pointer (the address of the first

structure in memory). Just change the value to the address of the next
 structure in memory and you're done. The new starting address is stored in

the first pointer.

If the current structure is not the first structure, then the following

takes place (the tricky part):

else

previous->next = current->next;

break;

The address of the previous structure is stored in the previous variable.

The value of previous->next then points to the current structure and

current->next points to the next structure. So to eliminate the current

structure, you change the value of previous->next to equal the value of

current->next. Get it? (Review the illustration.)

But where did previous come from? From this:

else

puts("No!");

previous=current;

If the user types N to not delete a record, then the address of the
 current structure is stored in the previous variable. This works since the

only time that wouldn't happen is when the first record is deleted – and

that's already happened earlier in the loop! (Tricky, eh?)

If it's still fuzzy as to why this function works, review it again and

think of the illustrations.

Remember that pointers are addresses. In this program, they hold the

address of structures at various locations in memory. The next value in

each structure is what makes this a linked list – and what allows you to

skip over structures without really having to delete anything or shuffle

stuff around.

Yes, technically you should use the free function to remove the

structure from memory. I will show you how that's done in a later

lesson. (It's okay at this point because the structures are small and

you're typing them all in; I don't expect you to type in more than 32K

of data, okay?)

Bonus C For Dummies Lesson 17-5Lesson 17-5 – Free At Last!

I've been bad. Bad author! Bad! I've done a lot of explaining how malloc

works, here on the Web and in the Books, but I left out one little,

teensy, weensy thing:

After it's allocated and used, memory should be freed.

The need to free memory shall become painfully apparent to you as your

programs grow. All that mallocing stuff gobbles up tons of memory, never

giving it back. Especially for the tiny programs you write in these
 lessons, 64K of memory is about all you have. Lose it and it's gone.

Now, I haven't been totally bad; for example, the BANK.C programs you've

been working with use manual entry. You must type in all the data from the

keyboard. So the odds are pretty long that you'll never chew up all memory

by typing something in. But if you're reading data from disk (which is the

topic of the next lesson), you're in trouble.

A Memory Consuming Example

The following program uses malloc to assign 32K chunks of data. Now malloc

returns the address of the chunk of memory it allocates, but in this

example, the address isn't saved. I'm just trying to prove a point:

Name: GOBBLE.C

#include <stdio.h>

#include <stdlib.h>

void main()

int x;

for(x=0;x<10000;x++)

printf("Allocating block %i\n",x);

if((char *)malloc(32767) == NULL)

 {

puts("Unable to allocate memory!");

exit(0);

Be a masochist and type the above source code into your editor, saving it

to disk as GOBBLE.C. Or, be a smart Web user and Shift+Click here to

download a copy of GOBBLE.C

Compile and Run!

You might see something like this:

Allocating block 0

Allocating block 1

Unable to allocate memory!

If you're lucky, you may get more output than what I got above. But

eventually, at some point in the run, the program stops because there is

no more memory to allocate. Tragic, eh?

The program uses the (char *) prototype to allocate 32,676 bytes (32K)

of memory.

Malloc returns the address of the memory block allocated. That value

isn't saved in this program. (No need to.)

When malloc returns NULL, it means no memory could be allocated. This

 program uses an if test to determine that.

The actual number of 32K chunks the program allocates depends upon which

memory module your compiler is using. For DOS programs, that's typically

the "small" or "tiny" module, which gives the program only 64K of

memory. If you can, try reconfiguring your compiler to the "huge" or

"outrageous" module to see how much memory gets allocated before it's

all gone.

Free up some memory with free

It's frustrating to have a PC with megabytes of RAM, yet you can write a

silly little program that supposedly gobbles up all memory in only two 32K

chunks. Now you're actually facing a conundrum many PC users have: "My

computer has megabytes of RAM, how come I'm running out of memory?

The truth is, memory must be allocated, used and then freed for use by

something else. It's only fair. But DOS (and Windows to some extent)

doesn't really, truly manage memory in a PC. For example, Microsoft Word

version 6 (for Windows) would typically grab all of a computer's resources

when it ran. If you ever worked with Word for a long period of time,

eventually you would get the cheery error message "Not enough memory to

save document." (That was just bad programming; Word's programmers were

hogging up resources and not returning them when they were done.)

To make life better for yourself and your users, it's always nice to free

up memory after you've used it. Say, for example, you write a program that

reads a file from disk, gathers information, changes settings, and then
 closes the file. It's just plain nice to free up the space used by the

buffer. And to do that, you use the free function. ( Click here for a

definition.)

Essentially, the free function de-allocates memory allocated by malloc.

And any time you allocate memory that's not used again in a program, you

should employ the free function to clear up that memory. The following

program, CHEW.C, is a fix-up of GOBBLE.C, which frees the memory and

allows all 10,000 blocks to be allocated:

Name: CHEW.C

#include <stdio.h>

#include <stdlib.h>

void main()

int x;

char *buffer;

for(x=0;x<10000;x++)

printf("Allocating block %i\n",x);

if((buffer=(char *)malloc(32767)) == NULL)

{
 puts("Unable to allocate memory!");

exit(0);

printf("Freeing block %i\n",x);

free((void *)buffer);

Type in the above source code or Shift+Click here to download yourself a

copy. Compile and Run.

You'll notice that the program runs all the way through, allocating memory

and freeing it up so that you really never run out.

free doesn't "erase" memory; it merely flags a chunk of memory as

available for re-allocation by another malloc function.

You don't really have to use free every time you use malloc. However,

any time you allocate memory that isn't used again, you should free it

up.

An example of not freeing memory would be creating a buffer to store

information that's used the entire time the program runs. In that case,

the memory is freed by the operating system when the program quits.

The next lesson, when it's published here on the Web, deals with

allocating and freeing memory using the BANK program.

Bonus C For Dummies Lesson 18-1Lesson 18-1 – Setting the Right Mode
Nothing puts the twinkle in a programmer's eye like graphics programming.

There's something about it that just seems naturally fun. And there's

nothing wrong with that. Graphics are fun. But unlike a lot of other

programming -- especially the stuff in C for Dummies -- graphics

programming is burdened with math.

Oh no! Math!

Relax. Always remember that the computer does the math. Still, it's up to

you to know the equations. If you've been through Algebra in high school

you may know some of them. Here are two popular equations you might

remember if you weren't sleeping too soundly:

The first is the equation for a line: Y and X are coordinates you'd plot

on a graph (or on the screen). M is the slope of the line, up or down, and

B is where the line crosses the Y (vertical) access.

The second equation is for plotting a circle, with X squared and Y squared

the coordinates and R (also squared) being the circle's radius. Simple.

Simple. Simple. (And I bet if you knew you'd be plotting graphics on a PCs

screen instead of fumbling around with a pencil on graph paper, you'd

probably have paid more attention in class!)

As long as you can work with those boogers in C, tie them into a loop, and

translate the results to the giant pixel grid on your computer's monitor,

you should be in great shape! (Or great shapes, depending on what your

program draws.)
A Quick [Boring] PC Graphics History Lesson

The PC has perhaps the worst graphics standards in the history of

computing. Oh, well, maybe the Apple II had a worse system (and God bless

the few who really understood it). The Mac is far better. Best? The Amiga.

But that's trivia. What's important to know is that the PC has several

weird graphics "standards," some of which aren't even standards in the

true sense of the word.

MDA. Originally, the PC came with the Monochrome Misplay Adapter (MDA)

card. It showed nice, crisp white text on a black background. Or green

text on a black background or amber text. (I liked the amber text myself.

In fact, my friend Tom still has my old amber monitor.) Color? Nope.

Graphics? Nope. Only text.

CGA. If you wanted color, you waited a few months until IBM released the

Color Graphics Adapter (CGA). It shows ugly, fuzzy text in one of 16

colors on a background of up to 8 colors. (See Lesson 13-5 for more

information on colored text.) Color? Yes, 16 of them in the text mode.

Graphics? Yes. There were two graphics modes in the CGA: The low

resolution 320 x 200 mode with four colors (black, white, magenta and

cyan) or a high resolution 640 x 200 with only two colors, black and

white. But that was enough for most PC users.

 Unintended moral: The higher the graphics resolution, the fewer colors

you get. The lower the resolution, the more colors. It's a nice

trade-off since you can use more colors to fool the eye into thinking it

sees higher resolution.

Let me be blunt: CGA sucked. The text was horrid. That was one reason I

used monochrome for years before switching to VGA in 1988. But I didn't

use the MDA. I used yet another graphics standard: Hercules, or

monographics. The Hercules card was a clone of the old MDA, but it also

had a special graphics mode. Sure, the graphics were black-and-white, but

they looked fabulous and the text was much better than CGA. I even had a

few games that used Hercules graphics modes. And my Hercules card even

displayed italics and underlined fonts in the old DOS WordPefect. (Man, I

was in heaven! Ha-ha!)

EGA. To improve upon CGA, IBM created the EGA (Enhanced Graphics Adapter)

standard. EGA offered better-looking text over CGA plus a few new graphics

resolutions. Oh, but this is all trivia.

VGA. EGA was quickly replaced by the VGA standard, which IBM introduced in

1987 with its PS/2 computers. Actually, there were two VGA standards. The

first was MCGA, which came with IBM's PS/1 (the "home" computer). MCGA,

which stands for something I can't remember, had lots of graphics

resolutions and color combinations. VGA includes the MCGA standards, so

MCGA is pretty much a dud.

SVGA. Eventually, the "industry" improved upon VGA and produced Super VGA

or SVGA, the graphics standard nearly all PCs have been sold with since

the early '90s.

Why is any of this important?

Because you must set your PC to the proper graphics mode before you can

program in graphics. Your programs do this for you automatically. When you

run a PC game, it typically switches to another graphics mode on the

computer. That's because the programmer has chosen a graphics mode

(resolution and colors) proper for the game. When you program in graphics,

it's up to you to set the proper mode.

As a preview, follow these steps in Windows:

Start an MS-DOS prompt window. In Windows, run the COMMAND.COM program,

or click on the MS-DOS Prompt icon to start up a DOS window.

Switch to full-screen mode. If the DOS prompt is in a window in Windows,

then press Alt+Enter so that it fills the screen -- just like the good

old days!

Type the following:

MODE 40

Press Enter. Welcome to 40-column mode! This is an old CGA graphics mode,

designed primarily so that people could use their PCs with a TV set as a
monitor. (I wonder if anyone ever did?) In fact, the DIR command was

formatted for 40-column output primarily for this reason. (They changed

the DIR format starting with DOS version 6.)

Type EXIT to return to Windows.

If you're using an SVGA video system, then your computer has 15 different

modes you can use to display text or graphics. You pick the resolution and

the number of colors and go from there. The following table lists the

graphics modes you have to choose from. (If you don't see a mode listed in

the table, then the SVGA adapter cannot produce it; these are typically

MDA, PCjr and EGA graphics modes on one uses anymore anyway anyhow.)

ModeTypeResolutionColors

0x00Text40 x 2516

0x01Text40 x 2516

0x02Text80 x 2516

0x03Text80 x 2516

0x04Graphics

(alt. color)320 x 2004

0x05Graphics320 x 2004

0x06Graphics640 x 2002

0x07Graphics720 x 350

720 x 400*2

0x0dGraphics320 x 20016
0x0eGraphics640 x 20016

0x0fGraphics640 x 3502

0x10Graphics640 x 3504 or 16**

0x11Graphics640 x 4802

0x12Graphics640 x 48016

0x13Graphics320 x 200256

* The vertical resolution is set using function 0x12 of Interrupt 0x10 --

something I'll never mess with.

** Only EGA adapters with 64K of video RAM display 4 colors in mode 0x10.

My favorite mode is 0x13, which is an old MCGA mode available on all VGA

adapters. It uses a rather low resolution, 320 x 200 pixels, but has a ton

of colors: 256.

More modes are available than those shown above. The table lists only the

standard VGA modes. I know that Super VGA has even more colors and

resolutions. If you get into this, consider buying a technical manual on

VGA graphics.

Set Some Dem Modes!

Wow! This chapter has way too much text. But the point's been made: Before

you do doodle with graphics, you must first set the graphics mode.

Actually, you should do several things: You should discover what graphics

mode the PC is currently in, save it, set the mode to what you want, then

reset the mode back to what it was when you're done. That's a lot of work,

but it's being nice to your user.

 Borland C++ has some graphics mode setting commands, but Microsoft C and

all other C language compilers probably don't. Because of that, I'm going

to use in-line Assembly language to run the graphics basics for this

Chapter. (See Chapter 16 for more information.)

To set the mode you use the PC's Video Interrupt, 0x10, to call the BIOS.

Here's the routine:

void set_mode(unsigned char mode)

_asm mov ah,00

_asm mov al,mode

_asm int 10h

The set_mode function is called with an unsigned char value, from 0

through 255, which you would pluck from the table shown earlier in this

Lesson. That mode is saved into the AL register, the value zero put into

the AH register and the function sped off to Interrupt 10h (which is

Assembler for 0x10). Voila, ze mode is set.

Of course, fetching the current graphics mode is also important. Otherwise

you may leave your user in some low-rez dorky mode and they won't have a

clue as to how to get back to normal. (The MODE 80 command does the trick

in DOS.) Here is the function to fetch the current graphics mode:

unsigned char get_mode(void)

{
 unsigned char mode;

_asm mov ah,0fh

_asm int 10h

_asm mov mode,al

return(mode);

The value mode is returned from Interrupt 0x10 in the AL register. That's

saved into the mode variable and returned from the get_mode function.

And now, your program:

Name: GRAFMODE.C

#include <stdio.h>

#define NORMAL 0x03

#define LOWGRAF 0x04

#define HIGRAF 0x06

#define VGA256 0x13

unsigned char get_mode(void);

void set_mode(unsigned char mode);

void main()

unsigned char save;

//Save the current mode (which is probably 3)

save = get_mode();

printf("You're in mode %i now.\n",save);

getchar();

//Switch to low rez CGA mode

set_mode(LOWGRAF);

puts("Hello Low CGA Graphics Mode!");

getchar();

//High rez CGA mode

set_mode(HIGRAF);

puts("Hello High CGA Graphics Mode!");

getchar();
//Low rez MCGA colorful mode

set_mode(VGA256);

puts("Hello VGA 256 Graphics Mode!");

getchar();

//Restore mode before exiting

set_mode(save);

unsigned char get_mode(void)

unsigned char mode;

_asm mov ah,0fh //Return mode value

_asm int 10h //Video interrupt

_asm mov mode,al //Save mode

return(mode);

void set_mode(unsigned char mode)

{

_asm mov ah,00 //Set mode

_asm mov al,mode //mode value

_asm int 10h //Video interrupt

Type in the above source code, or Shift+click on this link to download a

copy.

Compile!

Run the program in a DOS window (if you're using Windows). And switch the

window over to full-screen by pressing Alt+Enter if you need to. (Windows

will change modes otherwise, which is annoying).

Now, run. You should see the following:

You're in mode 3 now

You're probably in mode 3, which is the standard 80 x 25 text mode.

Press Enter.

Clunky, huh? Press Enter.

That's the old "high" resolution mode. Ick. Press Enter

Finally you have the high color VGA image. That's the one you'll be

working with in this Lesson. Press Enter and your system will be restored

back to its original graphics mode.

 Did you notice that the printf function works in graphics modes? Pretty

much so. You can mix graphics and text any time. The cursor moves the

same on a graphics screen as it does on a text screen. Just use any C

language text function to display text.

Windows is actually capable of viewing all these modes in a DOS window

on the screen. Just press Alt+Enter to switch between full-screen and

windowed mode. I prefer to work full screen when I do graphics.

Feel free to modify the program to display other modes shown in the

Table.

To see if the program really does save the graphics mode, type MODE 40

before you run it. If the program restores the 40-column mode when it's

done, then you've just done your users a favor.

If you're using DJGPP, then you'll need to use the ATT nottaion, not the

Intel Assembly notation shown here. Visit http://www.delorie.com/djgpp/

for more information. (Sorry!)

Bonus C For Dummies Lesson 18-2Lesson 18-2 – Hello, Pixel Fairy!

(If you have Microsoft Visual C++, click the icon below.)
The graphics screen is a big grid. Well, the text screen is a big grid,

too. So if you've been plotting out text on the screen using rows and

columns, you should get used to plotting graphics on its grid in no time.

But with graphics, you're not plotting characters, you're plotting pixels.

All graphics, no matter how complex they look or what they do, are merely

a collection of pixels on the screen. So when you start creating graphics,

you start by plotting pixels on a grid. And since you're programming, you

can afford to have the computer do all the plotting work for you.

Every graphics program you've ever seen -- including Windows itself --

starts with a basic routine to plot pixels on the screen.

Other routines, for drawing lines and circles and such, are all

functions designed to plot pixels in a certain manner.

Yes, even animation involves plotting pixels.

It's Just a Big Grid

Pixels have three attributes associated with them: X position, Y position

and color. (Again, this is like text, where each character has a row and

column position as well as a character code value.)

Pixels have an X (horizontal) position and a Y (vertical) position. The

pixel grid starts in the upper left corner of the screen at position 0,0.

This is the same for all graphics resolutions. The maximum values for X

and Y differ from graphics mode to graphics mode. In the high color VGA
mode 0x13, the maximum X value is 320 and the maximum Y value is 200.

(Incidentally, that's the size of the graphic image shown above: 320 x 200

pixels.)

In addition to the X and Y locations, pixels also have a color value. Like

the resolution, this value depends on the graphics mode. Color values

range from 2 (on and off, or "black" and "white") on up to 256. The colors

assigned to those values vary with the graphics mode.

That's the basics of all graphics. No matter what you see on the screen,

it's basically a collection of pixels set at certain locations in the

graphics grid, each pixel tuned to a certain color.

Pixel is short for "picture element." It's a dot on the screen.

Pixar is the name of the company that produced Toy Story.

X locations go from zero upward as you move from left to right across

the screen. This is the same as you learned in Algebra or Geometry

class, plotting on graph paper.

Y locations go from zero upward as you move from the top down on your

screen. This is different from the graph paper you used in school (where

Y started at the lower left corner of the graph).

X values correspond to column values on the text screen; Y values

correspond to row values on the text screen.

Plotting Pixels

Before you can do any graphics on any computer, you must have a function

that plots a single pixel on the screen. The good news is that it's simple
 to do. The bad news is that there are dozens of ways to do it.

For example, one programmer I know likes to plot pixels by writing

directly to video memory. Yikes! You can also plot pixels by manipulating

the graphics hardware inside your computer. I have technical books full of

VGA and SVGA documentation that tell über nerds how to manipulate the

screen and do all sorts of twisted things to it. Again: Yikes!

Some folks opt to get graphics libraries for their programming whims. For

example, you can buy a library full of high-speed graphics routines --

everything from plotting pixels to drawing lines to creating animations,

all programmed for you and ready to use.

For this chapter, I'm going to show you the tried and true method for

plotting pixels: using the PC's BIOS with in-line Assembly. This isn't the

fastest way to do things (people into computer graphics are into speed),

but it's standard. Here's the function:

void pset(int x,int y,unsigned char color)

_asm mov ah,0ch

_asm mov al,color

_asm mov cx,x

_asm mov dx,y

_asm int 10h

The pset function is called with three values: x, the pixel's X

 coordinate; y, the pixel's Y coordinate; and color, which sets the pixel's

color. These values are stuffed into the proper microprocessor registers

using in-line assembler. Register AH is loaded with the function number,

0x0C, which writes a pixel to the screen, and Video Interrupt 0x10 is

called.

Note that the pset function doesn't check to see if value numbers are

used. That's good in a way, since the BIOS function writes pixels in any

mode; the maximum value for X could be anything from 320 to 400 to 640 to

1024 or more. Of course, the pixel appears only in the proper graphics

mode.

I'll get into color values in a few pages, but generally speaking, zero is

always black (or turning a pixel off). Values from 1 on up could be any

color, though.

Time for a program to run pset through it's paces:

Name: PIXELS1.C

#include <stdio.h>

#define VGA256 0x13

#define XMAX 320

#define YMAX 200

void pset(int x,int y,unsigned char color);

unsigned char get_mode(void);

void set_mode(unsigned char mode);

void main()

unsigned char save,c;

int x,y;

/* Initialze things */

save = get_mode(); //Save current screen mode

set_mode(VGA256);

/* Main routine */

x=XMAX/2;

y=YMAX/2;

c=1;

pset(x,y,c);

getchar();

/* Reset stuff for end */

 set_mode(save); //Restore screen mode

void pset(int x,int y,unsigned char color)

_asm mov ah,0ch

_asm mov al,color

_asm mov cx,x

_asm mov dx,y

_asm int 10h

unsigned char get_mode(void)

unsigned char mode;

_asm mov ah,0fh

_asm int 10h

_asm mov mode,al

return(mode);

}
void set_mode(unsigned char mode)

_asm mov ah,00

_asm mov al,mode

_asm int 10h

Type in the above source code or, better still, Shift+click here to

download a copy.

Compile! Run! The screen goes graphical and you should see a single, blue

pixel -- one dot -- in the middle of the screen. That's one tiny building

block with which you can do anything you can visualize on your PC's

graphics screen.

As you know, I'm a big fan of defines. In the PIXELS1.C program there

are three of them: VGA256, XMAX and YMAX. VGA256 is the graphics mode

and XMAX and YMAX represent the maximum horizontal and vertical

resolutions for that mode, 320 and 200 respectfully.

Using the XMAX and Y MAX values in the program is a lot easier than

having to remember the values every time you write a graphics equation.

For example, if you properly define XMAX as the screen's horizontal

resolution, then you can always use XMAX or XMAX/2 to represent the far

left or center of the screen -- no matter which graphics mode you're in.

Refer to the Table in Lesson 18-1 for more information on graphics modes
 and resolutions.

More Dots

Heck ALL the Dots!

The following program uses a nested for loop to paint the entire low rez

VGA screen with all 256 possible color combinations. Dazzle your friends!

Name: PIXELS2.C

#include <stdio.h>

#define VGA256 0x13

#define XMAX 320

#define YMAX 200

#define MAXCOLORS 256

void pset(int x,int y,unsigned char color);

unsigned char get_mode(void);

void set_mode(unsigned char mode);

void main()

unsigned char save,c;

int x,y;
/* Initialze things */

save = get_mode(); //Save current screen mode

set_mode(VGA256);

/* Main routine */

for(c=0;c<MAXCOLORS;c++)

for(y=0;y<YMAX;y++)

for(x=0;x<XMAX;x++)

pset(x,y,c);

getchar();

/* Reset stuff for end */

set_mode(save); //Restore screen mode

}
void pset(int x,int y,unsigned char color)

_asm mov ah,0ch

_asm mov al,color

_asm mov cx,x

_asm mov dx,y

_asm int 10h

unsigned char get_mode(void)

unsigned char mode;

_asm mov ah,0fh

_asm int 10h

_asm mov mode,al

return(mode);

void set_mode(unsigned char mode)

_asm mov ah,00

 _asm mov al,mode

_asm int 10h

Doggedly type in all that source code, or just Shift+Click here to

download a freshly pressed copy, written by the author.

Compile! Run!

The nested for loops paint the entire screen, top to bottom. If you have a

slower PC, you may actually see the pixels fill the screen. (Yes, this

isn't the most efficient method for white washing a computer screen.)

Press ENTER to view the next color screen. Here's how the colors map out:

0 to 15 - Standard CGA colors

Black, Blue, Green, Cyan, Red, Magenta, Brown, Gray, Dark Gray, Bright

Blue, Bright Green, Bright Cyan, Bright Red, Bright Magenta, Yellow,

Bright White. (Incidentally, these are the same codes and colors for

color text on the screen. See Lesson 13-5.)

16 to 31 - Grayscale

Shades of gray from almost black on up through almost white.

32 to 103 - High intensity rainbow

Shades of Blue, Red and Green in high, moderate and low saturation.

104 to 175 - Moderate intensity rainbow

Shades of Blue, Red and Green in high, moderate and low saturation.

176 to 247 - Low intensity rainbow

Shades of Blue, Red and Green in high, moderate and low saturation.
 248 to 255 - Black

Black, black, black, black, black, black, black and black again.

Now you don't have to press the ENTER key 255 times to get the whole idea.

If you tire of it, press Ctrl+C to get out of the loop. You'll then have

to type MODE 80 to return to the normal text screen, or just keep working

in 40 column mode if you enjoy it.

Actually, if you're interested in viewing the VGA colors, you can refer to

a related program: click here.

The Old Randomizer Star Field Trick

Now you can get really silly with pixels. Don't worry about stringing them

into lines and circles yet. Instead, take that old chestnut the random

start field program out of the closet. This program combines the

randomizing functions from Volume I with the pset function introduced in

this Lesson. The results are predictable, but fun to watch nonetheless.

Name: PIXELS3.C

#include <stdio.h>

#include <stdlib.h>

#include <time.h> //for seeding randomizer

#define VGA256 0x13

#define XMAX 320

#define YMAX 200

#define MAXCOLORS 256

void pset(int x,int y,unsigned char color);

unsigned char get_mode(void);

void set_mode(unsigned char mode);

int rnd(int range);

void seedrnd(void);

void main()

unsigned char save,c;

int loop,x,y;

long int delay;

/* Initialze things */

seedrnd(); //Seed the randomizer

save = get_mode(); //Save current screen mode

set_mode(VGA256);

/* Main routine */

//Plot a pixel in a random location with a random color

 for(loop=0;loop<1000;loop++)

x=rnd(XMAX);

y=rnd(YMAX);

c=rnd(MAXCOLORS);

pset(x,y,c);

for(delay=0;delay<50000;delay++);

getchar();

/* Reset stuff for end */

set_mode(save); //Restore screen mode

void pset(int x,int y,unsigned char color)

_asm mov ah,0ch

_asm mov al,color

_asm mov cx,x

_asm mov dx,y

_asm int 10h

}
unsigned char get_mode(void)

unsigned char mode;

_asm mov ah,0fh

_asm int 10h

_asm mov mode,al

return(mode);

void set_mode(unsigned char mode)

_asm mov ah,00

_asm mov al,mode

_asm int 10h

/* Random number routines stolen from Volume I */

int rnd(int range)

int r;
 r=rand()%range; //spit up random num.

return(r);

void seedrnd(void)

srand((unsigned)time(NULL)); //seed randomizer

Type the above source code into your editor, or Shift+Click here to

download yourself a copy.

Compile! Run!

Are the starts out tonight?

I don't know if it's cloudy or bright . . .

Problem: the program loops only 1000 times. Can you imagine a better way

to halt it on your own? Something that would cause the stars to twinkle

until you pressed a key on the keyboard, say? I'll leave it to you to

divine a solution to that problem.

You can adjust the value of the delay for loop to make the stars twinkle

slower or faster. This depends on the speed of your computer.

Wow! Does this program remind you of the DIBBLE.C program from the end

of Volume I? Same concept, but pixels instead of happy faces.

Speaking of old programs, consider modifying the old DRUNK.C program

from the end of Volume I into a program that displays pixels on the
 screen. You'll need to adjust the maximum X and Y values, plus change

the drunk into a cursor of whatever color. Also, you'll need to

overwrite the drunk with color 0 (zero), which is black. See what you

come up with; in a few lessons I'll be tackling this program myself

using animation techniques.

Bonus C For Dummies Lesson 18-3Lesson 18-3 – Loverly Lines

A line is defined as connecting two points. A line is also defined as

that thing you wait in while you're at Disneyland. And not just

Disneyland, but Universal Studios, too. My son just went to Universal

Studios, Los Angeles to go on the Jurassic Park ride. The line was 3½

hours long. That line, too, was connecting two points: the point where my

son was and the point where people were boarding the 10-minute long ride.

Too much.

Back in geometry class, you drew line by first defining the two points on

graph paper. You drew a dot at the first location and a dot at the second

location. Then you used a ruler to connect the two dots. Amazing. On a

computer screen, which works like graph paper, you take the starting

location of the line and the ending location and then pray that there is

some simple line-drawing algorithm you can use like a ruler to connect

them. That's the gist of it. And don't worry -- this is still C for
Dummies. There won't be any math here.

Yes, I have the line drawing algorithm all worked out. It's explained in

the next section.

The line requires two locations on the computer screen. I'll refer to

the starting location as X1, Y1 and the ending location as X2, Y2.

All polygons are created with lines. So a rectangle, for example, is

merely a line drawn between four sets of coordinates.

It was really my mom's fault for taking my son to Universal Studios on a

Sunday in July, the last day of a promotional special. The park was

jammed. The line was 5 hours long to get into the Backdraft exhibit.

The Line Drawing Equation

All lines are drawn using the simple line drawing formula you probably

forget from algebra class:

You don't have to read this explanation: Y and X are the vertical and

horizontal coordinates of the line; for every Y position on the line there

is a corresponding X position. M is the slope of the line, or the ratio of

Y over X. And B is where the line crosses the Y-axis. Enough!

The idea is to convert that simple line-drawing equation into something

the C compiler can understand. After doing that, you work it out so that

the function can also plot pixels for all the X and Y coordinates of the

line. So, using the vast skills you honed after years of schooling, you

decided to look it up in a book where some propeller-head has already

 figured out the answer. But I won't let you get off that easy. At least

not all at once.

Drawing a Straight Line

The simplest type on line to draw is the straight line. For a horizontal

line, you would plot pixels from X1 to X2 with the same value of Y – a

simple for loop. For a vertical line, you plot pixels from Y1 to Y2 with

the same value of X – another simple for loop. However, a vertical line

has a slope of infinity (or some other wacky, nonsensical value). So it

must be treated as a special case, one that cannot follow the y=mx+b

formula. (Yes, an exception.)

Name: VERTLINE.C

#include <stdio.h>

#define VGA256 0x13

#define XMAX 320

#define YMAX 200

#define RED 4

void pset(int x,int y,unsigned char color);

unsigned char get_mode(void);

void set_mode(unsigned char mode);

void line(int x1,int y1,int x2,int y2,unsigned char color);

void main()

unsigned char save;

int x1,y1,x2,y2;

/* Initialze things */

save = get_mode(); //Save current screen mode

set_mode(VGA256);

/* Main routine */

printf("Coordinates [x1,y1,x2,y2]:");

scanf("%i,%i,%i,%i",&x1,&y1,&x2,&y2);

printf("(%i,%i) to (%i,%i)",x1,y1,x2,y2);

getchar();

line(x1,y1,x2,y2,RED);

getchar();

/* Reset stuff for end */

 set_mode(save); //Restore screen mode

void line(int x1,int y1,int x2,int y2,unsigned char color)

int x,y;

for(y=y1;y<=y2;y++)

pset(x1,y,color);

void pset(int x,int y,unsigned char color)

_asm mov ah,0ch

_asm mov al,color

_asm mov cx,x

_asm mov dx,y

_asm int 10h

unsigned char get_mode(void)

unsigned char mode;

 _asm mov ah,0fh

_asm int 10h

_asm mov mode,al

return(mode);

void set_mode(unsigned char mode)

_asm mov ah,00

_asm mov al,mode

_asm int 10h

This program uses the pset function from Lesson 18-2 to help you plot the

line. In addition to the line drawing function, I added code that lets you

input the starting and ending coordinates for the line. Now don't get

cocky with your values! It's easy to crash this program by typing in

negative numbers. (I'll leave it up to you to bulletproof the program on

your own.)

Type in the above source code or, if you're not a masochist, Shift+click

here to grab a copy the author typed in himself.

Compile! Run!
 Type in the four sets of coordinates, separating each by commas. Oh, heck,

just type in the following:

160,0,160,200

That's 160 for both values of X, which puts the line in the center of the

screen. Y1 is equal to 0 (at the top of the screen) and Y2 is equal to 200

(bottom). Press Enter to see the results. Press Enter again to quit.

Now. The program can draw any vertical line on the screen. It ignores the

X2 coordinate (which your compiler may alert you to when you created it),

but uses the two Y coordinates to draw the line. Alas, this isn't without

its problems.

Run the program again. This time type in the following for the

coordinates:

160,200,160,0

This is essentially the same thing, but with the coordinate sets reversed.

But will the program draw the line up? Press Enter to find out.

Nope. And it shouldn't surprise you. Review your program's for loop:

for(y=y1;y<=y2;y++)

The loop never works, not once. If Y1 is greater than Y2, then the for

loop quits before it loops even once. And it should because that's the way

the C language works. But, obviously, that's not going to fit into the

grand scheme of things when you draw a line. Obviously there will be times

when Y1 is greater than Y2 and your program will still need to draw the

line.
 The solution? Reverse the coordinates. Swap them. Add the following

function to the VERTLINE.C program:

/* Swap the values of two variables */

void swap(int *a,int *b)

int temp;

temp = *a;

*a = *b;

*b = temp;

Defining the variables as pointers here allows you to swap their values

without having to swap names. (It also allows the function to be used

elsewhere in the line function, which you'll discover later in this

lesson.)

Remember to prototype the function at the start of your source code:

void swap(int *a,int *b);

And finally, you'll need to modify the line function to take care of the

case when Y2 is larger than X2. Here's how that function looks:

void line(int x1,int y1,int x2,int y2,unsigned char color)

int x,y;
 if(y1 > y2)

swap(&y1,&y2);

for(y=y1;y<=y2;y++)

pset(x1,y,color);

A simple if test is all that's required to see if Y1 is greater than Y2.

If so, the values are swapped. (Refer to Chapter 10 in Volume II for more

information on how the ampersands are used when passing values to a

pointer function like swap.)

Make the proper adjustments to your program. Save. Compile. Run.

Now type in the following values for your coordinate sets:

160,200,160,0

The value of X is merely determines where the vertical line will slice the

screen. However, Y1 is greater than Y2 in this example. Will it work?

Press the Enter key to find out.

Well, you knew it would.

Drawing Lines Hither, Thither and Yon

There are actually many line drawing algorithms, but y=mx+b is the most

common. The puzzlement is how to shift the values of X1,Y1 and X2,Y2 into

that equation. The core code you can probably figure out yourself:

for(x=x1;x<=x2;x++)

{
 y = m*x+b;

pset(x,y,color);

A for loop whizzes the value X from X1 to X2. Then the value for Y is

calculated using y=mx+b. Finally, the pset function is used to plot the

actually points on the screen, X and Y. Cinchy!

Of course, where does M come from? And how do you calculate B?

M is the slope of the line, or the ratio of the change in Y over the

change in X. That's calculated using the formula shown above.

From your basic C knowledge, you should remember that anytime you use

division, you'll be dealing with a floating point number. So right off the

bat, you should know that M is a floating point variable. And you,

therefore, must typecast the results of the Y over X thing. But don't

think! Just look at this handy translation into C:

m = (float)(y2-y1)/(float)(x2-x1);

To solve for B you end up with something like the following:

Looks okay to me, boss! And it translates easier into C:

b = y1 - (m*x1);

Since the equation for B involves the value of M, B should also be a

floating point number. So both of them need to be declared at the start of

the new line function as floating point numbers. In fact, here is the
 entire new line function for you:

void line(int x1,int y1,int x2,int y2,unsigned char color)

int x,y;

float m,b;

// Plot straight vertical line

if (x2==x1)

if(y1 > y2)

swap(&y1,&y2);

for(y=y1;y<=y2;y++)

pset(x1,y,color);

return;

m = (float)(y2-y1)/(float)(x2-x1);

b = y1 - (m*x1);

for(x=x1;x<=x2;x++)
 {

y = m*x+b;

pset(x,y,color);

Above, you'll note that the new floating point M and B variables are

defined. Then comes the code for plotting a straight vertical line. That's

determined by matching the values of X1 and X2. If they're equal, then you

have a vertical line. The rest of the code between the if statements'

brackets is what you've already gone over in this lesson.

Next comes the new stuff. First the value of M is calculated, then the

value of B. Then a for loop steps through the points in the line using the

old y=mx+b chestnut. pset slaps down all the pixels in the proper spot.

Make the appropriate changes in your VERTLINE.C source code, adding in the

new stuff for the line function as shown above. Save your new creation to

disk under the name LINE1.C. Shift+Click here to download a copy of

LINE1.C.

Compile and run!

Type in the following as your first sample set of coordinates:

0,0,320,200

Press Enter. Thwoop! A nice red line from the upper left corner down to

the lower right. Press Enter to quit.

 Now you can mess with drawing lines all the doo-dah-day if you like.

Unfortunately, you're going to eventually encounter the following problem.

Run the program again and type in these coordinates:

160,200,0,0

That's a line from the bottom center of the screen up and back to the

upper left corner. Press Enter to draw the line.

Or not, as the case may be. You see, it's the for loop again. It expects

the values of X to crawl ahead from little to big. When you try to draw a

line in the other direction, the loop never loops. This isn't a major

hassle, though. It simply means you need to check to see if the

coordinates are "backwards" and then re-reverse them using the swap

function.

The following, final rendition of the line program, LINE.C, does the job

completely and utterly:

Name: LINE.C

#include <stdio.h>

#define VGA256 0x13

#define XMAX 320

#define YMAX 200

#define RED 4
void pset(int x,int y,unsigned char color);

unsigned char get_mode(void);

void set_mode(unsigned char mode);

void line(int x1,int y1,int x2,int y2,unsigned char color);

void swap(int *a,int *b);

void main()

unsigned char save;

int x1,y1,x2,y2;

/* Initialze things */

save = get_mode(); //Save current screen mode

set_mode(VGA256);

/* Main routine */

printf("Coordinates [x1,y1,x2,y2]:");

scanf("%i,%i,%i,%i",&x1,&y1,&x2,&y2);

printf("(%i,%i) to (%i,%i)",x1,y1,x2,y2);

getchar();
 line(x1,y1,x2,y2,RED);

getchar();

/* Reset stuff for end */

set_mode(save); //Restore screen mode

void line(int x1,int y1,int x2,int y2,unsigned char color)

int x,y;

float m,b;

// Plot straight vertical line

if (x2==x1)

if(y1 > y2)

swap(&y1,&y2);

for(y=y1;y<=y2;y++)

pset(x1,y,color);
 return;

// Reverse X1 and X2 if needed

if(x1 > x2)

swap(&x1,&x2);

swap(&y1,&y2);

m = (float)(y2-y1)/(float)(x2-x1);

b = y1 - (m*x1);

for(x=x1;x<=x2;x++)

y = m*x+b;

pset(x,y,color);

/* Swap the values of two variables */

void swap(int *a,int *b)

int temp;

temp = *a;

*a = *b;

*b = temp;

void pset(int x,int y,unsigned char color)

_asm mov ah,0ch

_asm mov al,color

_asm mov cx,x

_asm mov dx,y

_asm int 10h

unsigned char get_mode(void)

unsigned char mode;

_asm mov ah,0fh

 _asm int 10h

_asm mov mode,al

return(mode);

void set_mode(unsigned char mode)

_asm mov ah,00

_asm mov al,mode

_asm int 10h

}}

Shift+click here to download a final copy for your files. Then compile and

run. The program draws lines between any two points on the screen. No

problem.

The new chunk of code in the line function merely checks to see if X1 is

greater than X2. If so, then both coordinates are swapped, X1,Y2 and

X2,Y2. After all, it doesn't matter which direction the line goes, merely

that it connects those two points. Again, the handy swap function is used

to swap both coordinate sets.

The next lesson continues your graphics studies with some polygons drawn

using the basic line function.

On your own, you should try to devise a program that draws lines between
random coordinates using random colors.

Another program to consider: a line that bounces around the screen (like

a cheesy screen saver) in different colors. I did one of these on my

own. Shift+click here to see a copy, but don't cheat! Come up with your

own first before you compare it to mine. (And remember that there are

many different ways to work the same program.)